id
string | text
string | source
string | added
string | created
string | metadata
dict |
|---|---|---|---|---|---|
PMC1073694
|
Introduction {#s1}
============
The universal tree of life spans a diverse set of species with distinct phenotypic characteristics, here referred to as *traits.* These include specialized lifestyles (e.g., parasitism), usage of different energy sources (e.g., sunlight), and morphological properties (e.g., motility). Identification of genotype--phenotype relationships is one of the major goals of the genomics era \[[@pbio-0030134-b01]\], as it may lead to the discovery of novel biochemical and cellular processes and to the molecular understanding of complex phenotypic phenomena, including diseases.
Comparative genome analysis was proposed for resolving trait--gene relationships \[[@pbio-0030134-b01],[@pbio-0030134-b02]\] and has recently been used to predict genomic determinants for the well-known trait characteristics hyperthermophily \[[@pbio-0030134-b03],[@pbio-0030134-b04]\], flagellar motility \[[@pbio-0030134-b04],[@pbio-0030134-b05]\], and pili assembly \[[@pbio-0030134-b04]\]. The underlying principle is that species sharing a phenotype are likely to utilize orthologous genes in the involved biological process---thus correlations between the presence and absence of both genes and traits across species should indicate relevant genotype--phenotype associations (similarly, the co-occurrence of genes across species indicates functional links between proteins \[[@pbio-0030134-b06],[@pbio-0030134-b07]\]). While the applicability of the principle was demonstrated by the case studies above \[[@pbio-0030134-b03],[@pbio-0030134-b04],[@pbio-0030134-b05]\], they require manually curated knowledge on phenotypes in particular species, which is both time-consuming and unlikely to reveal unexpected relationships in the global context of all associations between genes and phenotypes.
We have thus developed an approach that involves an unbiased large-scale search for trait-descriptive terms leading to the discovery of several novel and unanticipated gene--phenotype relationships. Using literature mining, trait characteristics of species may be retrieved directly from MEDLINE abstracts (currently, the database contains more than 12 million abstracts) by identifying words that preferentially occur in abstracts referring to particular species. We focus on traits scattered across the universal tree of life (those characteristic for subsets of distantly related species), which allow species to be grouped by phenotypic rather than phylogenetic similarity. Finally, we identify systematically associations between genes and phenotypes based on the similarity of their phyletic distribution.
Results {#s2}
=======
A Systematic Approach Associates Genotypes and Phenotypes {#s2a}
---------------------------------------------------------
Historically, phenotypes are probably best understood as visible or measurable characteristics of species, which have been observed and described by biologists. The most complete resource of phenotypic knowledge is therefore the scientific literature itself. We thus identified associations between traits and genes using a multistep procedure involving genome and literature mining (see [Figure 1](#pbio-0030134-g001){ref-type="fig"}, and [Materials and Methods](#s4){ref-type="sec"} for a detailed description).
::: {#pbio-0030134-g001 .fig}
Figure 1
::: {.caption}
###### A Systematic and Unbiased Approach Combines Literature Mining and Comparative Genome Analysis with Associate Genes and Phenotypes
Words likely to describe phenotypic characteristics, that is, those preferentially co-occurring with certain species, are retrieved from MEDLINE abstracts. Phyletic distributions of genes are obtained using OGs from STRING \[[@pbio-0030134-b08]\]. As an example, phyletic distributions across bacteria for selected words and genes are shown in (I): we show species--word association scores for the words "flagellum", "flagellin", and "sewage", as well as presence/absence patterns of the selected genes *fliR* (COG1684) and *fliQ* (COG1987). Species--word association scores greater than 0 indicate that a word is likely to describe a trait of the species (colours indicate that green = true positive, i.e., the flagellar phenotype was correctly inferred; yellow = false negative; red = false positive). (I′) Black and grey bars indicate OG presence in a species (tree shown in \[I′′\]), while grey bars indicate presumably inactive genes \[[@pbio-0030134-b34],[@pbio-0030134-b35]\]. To identify informative phyletic distributions of traits and OGs, both species--word association and species--OG occurrence vectors are transformed using PCA. The similarity of the resulting transformed and normalized word and OG vectors (i.e., the word--OG association score) is computed from their inner vector products. A "heat map" (II) shows the distribution of word--OG association scores for the more than 300 words (y-axis) and over 500 OGs (x-axis) that reveal at least one significant, high-confidence association. Dendrograms are constructed by means linkage analysis, independently applying the inner products of transformed and normalized word and OG vectors as similarities. Clusters of associated words and OGs include many previously known trait--gene relationships. For example, terms mainly related to flagellar motility form a cluster with 29 OGs known to be involved in movement; see (III).
Abbreviations: Flagellum, flagellar function; Ct, involved in chemotaxis.
:::
:::
First, we computed associations between 172,967 nouns from MEDLINE abstracts and 92 prokaryotic *species,* for which both a complete genome sequence and a controlled vocabulary entry (Medical Subject Headings \[MeSH\] term) are available. Thereby, we assumed that *words* that preferentially co-occur with a subset of species are likely to be *trait-descriptive* (see for instance the examples given in [Table 1](#pbio-0030134-t001){ref-type="table"}).
::: {#pbio-0030134-t001 .table-wrap}
Table 1
::: {.caption}
###### Selected Representative Prokaryotic Species with Significantly Associated Words, Assumed to Directly Relate to Phenotypic Characteristics of the Species
:::
Here, words extracted from MEDLINE that are significantly associated with a species, i.e., those with species--word association score 2, are presented (the full list of species--word associations is available as Table S4). Trait-descriptive words are derived from the 30 most significantly associated words for a species. Words referring to a species or gene name, such as "bacillus", "subtilis", or "spoiiaa" were removed. The column "Score" indicates the level of significance (i.e., different levels of word--species association scores)
:::
Second, to record the presence and absence of 224,754 genes across these 92 genomes, 11,026 orthologous groups (OGs) of genes were obtained from the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database \[[@pbio-0030134-b08]\].
Third, to correct for the phyletic sampling of genome sequencing, we transformed the species association profiles of both trait-descriptive words and OGs, using principal component analysis (PCA).
Fourth, after comparing the resulting species--word and species--OG vectors, 2,700 significant associations between trait-descriptive words and OGs were obtained from a total of 1.9 × 10^9^ possible binary relationships. We consider as significant word--OG association scores causing at least 5-fold enrichment in true positives over expectation; this has been measured by comparing predicted associations to previously established trait--gene relationships, which were also extracted from MEDLINE (see [Figure 2](#pbio-0030134-g002){ref-type="fig"}, and [Materials and Methods](#s4){ref-type="sec"}).
::: {#pbio-0030134-g002 .fig}
Figure 2
::: {.caption}
###### Assessment of Prediction Quality
The figure demonstrates cumulative fractions of predicted OG--word associations that agree with previously known word--gene relationships (as extracted from MEDLINE). Independently confirmed predictions are enriched for high word--OG association scores.
:::
:::
As significant associations include substantial numbers of words and OGs with similar phyletic distributions, we independently generated *sets* of both words and OGs using means linkage clustering. The resulting sets were linked if they shared at least one significant word--OG association, leading to 323 significant word--OG clusters. Visual inspection of trait--gene associations in the well-studied processes of flagellar motility and plant constituent degradation revealed that for significant word--OG association scores the vast majority of all predicted associations between the phenotype and the respective cellular processes can be justified based on previous knowledge. At a more stringent level of \>7.5-fold enrichment over expectation, virtually all known trait--gene relations identified seem to agree with the clustering, and these "high-confidence" scores are analyzed here in detail (see, e.g., [Figures 1](#pbio-0030134-g001){ref-type="fig"} and [3](#pbio-0030134-g003){ref-type="fig"}, and [Table 2](#pbio-0030134-t201){ref-type="table"}). The entire list of significant clusters of trait--gene associations is available as Tables [S1](#st001){ref-type="supplementary-material"} and [S2](#st002){ref-type="supplementary-material"}.
::: {#pbio-0030134-g003 .fig}
Figure 3
::: {.caption}
###### Associations between Trait-Descriptive Words and OGs for Two Illustrative Clusters
"Heat maps" display word--OG association scores (scores greater than 0 are indicated; negative values are set to 0). We considered all words and OGs contributing to the respective cluster with at least one high-confidence association. Protein interaction networks, shown below, were derived from genomic context analysis (see [Materials and Methods](#s4){ref-type="sec"}). (A) Traits and genes related to plant constituent degradation. Functional descriptions are: Plant-degr., involved in plant constituent degradation; Ox, putative oxidoreductases; Arg, Arginine degradation protein/predicted deacylase; UV, UV damage repair endonuclease; those with no description are uncharacterized. Terms related to sporulation reflect a domination of exo- and endospore-forming species from different genera (e.g., *Streptomyces, Bacillus,* and *Clostridium*) in these degradation processes. (B) Traits and genes related to food spoilage and poisoning. Some proteins have previously been implicated in virulence of food pathogens such as ManR ("T"), a transcriptional antiterminator involved in resistance to natural food preservatives, and some propanediol degradation proteins ("Prop-diol"). We suggest the involvement of additional proteins in pathogenicity: for example, ethanolamine degradation proteins ("Eth.-amine-usage"; the phospholipid phosphatidyl-ethanolamine, cleaved to ethanolamine by phospholipase, is abundant in the gut \[[@pbio-0030134-b14]\]); the cobalt chelatase CbiK ("C"; cobalt is an essential factor for propanediol and ethanolamine utilization \[[@pbio-0030134-b14]\]); a phosphotransferase system ("PTS") involved in sorbitol transport \[[@pbio-0030134-b36]\] (sorbitol is an artificial food sweetener naturally found in fruits and may act as an additional carbon source; we suggest that alternatively the chemically similar inositol, cleavage product of another abundant phospholipid, may be utilized). Other proteins that may also be involved are a presumably anaerobically used butyrate kinase ("B"), gamma-glutamylcysteine synthetase ("G"), an electron transport complex protein ("O"), a predicted metal-binding enzyme ("E"), and several uncharacterized proteins (no description).
:::
:::
::: {#pbio-0030134-t201 .table-wrap}
Table 2
::: {.caption}
###### Significant Associations of Trait-Descriptive Words with Biological Functions
:::
The table demonstrates associations of words and biological functions retrieved from the functional annotation of the respective OGs. Related trait--gene associations are shown as clusters, ranked according to the best scoring association contributing to the cluster. Only large clusters are shown (i.e., Clusters 9, 11, 12, 18, and 22--24 \[see Tables S1and S2\] were not considered here as they contain only one word or OG). Clusters 15 (sporulation) and 21 (hyperthermophily) were manually refined (see Tables S2 and S3). The second column indicates trait-descriptive words; the third column indicates numbers of associated OGs (OGs comprising hypothetical proteins in parentheses). We considered all words and OGs contributing to the cluster with at least one high-confidence association. The fifth column lists annotated functions of the OGs; the best word--OG association score is given in column six. The table indicates a striking enrichment of disease-related terms and OGs (see fourth column). This is most likely due to a bias in genome sequencing towards disease-causing bacteria, and a bias in MEDLINE towards disease-related topics. Furthermore, horizontal gene transfer may be particularly frequent for pathogenic species \[[@pbio-0030134-b22]\]
^1^ Some photosynthetic cyanobacteria may reproduce using endospores, or may form spore-like structures during nitrogen fixation
:::
::: {#pbio-0030134-t202 .table-wrap}
Table 2
::: {.caption}
###### Continued
:::
:::
Altogether, we can infer at least one significant word--OG association for 811 OGs, corresponding to 28,888 genes. While many of these correspond to previously known trait--gene relationships, numerous relationships are novel or of unexpected character and complexity.
Identification of Previously Known Trait--Gene Associations {#s2b}
-----------------------------------------------------------
Among the many previously known trait--gene relationships are genes involved in pathogenicity, the degradation of plant constituents, capsule biosynthesis, and photosynthesis (see [Table 2](#pbio-0030134-t201){ref-type="table"}). Furthermore, determinants for flagellar motility and hyperthermophily, the target for previous case studies \[[@pbio-0030134-b03],[@pbio-0030134-b04],[@pbio-0030134-b05]\], were identified (see [Tables 3](#pbio-0030134-t003){ref-type="table"} and S3). For example, 72% of all genes involved in synthesizing or maintaining the bacterial flagellum, according to *Kyoto Encyclopedia of Genes and Genomes* (KEGG \[[@pbio-0030134-b09]\]), are recovered with high-confidence scores (see [Figure 1](#pbio-0030134-g001){ref-type="fig"} and [Table 3](#pbio-0030134-t003){ref-type="table"}).
::: {#pbio-0030134-t003 .table-wrap}
Table 3
::: {.caption}
###### OGs and Terms Associated with Motility (Cluster 2)
:::
We recovered 72% of all genes involved in synthesizing or maintaining the bacterial flagellum. Twenty-six out of 36 genes listed in the KEGG database \[[@pbio-0030134-b09]\] were identified unambiguously with high-confidence word--OG association scores. Flagellar OGs form a distinct cluster (Cluster 2) together with OGs involved in chemotaxis, a process tightly linked to motility \[[@pbio-0030134-b56]\]. Thus, all OGs in Cluster 2 were previously known to be motility-related. Analysis of conserved gene neighbourhood and gene fusions confirms the functional linkage between all motility-related OGs identified by our approach. The words "flagellin" (which represents a major component of the bacterial flagellum \[[@pbio-0030134-b57]\]) and "flagellum" are most tightly associated with their respective genes, followed by "motility", and the more general terms "sewage" and "vegetable". Although at first sight surprising, the latter two terms may refer to niches preferably colonized by motile bacteria \[[@pbio-0030134-b58]\]---that is, sewage and food---where bacteria that possess flagella may distribute most rapidly. In the table, the first column indicates the OG identifier; the second column shows representative gene names; the third column is annotated as a part of the flagellum in the KEGG database; the fourth column consists of associated words
:::
Prediction of Novel Genes in Known Processes: Enzymes Involved in Plant Degradation {#s2c}
-----------------------------------------------------------------------------------
We can predict trait relationships for many hypothetical genes. Within the 20 best scoring clusters (all of high-confidence trait--gene associations), 113 uncharacterized OGs (i.e., no functional annotations are given in SWISSPROT, TREMBL, or the Clusters of Orthologous Groups \[COGs\] database) are linked to trait-descriptive words (see [Tables 2](#pbio-0030134-t201){ref-type="table"} and [S1](#st001){ref-type="supplementary-material"}). For example, "hypothetical" and poorly characterized OGs are predicted to encode enzymes involved in the biodegradation of plant constituents ([Figure 3](#pbio-0030134-g003){ref-type="fig"}A); altogether 13 words, mainly describing reactions that break down plant polysaccharides, form a set with 15, mostly enzyme-encoding, OGs. Out of the latter, 75% (i.e., nine of the 12 OGs that have at least provisional functional annotations) are already known, or have been suggested, to break down plant polysaccharides (see Tables [S1](#st001){ref-type="supplementary-material"} and [S2](#st002){ref-type="supplementary-material"}). We predict involvement in plant degradation also for the remaining OGs, which include three uncharacterized OGs, and two with unspecific functional annotation (nonsupervised orthologous group \[NOG\]12385 is a "putative oxidoreductase"; COG4187 members are annotated as "Arginine degradation protein" and "predicted deacylase").
Prediction of Unexpected Associations: Novel Genomic Determinants for Food Poisoning? {#s2d}
-------------------------------------------------------------------------------------
Of the many unanticipated trait--gene relationships, we will discuss in detail a cluster of word--OG associations that groups several terms related to food and food poisoning with a number of OGs encoding metabolic enzymes ([Figures 3](#pbio-0030134-g003){ref-type="fig"}B and [4](#pbio-0030134-g004){ref-type="fig"}). The words of the respective set refer, for instance, to common habitats of food pathogens, such as processed and spoiled food (e.g., "cheese", "sausage", "broiler", "spoilage", and "carcass"), and to natural food preservatives used to inhibit their growth ("bacteriocin"). Furthermore, several specific pathogenicity-related terms were revealed: for example, "monocytogene", "gastroenteritis", and "abortion" that refer to immune response, direct and indirect results of intestinal infections, respectively \[[@pbio-0030134-b10]\]. Also the terms "phospholipase" and "lecithinase" in this word set have been implicated in toxicity mechanisms \[[@pbio-0030134-b11],[@pbio-0030134-b12]\], despite the general cellular roles of the proteins. Other words may refer to unspecific connotations of more general words, which play an important role in clinical praxis or the food industry, such as "starter" (culture) and "vacuum" (packaging). We predict for 37 OGs with high-confidence association to these words that they are involved in food spoilage and toxicity. All of these OGs are present in food-borne pathogens, and absent from most other prokaryotes. One of them has already been demonstrated experimentally to be involved in food spoilage and pathogenicity: the *manR* gene (COG3933) of the food pathogen Listeria monocytogenes encodes a transcriptional regulator that was shown to be involved in resistance to natural food preservatives \[[@pbio-0030134-b13]\].
::: {#pbio-0030134-g004 .fig}
Figure 4
::: {.caption}
###### Phyletic Distributions across Bacteria of Genes and Associated Representative Trait-Descriptive Words Related to Food, Food Spoilage, and Food Poisoning (Cluster 1)
The complete figure, including phylogenetic distributions of all trait-descriptive words and OGs in Cluster 1, is available online as [Figure S1](#sg001){ref-type="supplementary-material"}. Black squares indicate gene occurrences across species for the respective OGs. Blue squares indicate predicted associations between trait-descriptive words and species (species--word association scores greater than 0). Function descriptions (grey bar) are the same as in [Figure 3](#pbio-0030134-g003){ref-type="fig"}.
:::
:::
Of the remaining 36 OGs, four are components of the utilization pathway for 1,2-propanediol and eight participate in ethanolamine usage. Among the remaining, mostly poorly characterized OGs, one is a cobalt chelatase likely to be involved in providing the essential cobalamin for both pathways \[[@pbio-0030134-b14]\]. Intriguingly, propanediol and ethanolamine are abundant compounds in the human gut \[[@pbio-0030134-b14],[@pbio-0030134-b15]\] and in processed food \[[@pbio-0030134-b16],[@pbio-0030134-b17]\] that can be utilized as the main source of carbon, nitrogen, and energy under aerobic and anaerobic conditions \[[@pbio-0030134-b10],[@pbio-0030134-b14]\]. The corresponding genes form conserved operons in three of the most hazardous food-borne pathogens---L. monocytogenes (a low-GC Gram-positive bacterium, *Bacillales* family), Clostridium perfringens (high-GC Gram-positive), and Salmonella typhimurium (Gram-negative)---but are absent from almost all other species. Further sequence analysis in National Center for Biotechnology Information\'s NRDB revealed the presence of ethanolamine usage genes in the recently sequenced species Enterococcus faecalis---another gut pathogen from a distinct phylogenetic clade *(Lactobacillales).* Based on these associations we predict that both propanediol and ethanolamine utilization pathways are crucial genomic determinants of pathogenicity associated with food poisoning, presumably by promoting anaerobic growth both in the host and in processed food. Thereby, they may provide a growth advantage over natural gut inhabitants not possessing those genes ([Figure 4](#pbio-0030134-g004){ref-type="fig"}). In agreement with these predictions, genes involved in propanediol utilization were previously proposed to be involved in pathogenicity of S. typhimurium as deletion of the respective genes specifically impairs growth in the host \[[@pbio-0030134-b18]\].
Discussion {#s3}
==========
We present here a computational approach that combines literature data mining with comparative genome analysis to systematically identify numerous novel phenotype--genotype relationships. Text-mining methods have previously been used for function prediction, for example, to associate functionally interacting proteins (e.g., \[[@pbio-0030134-b19],[@pbio-0030134-b20]\]) and to link human genes to hereditary diseases (e.g., \[[@pbio-0030134-b21]\]). However, this is, to the best of our knowledge, the first unbiased and systematic approach that enables the identification of genomic determinants for numerous phenotypes, without requiring manually curated knowledge on phenotypic characteristics.
Besides identifying several previously known relationships, our approach predicts many novel and unanticipated trait--gene associations with high confidence ([Figures 3](#pbio-0030134-g003){ref-type="fig"} and [4](#pbio-0030134-g004){ref-type="fig"} and [Table 2](#pbio-0030134-t201){ref-type="table"}). The association of food- and spoilage-related terms with genes presumably involved in food intoxication exemplifies the identification of unforeseen relationships using an unsupervised strategy. A typical example is the significant association of "cheese", "abortion", and predicted pathogenicity factors, presumably explained by *Listeria* infections following raw-milk cheese consumption, which may cause miscarriage.
Given the focus of research and, in particular, genome sequencing on tackling human health issues, it is not surprising that a considerable proportion of all inferred genes are predicted to be involved in bacterial pathogenicity ([Table 2](#pbio-0030134-t201){ref-type="table"}). It was furthermore proposed that genes responsible for virulence are frequently transferred horizontally across distant clades \[[@pbio-0030134-b22]\], which would favour them in our analysis. These arguments imply that our approach represents a well-suited tool for identifying disease-related genes that may serve as promising novel classes of drug targets.
Altogether, we have identified 2,700 significant OG--word associations, which link more than 800 OGs (encoding over 28,000 proteins) to at least one trait-descriptive word. Our approach to function prediction is complementary to computational methods that utilize evolutionary signals, such as genomic context methods that analyze conserved gene neighbourhood \[[@pbio-0030134-b23],[@pbio-0030134-b24],[@pbio-0030134-b25]\], or gene fusion \[[@pbio-0030134-b26],[@pbio-0030134-b27]\], to predict functional associations in terms of involvement in a common cellular process. Notably, the majority of poorly characterized OGs we identify cannot be functionally annotated by these methods (see [Figure 3](#pbio-0030134-g003){ref-type="fig"} and [Materials and Methods](#s4){ref-type="sec"}). Furthermore, even if genomic context methods link to a set of OGs, our approach can associate these sets to phenotypic characteristics. For example, while genes involved in ethanolamine utilization can be inferred using existing genome context methods, our approach enables linking this cellular process with food-borne pathogenicity. Furthermore, grouping of ethanolamine utilization with the food preservative resistance factor *manR* and other genes indicates that independent cellular processes can be combined by the approach, if they are involved in the same trait.
As more genomes are sequenced and the available literature in MEDLINE is constantly increasing in size, our approach is expected to predict more fine-grained trait--gene relationships in the future. This may pave the way for mapping numerous distinctive phenotypic characteristics observed in nature to the genes responsible.
Materials and Methods {#s4}
=====================
{#s4a}
### Identification of species--word associations {#s4a1}
We first identified the distribution of phenotypic traits across species by analyzing the co-mentioning of species and trait-descriptive words across MEDLINE abstracts. Namely, nouns that preferentially co-occur with a subset of species are likely to be trait-descriptive (e.g., the words "flagellum" and "motility" are enriched in abstracts dealing with motile species; see [Figure 1](#pbio-0030134-g001){ref-type="fig"}). We focused on nouns because these presumably carry a more considerable proportion of the relevant information represented in the scientific literature than verbs and adjectives \[[@pbio-0030134-b28]\]. Nouns were extracted from MEDLINE abstracts using a part-of-speech tagger (i.e., Tree Tagger \[[@pbio-0030134-b29]\]). Words of five characters or less were excluded from the analysis, as many of those are gene names and other noninformative words leading to an increase in noise. Species names were taken from the corresponding MeSH terms associated with the abstracts, that is, from the MeSH B category corresponding to "organisms" (applying the controlled MeSH vocabulary reduces errors in species name recognition---for example, in the case of synonym usage; on the other hand, using all nouns in MEDLINE abstracts for identifying trait-descriptive words allows searching a variety of traits not accessible via a controlled vocabulary). Some species were not represented in MeSH, and were thus mapped to their genus. A total of 255,249 MEDLINE abstracts connected with any of the 92 species analyzed were considered in the analysis. We considered the occurrence of distinct species in abstracts. Frequencies of words within abstracts were not taken into account (single and multiple occurrences were equally treated as "word presence"). Given the set of *n* ~1~ words and *n* ~2~ species associated with an abstract, we counted all possible species--word pairs *(n* ~1*×*~ *n* ~2~ *).* For each species--word pair, a species--word association score *s~sw~* was determined using a regularized log-odds score:
where *n~s~* is the number of abstracts mentioning a particular species, *n~w~* refers to the number of abstracts mentioning a particular word, *n~sw~* is the number of abstracts that co-mention species and word, and *N* is the sum of all *n~sw~*. The log-odds framework quantifies correlation strengths and, in particular, facilitates the handling of species or words for which only sparse scientific literature exists. To allow the handling of sparse data, the standard log-odds formula was augmented with pseudocounts, *p* = 1. The resulting score, *s~sw~,* yields positive values for enriched species--word pairs and negative values for underrepresented pairs. The magnitude of the score provides a measure of the strength of the association, indicating its potential relevance in describing a characteristic trait. To record overrepresentation, the species--word association score requires frequently used words and species (such as "flagellum" and Escherichia coli) to be co-mentioned more often than infrequent ones (e.g., "oligosaccharide" and Ralstonia solanacearum).
Associations were calculated for each species--word pair, and a species association vector was subsequently constructed for each word, representing its association scores with each of the 92 prokaryotic species studied.
### Identification of orthologues and species--gene associations {#s4a2}
We obtained groups of OGs covering those 92 completely sequenced genomes from the STRING server \[[@pbio-0030134-b08]\], version 4 (the raw data can be downloaded from <http://string.embl.de>). The OGs include protein families originally obtained from the COG database \[[@pbio-0030134-b30]\], which were subsequently expanded and extended to accommodate more recently sequenced species \[[@pbio-0030134-b08]\]. Species--OG association vectors were constructed for each OG; +1 signifies presence of a gene (i.e., orthologue) in a certain species, while −1 signifies absence.
### PCA {#s4a3}
To reduce the sampling bias introduced by genome sequencing projects, which have been focused on certain groups of closely related species, we performed PCA (also known as singular value decomposition) on the species--OG association matrix. The PCA transformation collapsed groups of species with very similar gene content to a single dimension, thereby eliminating inherent correlations from our representation. Subsequently, the same linear transformation was applied to the species--word association vectors. For both species--OG and species--word association vectors, the first 32 principal components were further considered, yielding an acceptable signal-to-noise ratio. The performance of the approach was comparable (i.e., slightly weaker) when applying distinct numbers of components in the range from approximately 25 to 40. Considering even smaller or larger numbers of components led to performance drops, as too little information or too much noise was included in the further analysis.
### Mapping genes to phenotypes and vice versa {#s4a4}
Before mapping genes and traits, further filtering was applied to diminish the contribution of rarely occurring OGs and words; that is, only OGs occurring in at least four distantly related species clades were considered; similarly, we focused on words yielding positive species--word associations in at least four distantly related species clades (thereby utilizing clades of closely related species from STRING \[[@pbio-0030134-b08]\]; see [Table S1](#st001){ref-type="supplementary-material"} in \[[@pbio-0030134-b25]\]). OGs encoding phage-associated proteins (i.e., those with description lines including the terms "phage", "transposase", and "integrase") were regarded as a source of "contamination" within the genomes of analysed species and thus ignored. Finally, we eliminated words that did not display sufficiently strong association with any of the species studied; this was done by removing all but the 1,000 longest transformed species--word vectors (these were considered to be the most informative vectors). The remaining vectors were normalized to a length of 1; similarly, all species--OG vectors were normalized. Subsequently, the pair-wise similarity of each word--OG pair was computed, that is, the word--OG association score, defined as the inner product of normalized species--OG and species--word vectors (the highest word--OG association score obtained is 0.812; see [Table 2](#pbio-0030134-t201){ref-type="table"}). Furthermore, the similarity score for pairs of OGs and for pairs of words was computed in the same way as described for the word--OG associations. Using means linkage clustering analysis as implemented in OC \[[@pbio-0030134-b31]\] ("similarity mode"; cutoff = 0.45), sets of words and OGs were independently generated. Finally, combined clusters were constructed by combining word sets with OG sets, if these were linked by at least one significant word--OG association. Note that this "loose" clustering procedure allows word sets to be combined with several OG sets, and vice versa (i.e., words or OGs may in principle be part of several clusters).
### Assessment of prediction quality {#s4a5}
We reasoned that the quality of the predictions may be examined using an orthogonal strategy---comparing the predicted word--OG associations to previously established trait--gene relationships, which can be extracted from MEDLINE when focusing on significantly associated word--gene pairs. Namely, previously established relations were extracted from scientific abstracts, by detecting significant co-mentioning of trait-descriptive words and gene names, using the hypergeometric distribution \[[@pbio-0030134-b32],[@pbio-0030134-b33]\]. Thereby, gene names were associated to species, considering MeSH terms and organism names occurring in abstracts, and including gene synonyms retrieved from <http://www.bork.embl-heidelberg.de/synonyms>. (For instance, the word "flagellum" co-occurs significantly with *fliR* from *S. typhimurium, p* = 0.00063, consistent with a gene function in motility). We assume confirmation of a predicted word--OG association *("true positive"),* if any gene within an OG significantly co-occurs with a word (i.e., when *p* ≤ 1/10^1.5^, roughly corresponding to *p* ≤ 0.03). [Figure 2](#pbio-0030134-g002){ref-type="fig"} demonstrates the enrichment of true positives among the highest scoring predictions. Words not significantly associated with any gene, or OGs lacking genes significantly linked to any word were ignored. Furthermore, we conservatively estimated expected fractions of true positives by shuffling both OGs and the 1,000 most informative words, and subsequently repeating the assessment with previously established trait--gene relationships on these randomized associations. By comparing to expected scores, we estimated two significance thresholds: word--OG association scores ≥0.5675 (true positives are 5-fold enriched over expectation) are regarded as significant; scores ≥0.6125 (7.5-fold enrichment) indicate "high-confidence". OGs and words discussed in detail (see, e.g., [Figures 1](#pbio-0030134-g001){ref-type="fig"}, [3](#pbio-0030134-g003){ref-type="fig"}, and 4 and [Tables 2](#pbio-0030134-t201){ref-type="table"} and [3](#pbio-0030134-t003){ref-type="table"}) all contribute with at least one high-confidence association to the respective clusters.
### Construction of genomic context association networks {#s4a6}
We tested whether functional annotations predicted from our method can also be inferred with existing computational methods (which utilize different methodologies than the approach described here). We compared predictions from our approach to functional associations between proteins inferred from genomic context methods (see, e.g., [Figure 3](#pbio-0030134-g003){ref-type="fig"}), which predict involvement in a common metabolic pathway or biochemical process. Thereby, we considered OGs to be functionally characterized by another method, if the corresponding genes can be significantly associated to genes of known function, that is, if they are fused to such genes \[[@pbio-0030134-b26],[@pbio-0030134-b27]\], or if they occur in conserved proximity with these (we analyzed conserved organization of putative operons \[[@pbio-0030134-b23],[@pbio-0030134-b24]\], and of divergently transcribed gene pairs \[[@pbio-0030134-b25]\]; gene fusions and conserved \[putative\] operon structures were examined using STRING \[[@pbio-0030134-b08]\] with the default probability cutoff of 0.400, excluding evidence from sources other than gene fusion or neighbourhood; for the divergently transcribed gene pair method, all pairs conserved across at least three distant evolutionary species clades were considered \[[@pbio-0030134-b25]\]).
### Species analyzed {#s4a7}
The following species were analyzed:
*Aeropyrum pernix, Agrobacterium tumefaciens* (Cereon), A. tumefaciens (Wash.), *Aquifex aeolicus, Archaeoglobus fulgidus, Bacillus halodurans, B. subtilis, Bifidobacterium longum, Borrelia burgdorferi, Bradyrhizobium japonicum, Brucella melitensis, Buchnera aphidicola, B. aphidicola* Schiz*, Campylobacter jejuni, Caulobacter crescentus, Chlamydia muridarum, C. trachomatis, Chlamydophila pneumoniae* AR39, C. pneumoniae CWL029, C. pneumoniae J138, *Clostridium acetobutylicum, C. perfringens, Corynebacterium efficiens, C. glutamicum, Deinococcus radiodurans, Escherichia coli* K12, E. coli O157:H7, E. coli O157:H7 EDL933, E. coli O6, *Fusobacterium nucleatum, Haemophilus influenzae, Halobacterium* sp. NRC-1, Helicobacter pylori 26695, *Lactococcus lactis, Leptospira interrogans, Listeria innocua, L. monocytogenes, Mesorhizobium loti, Methanococcus jannaschii, Methanosarcina acetivorans, M. mazei, Mycobacterium leprae, M. tuberculosis* CDC1551, M. tuberculosis H37Rv, *Mycoplasma genitalium, M. pneumoniae, M. pulmonis, Neisseria meningitidis* A, N. meningitidis B, *Nostoc* sp. PCC 7120, *Pasteurella multocida, Pseudomonas aeruginosa, P. putida, Pyrobaculum aerophilum, Pyrococcus abyssi, P. furiosus, P. horikoshii, Ralstonia solanacearum, Rickettsia conorii, R. prowazekii, Salmonella typhi, S. typhimurium, Shewanella oneidensis, Shigella flexneri, Sinorhizobium meliloti, Streptococcus agalactiae, S. mutans, S. pneumoniae* R6, S. pneumoniae TIGR4, *S. pyogenes, S. pyogenes* M3, S. pyogenes MGAS8232, Staphylococcus aureus Mu50, S. aureus MW2, S. aureus N315, *S. epidermidis, Streptomyces coelicolor, Sulfolobus solfataricus, S. tokodaii, Synechococcus elongatus, Synechocystis* sp. PCC 6803, *Thermoplasma acidophilum, T. volcanium, Thermotoga maritima, Treponema pallidum, Ureaplasma parvum, Vibrio cholerae, Xanthomonas axonopodis, X. campestris, Xylella fastidiosa, Yersinia pestis, and Y. pestis* KIM*.*
Supporting Information {#s5}
======================
Figure S1
::: {.caption}
###### Genes and Trait-Descriptive Words Related to Food Poisoning
Phyletic distributions across bacteria of genes and associated trait-descriptive words related to food, food spoilage, and food poisoning (complete figure).
(763 KB JPG).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S1
::: {.caption}
###### Complete List of Trait-OG Clusters Based on All Significant Associations
(221 KB TXT).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S2
::: {.caption}
###### Complete List of Trait-OG Clusters Based on High-Confidence Associations Only
The clusters have been refined as described in [Table S3](#st003){ref-type="supplementary-material"}.
(116 KB TXT).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S3
::: {.caption}
###### OGs Predicted to Be Involved in Hyperthermophily
(52 KB DOC).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S4
::: {.caption}
###### Complete List of Species--Word Associations Identified by Mining MEDLINE
(10.3 MB ZIP).
:::
::: {.caption}
######
Click here for additional data file.
:::
We wish to thank members of the Bork group for helpful discussions, Eoghan Harrington for critical reading and comments, and in particular Christian von Mering for invaluable input and suggestions, and for providing data from the STRING database.
**Competing interests.** The authors have declared that no competing interests exist.
**Author contributions.** JOK, TD, LJJ, MAA, and PB conceived and designed the experiments. JOK and TD performed the experiments. JOK, TD, LJJ, CPI, SDH, MAA, and PB analyzed the data. LJJ, CPI, SK, and SDH contributed reagents/materials/analysis tools. JOK, TD, and PB wrote the paper.
Citation: Korbel JO, Doerks T, Jensen LJ, Perez-Iratxeta C, Kaczanowski S, et al. (2005) Systematic association of genes to phenotypes by genome and literature mining. PLoS Biol 3(5): e134.
COGs
: Clusters of Orthologous Groups
KEGG
: *Kyoto Encyclopedia of Genes and Genomes*
MeSH
: Medical Subject Headings
NOG
: nonsupervised orthologous group
OG
: orthologous group
PCA
: principal component analysis
STRING
: Search Tool for the Retrieval of Interacting Genes/Proteins
|
PubMed Central
|
2024-06-05T03:55:55.053165
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1073694/",
"journal": "PLoS Biol. 2005 May 5; 3(5):e134",
"authors": [
{
"first": "Jan O",
"last": "Korbel"
},
{
"first": "Tobias",
"last": "Doerks"
},
{
"first": "Lars J",
"last": "Jensen"
},
{
"first": "Carolina",
"last": "Perez-Iratxeta"
},
{
"first": "Szymon",
"last": "Kaczanowski"
},
{
"first": "Sean D",
"last": "Hooper"
},
{
"first": "Miguel A",
"last": "Andrade"
},
{
"first": "Peer",
"last": "Bork"
}
]
}
|
PMC1073695
|
Introduction {#s1}
============
Understanding the mechanisms that underlie phenotypic variation within species is crucial to addressing fundamental issues in medicine, agriculture, and evolutionary biology \[[@pbio-0030135-b01]\]. Identifying genes that contribute to variation in traits affected by multiple genetic and environmental factors has proven extremely difficult \[[@pbio-0030135-b02]\], although the molecular basis of a few quantitative trait loci (QTLs) has been elucidated \[[@pbio-0030135-b03],[@pbio-0030135-b04],[@pbio-0030135-b05]\]. Despite these successes, several general questions remain, such as whether genes involved in Mendelian disorders also contribute to complex trait variation \[[@pbio-0030135-b06]\], and the extent to which coding sequence versus regulatory variation is responsible for complex trait variation. In cases where there is heritable variation in gene expression, it is not clear what magnitude of difference is sufficient to contribute to phenotypic variation without substantial deleterious effects. These issues are particularly relevant to the further identification of genes responsible for complex trait variation. Numerous studies use expression microarrays to identify genes underlying trait variation \[[@pbio-0030135-b07],[@pbio-0030135-b08],[@pbio-0030135-b09],[@pbio-0030135-b10]\], and yet such approaches will not detect coding sequence variation or subtle differences in expression.
An archetypal model for complex trait variation is body size, but with the exception of a few Mendelian mutations \[[@pbio-0030135-b11],[@pbio-0030135-b12]\], no gene contributing to quantitative variation in this trait has been identified in animals. In previous work that examined lines of mice divergently selected for body size, we showed that much of the selection response is due to a large-effect QTL on chromosome (Chr) X that causes an approximately 20% difference in growth rate between homozygotes \[[@pbio-0030135-b13],[@pbio-0030135-b14]\] and explains 14% of the phenotypic variance at 6 wk in an F~2~ cross between the selection lines \[[@pbio-0030135-b15]\]. A large X-linked effect was observed in replicate selection lines derived independently from the same base population \[[@pbio-0030135-b13]\], indicating that the QTL is not due to a mutation that occurred during the selection process. Rather, it is due to variation segregating within the initial population, which was derived from a cross between two inbred and one outbred strains. This scenario contrasts with the "high growth" mutation *(hg),* which arose during selection for increased growth rate in a different set of selection lines \[[@pbio-0030135-b16]\] and resulted from a disruption of *Socs2 (suppressor of cytokine signalling 2),* which eliminated the expression of this gene \[[@pbio-0030135-b12]\].
To determine the molecular basis of the X-linked QTL, we fine-mapped the QTL by progeny testing, searched for sequence polymorphisms in annotated coding regions, and examined the expression of all genes within the target region.
Results/Discussion {#s2}
==================
Fine-Mapping {#s2a}
------------
The QTL had previously been mapped to a region of approximately 2 cM, or 2.6 Mb \[[@pbio-0030135-b14]\]. By further progeny testing, we refined the location of the QTL to an approximately 660-kb region ([Figure 1](#pbio-0030135-g001){ref-type="fig"}). The entire effect of the QTL is attributable to this region, as demonstrated by three recombinant families: families 103 and 105 segregate for the QTL region and for the phenotypic effect of the QTL, whereas family 101 does not segregate for either ([Figure 1](#pbio-0030135-g001){ref-type="fig"}; [Table 1](#pbio-0030135-t001){ref-type="table"}). In both sexes, the differences in effect size between families 103 and 101 and between families 105 and 101 are significant (*p* \< 0.02 in all cases), whereas the differences between families 103 and 105 are not significant (*p* \> 0.2 in both cases). Thus, in contrast to previous studies that have either found QTLs to be composed of multiple QTLs (e.g., \[[@pbio-0030135-b17],[@pbio-0030135-b18]\]), or have lacked the statistical power to dissect a single QTL, this large-effect QTL is caused entirely by one small chromosomal region.
::: {#pbio-0030135-g001 .fig}
Figure 1
::: {.caption}
###### QTL Region
At the top is shown the extent of Chr X segregating in three recombinant families. The horizontal grey bars indicate the regions known to segregate, while the error bars show the uncertainty in the location of recombination. Black bars indicate genes within the QTL region according to the Ensembl database \[[@pbio-0030135-b20]\]. Below is a LOD score plot for body mass at 6 wk in entire progeny test population (*n* = 1,909). Triangles indicate the locations of markers. At the bottom, recombination rates are shown for the intervals delimited by diamonds (the Chr X average is 0.40 cM/Mb \[[@pbio-0030135-b40]\]).
:::
:::
::: {#pbio-0030135-t001 .table-wrap}
Table 1
::: {.caption}
###### Body Mass at 6 Wk of Age in Families 103 and 105, Which Segregate for the QTL Region, and Family 101, Which Does Not
:::
Values are least squares means ± standard errors from a general linear model including family, litter nested within family, genotype, and family-by-genotype interaction
^a^ An asterisk indicates that the difference between genotypes within a family is significant (*p* \< 0.0001); no asterisk indicates *p* \> 0.05
:::
Further fine-mapping of the QTL has not been possible because the target region appears to be located in a recombination "cold spot" ([Figure 1](#pbio-0030135-g001){ref-type="fig"}). There is substantial heterogeneity in the recombination rate within the region, roughly similar in magnitude to variation observed in humans \[[@pbio-0030135-b19]\], although the cold spot may be unusually wide.
Genes within QTL Region {#s2b}
-----------------------
The QTL region contains four genes according to the Ensembl database \[[@pbio-0030135-b20]\], and function is known for only two of these: *Gpc3* and *Gpc4* ([Figure 1](#pbio-0030135-g001){ref-type="fig"}). Both of these genes encode members of the glypican family of membrane-bound heparin sulphate proteoglycans that are involved in morphogenesis and growth regulation \[[@pbio-0030135-b21]\]. Loss-of-function mutations in *Gpc3* lead to Simpson--Golabi--Behmel syndrome in humans, a disorder with numerous phenotypic effects, including overgrowth, skeletal and renal developmental abnormalities, an increased frequency of embryonic cancers, and neonatal mortality \[[@pbio-0030135-b22],[@pbio-0030135-b23],[@pbio-0030135-b24]\]. *Gpc3* knock-out mice show similar phenotypes, including increased body mass, renal dysplasias, and increased perinatal mortality \[[@pbio-0030135-b25]\]. In contrast, no obvious phenotypes are seen in *Gpc4* knock-out mice \[[@pbio-0030135-b26]\].
DNA sequencing revealed no differences in coding sequence between the high- and low-line QTL alleles at *Gpc3, Gpc4,* or *Q8C9S7,* one of the genes of unknown function. In the other gene of unknown function, *Q9D9G4,* there was one synonymous single nucleotide polymorphism ([Table S1](#st001){ref-type="supplementary-material"}).
Quantitative Measurement of Expression of *Gpc3* and *Gpc4* {#s2c}
-----------------------------------------------------------
The lack of non-synonymous differences indicates that the effect of the QTL must be due to regulatory variation. We therefore measured mRNA transcript levels in tissues of congenic mice from litters in which the QTL segregated. Newborns were examined, since the effect of the QTL on neonatal body weight is as large as that in adults \[[@pbio-0030135-b27]\].
Transcript levels of *Gpc3* and *Gpc4* were examined in kidney and liver, since both genes are expressed in these tissues in mice \[[@pbio-0030135-b28]\], and kidney abnormalities are often observed in *Gpc3* loss-of-function mutations in humans and mice \[[@pbio-0030135-b22],[@pbio-0030135-b23],[@pbio-0030135-b25]\]. Mice with the high-line allele showed 15% lower expression of *Gpc3* in liver and kidney (*p* = 0.017 and *p* = 0.012, respectively, from a general linear model fitting effects of genotype, sex, and litter; [Figure 2](#pbio-0030135-g002){ref-type="fig"}), whereas the differences in transcript levels for *Gpc4* were non-significant (*p* = 0.08 and *p* = 0.74, respectively), and the trends varied in direction between tissues ([Figure 2](#pbio-0030135-g002){ref-type="fig"}; [Table S2](#st002){ref-type="supplementary-material"}). Transcript levels of *Gpc3* and *Gpc4* were adjusted by dividing by *β-actin* levels; correcting for *β-actin* by including it as a co-variate in the model yielded qualitatively similar results. The lower level of expression of *Gpc3* in mice with the high body mass QTL allele is consistent with the overgrowth seen in *Gpc3* knock-out mice \[[@pbio-0030135-b25]\].
::: {#pbio-0030135-g002 .fig}
Figure 2
::: {.caption}
###### Transcript Levels of *Gpc3* and *Gpc4* (Divided By *β-actin*)
Expression was measured in newborn liver and kidney in homozygous low-allele females and hemizygous low-allele males (black bars) and in heterozygous females and hemizygous high-allele males (grey bars). Data are from 23 low-allele mice and 24 high-allele males/heterozygous females, and values are least squares means (± 2× standard error); \*, *p* \< 0.05.
:::
:::
Given the phenotypic effects of the QTL ([Table 1](#pbio-0030135-t001){ref-type="table"}), we would expect the difference in *Gpc3* expression between hemizygous low-allele males and hemizygous high-allele males to be greater than the difference between homozygous low-allele females and heterozygous females. Although there appeared to be some indication of sex-specific differences in *Gpc3* expression in liver ([Table S2](#st002){ref-type="supplementary-material"}), this was largely due to a marginally non-significant sex-by-genotype interaction in *β-actin* levels used to normalise *Gpc3* expression (*p* = 0.06), which generated the pattern shown in [Table S2](#st002){ref-type="supplementary-material"}; untransformed liver *Gpc3* levels did not show a significant sex-by-genotype interaction (*p* = 0.13). The lack of significant sex-specific differences in *Gpc3* expression is likely due to low statistical power to detect interactions.
Expression Profiling of Genes of Unknown Function {#s2d}
-------------------------------------------------
To examine whether the two genes of unknown function might contribute to the effect of the QTL, we examined their expression using a 24-tissue gene-expression panel. *Q9D9G4* (also known as *1700080O16*) was originally identified in adult male testis cDNA \[[@pbio-0030135-b20]\], and we observed clear expression in this tissue, as well as very low levels of expression in muscle, lung, and small intestine; no expression was detected in embryos. Others have also found much greater expression of this gene in testis than in any other tissue in a 61-tissue panel \[[@pbio-0030135-b29]\] and in a 55-tissue panel \[[@pbio-0030135-b30]\].
*Q8C9S7* (also known as *A630012P03*) was originally identified in 3-d neonate thymus cDNA \[[@pbio-0030135-b20]\], and while we were able to detect very low levels of expression in the thymus of 3-d-old mice, we were unable to detect its expression in any adult tissue or embryonic stage using the commercially available expression panel. *Q8C9S7* could not be found and/or there was inconsistent annotation in other expression panels \[[@pbio-0030135-b29],[@pbio-0030135-b30]\]. Furthermore, this gene appears to be homologous to an annotated human pseudogene (AF003529.2) \[[@pbio-0030135-b20]\]. Because genes of unknown function with restricted patterns of expression did not present strong candidates for the causative factor underlying a QTL with a general effect on growth rate in both sexes, we did not pursue these two genes further.
Expression of a Gene Downstream from *Gpc3* {#s2e}
-------------------------------------------
To investigate the pathways through which *Gpc3* might exert its effect, we examined the expression of *Smad6 (mothers against decapentaplegic homolog 6); Gpc3* has been shown to affect BMP-7 (Bone morphogenic protein 7) signalling \[[@pbio-0030135-b31]\], which in turn promotes the expression of *Smad6* \[[@pbio-0030135-b32]\]. However, *Smad6* transcript levels did not differ significantly between genotypes in newborn liver or kidney (data not shown), suggesting that *Gpc3* exerts its effect through a different pathway. Glypican-3, the protein encoded by *Gpc3,* has been shown to bind to FGF-2 (fibroblast growth factor 2) \[[@pbio-0030135-b32],[@pbio-0030135-b33]\]. Therefore, the lower *Gpc3* levels in mice with the high-line allele may lead to higher levels of unbound FGF-2 or other growth factors that may promote growth. However, insulin-like growth factors do not appear to be targets of *Gpc3* binding \[[@pbio-0030135-b33],[@pbio-0030135-b34]\].
Polymorphisms in Non-Coding DNA Adjacent to *Gpc3* {#s2f}
--------------------------------------------------
To identify candidate polymorphisms that might be responsible for the difference in *Gpc3* transcript levels, we sequenced the 5′ and 3′ untranslated regions (UTRs), 2,876 bp upstream from the 5′ UTR of *Gpc3* (including its promoter region \[[@pbio-0030135-b35]\]), 1,724 bp downstream of the 3′ UTR, the first 1,048 bp of intron 1, as well as 3,377 bp of other regions of intron 1 that were identified as having high conservation with human. These are the non-coding regions near genes that show the highest levels of sequence conservation in rodents \[[@pbio-0030135-b36]\]. The only sequence differences between the high- and low-line-derived regions were three mononucleotide repeat polymorphisms (one in the first intron of *Gpc3* and two in the 3′ UTR), two dinucleotide repeat polymorphisms downstream from *Gpc3,* and a single nucleotide polymorphism 1,455 bp downstream of the 3′ UTR (see [Table S1](#st001){ref-type="supplementary-material"}). This low level of polymorphism is consistent with previous findings \[[@pbio-0030135-b14]\] and a low frequency of microsatellite polymorphism between the lines. The 3′ UTR polymorphisms present strong candidate polymorphisms for the differential expression of *Gpc3,* since 3′ UTRs are known to play a role in mRNA stability \[[@pbio-0030135-b37],[@pbio-0030135-b38]\]. For instance, the 3′ UTR of *dally,* a *Drosophila* member of the glypican family, affects the mRNA levels of this gene \[[@pbio-0030135-b39]\]. Furthermore, the polymorphic segments of the 3′ UTR show high conservation across mammals ([Figure 3](#pbio-0030135-g003){ref-type="fig"}A and [B](#pbio-0030135-g003){ref-type="fig"}). A BLAST search of a 450-bp region surrounding the downstream single nucleotide polymorphism yielded hits in the region of *Gpc3* in both human and rat, and indicated that this base pair is also conserved across these species ([Figure 3](#pbio-0030135-g003){ref-type="fig"}C). While the 3′ UTR polymorphisms are promising candidates, it should be noted that the causative polymorphism(s) may be located further upstream or downstream than was sequenced, or in an intron (e.g., \[[@pbio-0030135-b04]\]).
::: {#pbio-0030135-g003 .fig}
Figure 3
::: {.caption}
###### Polymorphisms between High- and Low-Line-Derived Chr X
\(A) 80 bp from the stop codon of *Gpc3* in the 3′ UTR.
\(B) 332 bp from the stop codon of *Gpc3* in the 3′ UTR.
\(C) 1,455 bp downstream of *Gpc3*.
Sequence in common with reference mouse sequence \[[@pbio-0030135-b20]\] is denoted by dots.
:::
:::
Pleiotropic Effects of Altered *Gpc3* Expression {#s2g}
------------------------------------------------
Knock-out mutations of *Gpc3* generate a range of pathological phenotypes, and it might be expected that QTL-associated regulatory variation at *Gpc3* would generate milder forms of these pleiotropic effects. We therefore conducted post-mortem and histological analyses on a sample of 34 age- and sex-matched individuals. Some of the most prominent pathological conditions of *Gpc3*-deficient mice are cystic and dysplastic kidneys, imperforate vaginas leading to swelling of the perineum and fluid-filled uteri, and susceptibility to respiratory infections \[[@pbio-0030135-b25]\]. However, there was no evidence of cystic medullary dysplasia resembling that seen in the *Gpc3*-deficient phenotype in mice carrying the high- or low-line allele. Although a range of incidental and pathological features were recorded ([Table S3](#st003){ref-type="supplementary-material"}), no phenotype was consistently associated with either genotype.
Since *Gpc3*-deficient mice have a reduced survival probability to weaning \[[@pbio-0030135-b25]\], we compared the numbers of high- and low-genotype mice surviving to weaning age in segregating litters. There is no evidence of a significant effect of genotype on numbers of high- and low-allele mice at weaning (512 and 554, respectively; χ^2^ ~1~ = 1.65; *p* = 0.2). For litter size, congenic females homozygous for the high-line QTL allele have somewhat higher performance than females homozygous for the low-line allele (mean ± standard error, 5.38 ± 0.22 versus 4.78 ± 0.18, respectively; *t* ~198~ = 2.13; *p* = 0.03).
Conclusions {#s2h}
-----------
In this study, we fine-mapped a growth QTL to a region containing only two genes of known function, found no coding sequence variation in these two genes, and demonstrated significant differences in the transcript levels of *Gpc3*. The phenotypic and expression differences between QTL genotypes are consistent with known loss-of-function mutations and knock-out phenotypes (i.e., reduced or absent *Gpc3* expression leads to increased body size). These results underscore the potential impact of relatively small changes in expression levels on phenotype.
Our results show that a gene underlying a Mendelian disease in humans can contribute to quantitative variation in mice. Unlike loss-of-function mutations, allelic variation in *Gpc3* had no pathological side-effects that we were able to detect; it affected growth rate only, and did so at all ages and in all tissues that we studied \[[@pbio-0030135-b27]\]. This work provides further evidence that the glypicans are involved in normal growth processes in addition to their role in Simpson--Golabi--Behmel syndrome and a variety of cancers \[[@pbio-0030135-b32]\].
Materials and Methods {#s3}
=====================
{#s3a}
### Experimental mice {#s3a1}
The inbred low line and a congenic for a high-line segment of Chr X were described previously \[[@pbio-0030135-b14]\]. We continued marker-assisted backcrossing to the low line to produce an interval-specific congenic strain containing a 14-cM segment of Chr X from the high line on the low-line background, with a contribution from high-line autosomes of less than 0.1%. The mice used in this study were at backcross generation 10--12. All experiments were carried out in accordance with U.K. Home Office regulations.
### Progeny testing {#s3a2}
Heterozygous females from the interval-specific congenic strain were crossed with low-line males, and mice recombinant between DXMit226 and DXMit68 were used for progeny testing. Recombinant males and females were crossed with low-line mice to produce families that segregated for the recombinant segment. Body weights at 6 wk of age from the progeny were recorded and flanking markers genotyped. Further genotyping using a range of microsatellite markers established the recombination breakpoints; microsatellite primer sequences are available in [Table S4](#st004){ref-type="supplementary-material"}. PCR genotyping was carried out on DNA extracted from ear clip or tail clip samples \[[@pbio-0030135-b14]\].
### Maximum likelihood analysis {#s3a3}
The marker allelic states and phenotypes of the progeny test dataset were analysed by maximum likelihood interval mapping \[[@pbio-0030135-b14]\]. Briefly, each recombination event was assumed to have been replicated across litters, and the phenotypic and flanking marker data at a given chromosomal position were used to estimate a hemizygous effect in males; homozygous and heterozygous effects in females; normally distributed litter effects; and effects for litter size, parity, and sex. Likelihood ratio for the model with a QTL relative to that for the reduced model with no QTL was calculated every 0.1 cM in the region of interest, and converted to a LOD score. There were 937 males and 972 females in the dataset.
### Post-mortem and histology analysis {#s3a4}
A total of 34 mice, matched for genotype and sex, were sacrificed between 8 and 16 wk of age and immediately underwent a comprehensive post-mortem and histological investigation. Tissue samples were fixed in 10% phosphate-buffered formalin and processed routinely. Sections were cut at 4 μm and stained with haematoxylin and eosin. Samples of all major organ systems were examined (urinary, cardiovascular, respiratory, alimentary, endocrine, reproductive, haemolymphatic, integumentary, musculoskeletal, and central nervous systems). Standard histopathological analysis was carried out and morphologic abnormalities recorded (see [Table S3](#st003){ref-type="supplementary-material"}).
### DNA sequencing {#s3a5}
Sequencing was carried out in forward and reverse directions using DYEnamic ET Terminator Cycle Sequencing Kits (Amersham Biosciences, Little Chalfont, United Kingdom) on an ABI Prism 3730 DNA Analyzer (Applied Biosystems, Foster City, California, United States) according to manufacturer\'s instructions. Sequencing primer sequences are shown in [Table S5](#st005){ref-type="supplementary-material"}. *Gpc3* is a large gene with almost 340 kb of intronic sequence. We therefore sequenced only a subset of the intronic regions, focusing on regions with high sequence conservation between mouse and human to increase the likelihood of finding functional sequences. Conserved regions were identified using the "Detailed view" of ContigView at the Ensembl Web site \[[@pbio-0030135-b20]\] (displayed using the "Compara" menu).
### RT-PCR {#s3a6}
Transcript levels were examined in kidney and liver from 47 newborn mice from seven litters that were segregating for the QTL region (23 low-allele mice and 24 high-allele males or heterozygous females). Tissue samples were collected into RNAlater solution (Qiagen, Valencia, California, United States) and stored at −20 °C until required. Total RNA was isolated from tissue using Qiashredder homogenisers (Qiagen) and RNAEasy Extraction kits (Qiagen) according to manufacturer\'s instructions. We performed RT-PCR using One Step RT-PCR kits (Qiagen) with the addition of RNAsin RNase inhibitor (Promega, Madison, Wisconsin, United States). Reaction conditions were optimised for each gene tested and for each tissue type to ensure the PCR reactions did not reach saturation. Specifically, we determined the number of PCR cycles and starting RNA concentration such that the amount of product varied linearly with RNA concentration. RT-PCR primer sequences are provided in [Table S6](#st006){ref-type="supplementary-material"}, and RT-PCR conditions are listed in [Table S7](#st007){ref-type="supplementary-material"}. To check for DNA contamination, 5 μl of each RT-PCR product was run out on a 1% agarose gel.
Although no splice variants of *Gpc3* or *Gpc4* are known, we designed three primer pairs for each gene (for *Gpc3,* these spanned introns 2, 3, and 7; for *Gpc4,* these spanned introns 1, 3, and 8). For both genes, all three primer pairs yielded products of the expected size from "high" genotype RNA. Furthermore, because we sequenced the coding region using cDNA (see [Table S5](#st005){ref-type="supplementary-material"}), we know that the entire genes are expressed for both alleles. Quantitative measurement of expression levels was performed using only one of the primer pairs per gene (see [Table S6](#st006){ref-type="supplementary-material"}).
### RT-PCR product quantification by DHPLC {#s3a7}
RT-PCR products were quantified using a WAVE denaturing high-pressure liquid chromatography instrument at an oven temperature of 50 °C. We sampled 5 μl of each RT-PCR product on a DNASep column. Samples were eluted from the column using an acetonitrile gradient in a 0.1 M triethylamine acetate buffer (pH 7), at a constant flow rate of 0.9 ml min^−1^. The gradient was created by mixing eluent A (0.1 M triethylamine acetate and 0.1 M tetrasodium EDTA) and eluent B (25% acetonitrile in 0.1 M triethylamine acetate) according to the manufacturer\'s specifications (Transgenomic, Omaha, Nebraska, United States). Each litter of mice was measured for all three genes in one assay to eliminate variation due to differences between runs. Transcript levels of *Gpc3* and *Gpc4* were expressed relative to that of *β-actin* by dividing the amount of *Gpc3* or *Gpc4* product by that of *β-actin*. Because the RT-PCR and quantification provided only an index of transcript levels, these are in arbitrary units. All samples were analysed in triplicate and the average within-assay coefficient of variation was less than 5%.
### Expression of genes of unknown function {#s3a8}
The tissue distribution patterns of expression of *Q8C9S7* and *Q9D9G4* were surveyed using mouse Rapid-Scan Gene Expression panels (OriGene Technologies, Rockville, Maryland, United States; MSCB101) that included cDNA from brain, heart, kidney, spleen, thymus, liver, stomach, small intestine, muscle, lung, testis, skin, adrenal gland, pancreas, uterus, prostate gland, breast (virgin, pregnant, lactating, and involuting), and embryo (e8.5, e9.5, e12.5, and e19). The kits were used in accordance with manufacturer\'s instructions using an initial activation of 3 min at 94 °C; followed by 35 cycles of 30 s at 94 °C, 30 s at 55 °C, and 2 min at 72 °C; and a final extension of 5 min at 72 °C (primer sequences are provided in [Table S6](#st006){ref-type="supplementary-material"}).
Supporting Information {#s4}
======================
Table S1
::: {.caption}
###### Sequence Polymorphisms between the High-Growth Line, Low-Growth Line, and Reference Mouse Sequence \[[@pbio-0030135-b20]\]
(41 KB DOC).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S2
::: {.caption}
###### Transcript Levels (Expressed as Ratio of *β-actin* Levels) in Newborn Mice from Litters Segregating for the QTL
Values are least squares means (LSM) and standard errors (SE) from a general linear model including litter, sex, genotype, and sex-by-genotype interaction.
(40 KB DOC).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S3
::: {.caption}
###### Summary of Post-Mortem and Histological Investigations
All females were homozygous high or low (i.e., not from segregating litters as in the expression and mapping studies).
(65 KB DOC).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S4
::: {.caption}
###### Microsatellite Primer Sequences
Physical positions are from Ensembl Build 24.33.1 \[[@pbio-0030135-b20]\].
(56 KB DOC).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S5
::: {.caption}
###### Sequencing Primer Sequences
Primers for *Gpc3* and *Gpc4* were designed to span introns and were used to sequence cDNA. All other primers were designed for direct sequencing of genomic DNA.
(79 KB DOC).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S6
::: {.caption}
###### RT-PCR Primer Sequences
(32 KB DOC).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S7
::: {.caption}
###### RT-PCR Conditions
In all cases, the RT-PCR consisted of an initial reverse transcription of 30 min at 50 °C; initial activation of 15 min at 95 °C; a variable number of cycles of 1 min at 94 °C, 1 min at 55 °C, and 1 min at 72 °C; and a final extension of 10 min at 72 °C.
(35 KB DOC).
:::
::: {.caption}
######
Click here for additional data file.
:::
Accession Numbers {#s4a8}
-----------------
The LocusLink (<http://www.ncbi.nlm.nih.gov/LocusLink/>) accession numbers for the genes and gene products discussed in this paper are *β-actin* (LocusLink ID 11461), *BMP-7* (LocusLink ID 12162), *dally* (LocusLink ID 39013), FGF-2 (LocusLink ID 14173), *Gpc3* (LocusLink ID 14734), *Gpc4* (LocusLink ID 14735), *Smad6* (LocusLink ID 17130), and *Socs2* (LocusLink ID 216233). The Vega Gene ID (<http://www.ensembl.org/>) for human pseudogene AF003529.2 is OTTHUMG00000022447.
We thank Pauline Baird, Jill Lovell, and Steven Macaskill for technical assistance; Mark Dorris for advice on the quantification of RT-PCR products; Bill Hill, Kellie Rance, and three anonymous reviewers for helpful comments on the manuscript; Lutz Bünger for the image of the high and low mice; and the Biotechnology and Biological Sciences Research Council and the Medical Research Council for funding.
**Competing interests.** The authors have declared that no competing interests exist.
**Author contributions.** FO, JKC, SDMB, PD, and PDK conceived and designed the experiments. FO, JKC, XL, SR, VV, and CD performed the experiments. FO, JKC, and PDK analyzed the data. FO, JKC, and PDK wrote the paper.
Citation: Oliver F, Christians JK, Liu X, Rhind S, Verma V, et al. (2005) Regulatory variation at glypican-3 underlies a major growth QTL in mice. PLoS Biol 3(5): e135.
Chr
: chromosome
QTL
: quantitative trait locus
UTR
: untranslated region
|
PubMed Central
|
2024-06-05T03:55:55.056458
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1073695/",
"journal": "PLoS Biol. 2005 May 5; 3(5):e135",
"authors": [
{
"first": "Fiona",
"last": "Oliver"
},
{
"first": "Julian K",
"last": "Christians"
},
{
"first": "Xiaojun",
"last": "Liu"
},
{
"first": "Susan",
"last": "Rhind"
},
{
"first": "Vinesh",
"last": "Verma"
},
{
"first": "Claire",
"last": "Davison"
},
{
"first": "Steve D. M",
"last": "Brown"
},
{
"first": "Paul",
"last": "Denny"
},
{
"first": "Peter D",
"last": "Keightley"
}
]
}
|
PMC1073696
|
Born on July 5, 1904, in Kempten in southern Germany, Ernst Mayr passed away peacefully at the Methuselah-like age of 100 on February 3, 2005, in Bedford near Cambridge, Massachusetts. Mayr was, by the accounts of his Harvard colleagues the late Stephen Jay Gould and Edward O. Wilson, not only the greatest evolutionary biologist of the 20th century, but even its greatest biologist overall. Thomas Henry Huxley was dubbed "Darwin\'s bulldog" for fighting for the acceptance of Darwinian ideas soon after their inception in the last decades of the 19th century. Similarly, Ernst Mayr has been called "Darwin\'s apostle" or the "Darwin of the 20th century" for promoting and dispersing Darwin\'s hypotheses throughout the past century.
Mayr lived for a century and accomplished more than several lifetime\'s worth of science in different biological disciplines. Brought up by parents who loved nature and who took the young Ernst on long hikes, he was exposed to natural history early on, but although birds were his passion all his life, he was, like Darwin, first compelled to study medicine. He began his studies at Greifswald---a prime birding spot---and through the chance observation of a rare species of duck that had not been seen in Germany for many years, he came in contact with the Berlin ornithologist Erwin Stresemann, who proposed that he switch to biology. Mayr abandoned medicine for biology and published his first scientific paper (of a total of almost 700) at the age of 19 in 1923, receiving his Ph.D. from Humboldt University in Berlin after only 16 months of graduate work and dissertation research; he was just 22. Ernst Mayr\'s last book (of a total of 25) was published in August 2004, a month after he turned 100 \[[@pbio-0030152-b1]\].
In 1931, thinking that he would not be offered a permanent post in Germany, he moved from Berlin to the American Museum of Natural History in Manhattan. In New York he called himself an ornithologist, and believed then, like many of his contemporaries, in Lamarkian inheritance. Sent by his advisor Erwin Stresemann from Berlin and financed by Lord Rothschild, he had just returned from over two years of perilous fieldwork in New Guinea and the Solomon Islands. The parallels to the lives of Darwin and Wallace may not be coincidental. During these expeditions, forlorn, at times given up for dead, exposed to tropical diseases and the danger of headhunters, he collected the skins of thousands of specimens, eating the flesh of many. Mayr was not only the ornithologist who probably tasted the largest number of different species of birds, but he also named 26 new species and over 400 new subspecies, more than any other taxonomist. In over 300 publications throughout his life, he discussed and described the geographic variation and distribution of birds and he also edited the last eight volumes of the *Checklist of the Birds of the World*. His main occupation during his 20 years at the American Museum was to curate and research the 280 000 specimens of the Rothschild collection.[](#pbio-0030152-g001){ref-type="fig"}
::: {#pbio-0030152-g001 .fig}
::: {.caption}
###### Ernst Mayr in 1994, after receiving an honorary degree at the University of Konstanz
(Photo: University of Konstanz)
:::
:::
In the 1930s, Mayr\'s friendship and interactions with the Russian-born Columbia University population geneticist Theodosius Dobzhansky, author of the landmark text *Genetics and the Origin of Species* published in 1937 \[[@pbio-0030152-b2]\], started to influence his thinking. Mayr\'s interests subsequently began to diversify beyond taxonomy into evolutionary biology, and this expansion of his interests culminated in his first, and possibly still most important book, *Systematics and the Origin of Species*, published in 1942 \[[@pbio-0030152-b3]\]. This was his main contribution to the so-called Modern Synthesis of the 1930s and 1940s, a scientific sea change that came about largely through the contributions of Mayr and Dobzhansky and other scientists such as Ronald A. Fisher and George G. Simpson. Mayr\'s first book combined insights and methods from paleontology, population genetics, systematics, and natural history, thus providing a unified modern evolutionary theory. Patterns and processes in natural populations would now be seen as consistent with Darwinian natural selection and Mendelian mechanisms of inheritance, and the behavior of genes in populations came to be understood through laboratory population genetic experiments and theoretical mathematical predictions. Mayr was the last survivor, and historical eyewitness, among the architects of the Modern Synthesis.
Ernst Mayr had many fundamental insights into evolutionary biology, and almost every topic of importance in evolution was advanced by his ideas. Perhaps his most widely known contribution is to the current notion of what constitutes a species. Darwin did not think that species were real in the philosophical sense, but rather that they were the result of the human predilection to perceive discontinuity among continuously varying individuals. Most biologists nowadays disagree with Darwin\'s view of species, largely because of Mayr\'s "biological species concept". Together with Dobzhansky, Mayr developed this definition of species "as groups of interbreeding populations in nature, unable to exchange genes with other such groups living in the same area" \[[@pbio-0030152-b2],[@pbio-0030152-b3]\]. Barriers to gene flow between species---termed reproductive isolating mechanisms---keep biological species distinct through processes such as species-specific mate choice and hybrid sterility. Although there are theoretical and operational problems with the biological species concept (e.g., it does not apply to asexually reproducing organisms such as bacteria), it is still, by far, the most widely used species concept among the 20 or so competing definitions that have been proposed in the past several decades. Students of biology all over the world have memorized Mayr\'s definition of species for more than half a century.
The biological species concept made it possible to study how species arise, since the criterion of reproductive isolation provided a scientifically rigorous litmus test. The origin of species is a topic to which Darwin himself, in spite of the promising title of his famous book, did not say all that much. Mayr\'s understanding of the biogeographic distributions of bird species, overlaid with extensive knowledge about variation in morphology, led him to develop concepts about the geographic mechanisms of speciation---cornerstones for those studying speciation today. The geographic separation of populations, such as by rivers or valleys, he argued, prohibits homogenizing gene flow between them. If such isolated (termed allopatric) populations accumulate mutations over time, this might lead to the divergence of such populations from each other, and reproductive isolation might arise as a simple byproduct of these separate evolutionary histories. Mayr staunchly defended this idea during sometimes heated debates and further developed it and other hypotheses regarding geographic mechanisms of speciation over many decades (outlined in depth in the 797 pages of *Animal Species and Evolution* \[[@pbio-0030152-b4]\]).
One mechanism of speciation in particular is still contested (see \[[@pbio-0030152-b5]\] and \[[@pbio-0030152-b6]\]). Mayr called it "peripatric speciation" or "founder-effect speciation." And it is an idea that Ernst Mayr was particularly fond of. He believed it to be his most important contribution to evolutionary biology. This model was again inspired by Mayr\'s own natural history observations. He noted that on some New Guinea islands, populations of birds differed markedly from individuals of the mainland population. He reasoned that this differentiation and speciation could result from a small number of individuals founding the island population. By bringing only a subset of all the genes of the main population (causing a genetic bottleneck), genetic drift (random fixation) and natural selection (due to a different set of selection pressures on these islands) would not only promote the formation of new species but would do so rapidly. This mechanism might also account for the paleontological pattern called "punctuated equilibrium," which was proposed by Nils Eldredge and Gould in 1972 \[[@pbio-0030152-b7]\]. They noted that long periods of morphological stasis were sometimes interrupted (punctuated) by short periods of drastic phenotypic change in the fossil record. Somewhat ironically, Mayr, who considered himself a "gradualist" all his life, seems to have also provided a mechanism for variability in rates of evolution along evolutionary lineages.
After establishing the Society for the Study of Evolution and serving as the first editor of its journal, *Evolution*, Mayr moved to Harvard University in 1953 as the Alexander Agassiz Professor of Zoology and curator of birds at the Museum of Comparative Zoology. By this time, one surely would have labeled him primarily an evolutionary biologist rather than an ornithologist. His interests expanded even further into the theory of systematics---another field to which he made many contributions (see *Principles of Systematic Zoology* \[[@pbio-0030152-b8]\]). A lifelong and, it seems fair to say, futile fight with the then emerging idea of cladistics in systematic biology began. Ernst Mayr also served as director of the Museum of Comparative Zoology before his retirement in 1975 and oversaw the building of a new addition to the museum, whose library was renamed after him ten years ago.
What of his retirement? Mayr published 14 of his 25 books in the 30 years that followed after his official retirement. During the last two decades of his life, Mayr began to think and write more about the history and philosophy of biology. His most important work of this period was *The Growth of Biological Thought* \[[@pbio-0030152-b9]\], a monumental 974 pages. Here, and in later books and publications (he also founded the *Journal of the History of Biology*), he laid out why he thought that the philosophy of biology is an autonomous science that differs fundamentally from the philosophy of science, which, Mayr implied, was largely derived from physics. He argued that biology is a science that is based on historical contingency as well as on many unpredictable and coincidental factors that make it impossible to discover laws. Rules, not laws, are all that one will be able to find in biology.
Clearly, Ernst Mayr felt very strongly that he had something of importance to say to the world. And the world, not only in its scientific realms, seemed to think likewise. He received almost 20 honorary degrees from major universities, was a member of more academies than any other scientist before him, and received most of the prizes that could possibly be awarded to a biologist, including the Japan Prize, the Balzan Prize, and the Crafoord Prize, the "Nobel Prize for ecologists and evolutionary biologists."
How could one person possibly fit so much into one lifetime, even such an astonishingly long one? For a start, he was a man of stringent self-discipline, who would get up with (or before) the birds, like a good ornithologist should. Writing (longhand or dictation) was done mostly in the mornings, and long walks were part of every day, as were extended periods of reading and corresponding with his colleagues. Just like Darwin, Mayr wrote thousands of letters minding the business of others, telling his fellow scientists what he thought of their work, praising them but also advising them on missed literature and new directions for further study. He did not like to be bothered with those other menial things that also belong to living on this planet, and, luckily for him, Gretel, his wife of 55 years, mostly took care of that part. So after her death ten years ago, when he was in his early 90s, he had to learn how to cook a hamburger for himself.
Ernst was gifted with an astonishing clarity of thought. Something that always impressed and humbled me was that the transcripts of his dictated manuscripts required very little further editing. Even Mayr\'s native-English-speaking competitors praised him, obviously a nonnative, for his lucid and clear writing style. Ernst also had the ability to store an astonishing amount of information drawn from many different sources---his memory was spectacular. His ability to synthesize ideas, combined with an amazing recall of natural history, an exact visual memory, and an overall wide scientific horizon, was awe-inspiring and, more than in anyone else I\'ve ever met, produced a plethora of novel ideas. His vitality was also legendary. He would still climb trees in his mid-80s to inspect birds\' nests, and he bought his last new car after having passed the age of 90, much to the astonishment of the car salesman, as he once told me.
Pulitzer Prize winner Natalie Angier from the New York Times once described Ernst Mayr in an interview as "opinionated and elitist, courtly and generous." Ernst Mayr was all that. He was strong-willed, had little patience for people who had not done their homework or talked without having their natural history straight or their line of reasoning well thought out, and he could be damning in his judgment of the ideas of others. Yet he generously shared not only his thoughts, but also charitably donated a good portion of his salary and most of his significant prize money to causes such as the Nature Conservancy and to endow prizes for young evolutionary biologists.
Although Ernst Mayr lived only about a tenth of the 969 years that Methuselah is purported to have lived, he still accomplished much more than one might expect to get done, even in a 100 years. More important than his scientific output is the overarching influence he has had on the thinking of three or four generations of biologists that he was a contemporary of and interacted with. The scientific world lost a giant and an inspirational thinker. His doctoral advisor Stresemann once called Mayr a rising star. Nobody since Darwin shed the light of insight as bright over the firmament of evolutionary biology as Ernst Mayr did. His star has not stopped shining, and his ideas will continue to live on for generations of young evolutionary biologists to come.
On the occasion of his 100th birthday Mayr published an article in Science \[[@pbio-0030152-b10]\] looking back over eight decades of research in evolution that he closed with the following words: "The new research has one most encouraging message for the active evolutionist: it is that evolutionary biology is an endless frontier and there is still plenty to be discovered. I only regret that I won\'t be present to enjoy these future developments."
**Citation:** Meyer A (2005) On the importance of being Ernst Mayr. PLoS Biol 3(5): e152.
Axel Meyer is Professor of Zoology and Evolutionary Biology at the University of Konstanz in Germany. E-mail: <axel.meyer@uni-konstanz.de>
|
PubMed Central
|
2024-06-05T03:55:55.059021
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1073696/",
"journal": "PLoS Biol. 2005 May 5; 3(5):e152",
"authors": [
{
"first": "Axel",
"last": "Meyer"
}
]
}
|
PMC1073697
|
Protecting the liver from infection presents the immune system with a tough challenge. Because it is the first stop for food absorbed from the gut, the liver is constantly bathed in a rich broth of mostly harmless foreign molecules, which need to be immunologically tolerated. On the other hand, the slow rate of blood flow and high internal surface area makes the liver\'s many sinuses ideal sites for infections to take hold. One unusual type of immune cell patrolling this complex maze in the liver is the natural killer T (NKT) cell. In this issue, Dan Littman, Mike Dustin, and colleagues shed some light on the behavior of this little-appreciated and important hepatic guardian, showing its unexpected ability to perform immune surveillance entirely within the vasculature. NKT cells crawl along vascular passages in the liver (hepatic sinusoids) at high speeds for a cell moving under its own power, but slow speeds compared to blood flow, and never or rarely venture into the surrounding tissue.
To track the movements of these cells, the authors used a mouse in which the gene for the CXCR6 receptor was replaced with a gene for green fluorescent protein (GFP). In these mice, cells that express CXCR6 also express GFP, and because they fluoresce, they can be visualized and tracked under a microscope. The study showed that cells remained entirely within the blood vessels, crawling along the endothelial lining at approximately 16 microns per minute. Despite directional blood flow, the NKT cells moved and changed directions randomly, and could occasionally be seen migrating past one another in opposite directions within a single vessel. When the authors injected antigen into the blood stream, the cells abruptly stopped, and remained stationary, suggesting that they had fulfilled their first duty to find antigen and were beginning their next: to alert the rest of the immune system. Surprisingly, antigen detection occurred within the blood stream---it is normally thought to occur after lymphocytes leave the blood.[](#pbio-0030154-g001){ref-type="fig"}
::: {#pbio-0030154-g001 .fig}
::: {.caption}
###### Tracking the movment of glowing T cells in the vasculature
:::
:::
While they make up less than 1% of lymphocytes in other tissues, NKT cells comprise 30% of the lymphocytes in the liver. Mice lacking both copies of the *cxcr6* gene had only one third the normal number of hepatic NKT cells. Though chemokine receptors are typically thought to function by directing cell homing, the authors observed that CXCR6-deficient cells died rapidly in vitro, suggesting that CXCR6 delivers survival signals and thereby controls the number of hepatic NKT cells and liver immune surveillance in general.
While cell numbers were reduced, the movement of the remaining cells in the homozygous-deleted mice was no different than that in the heterozygotes. Because of this, the authors propose that cell motility in wild-type mice is likely to be similar to what they observed in these experiments. Based on the speed and density of the NKT cells, they calculate that each cell can visit a new hepatocyte every two minutes, and that every hepatocyte is surveyed approximately once every 15 minutes. This is in stark contrast to surveillance in the lymph nodes, in which a typical dendritic cell receives 5,000 visits per hour from its resident T cells, but it may make biological sense given that NKT cells have a much more restricted range than conventional T cells.
|
PubMed Central
|
2024-06-05T03:55:55.060639
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1073697/",
"journal": "PLoS Biol. 2005 Apr 5; 3(4):e154",
"authors": []
}
|
PMC1073698
|
The presence of a small number of discrete forms---as you find in a classic Mendelian trait like eye color---suggests that the phenotype is controlled by a very small number of genes. In contrast, a complex trait such as body size is influenced by multiple genes as well as environmental factors, giving rise to a continuous spectrum of phenotypes. This causal complexity makes discovery of the genetic determinants of the trait---so-called quantitative trait loci (QTLs)---very difficult. In this issue, Fiona Oliver, Julian Christians, and colleagues extend their work on a single QTL with a large effect on body size variation in mice, and show that the responsible gene is one already linked to a Mendelian growth disorder in humans.
In previous work, mice were divergently selected for large or small body size, revealing a QTL on the X chromosome that causes a 20% difference in growth rate. In this study, the authors further refined the map of the area to 660 kilobases, and found it contains only four genes: glypican-3 and glypican-4 (Gpc3 and Gpc4), and two other genes of unknown function. The glypicans are membrane-bound growth regulators; loss of function of Gpc3 in humans causes a rare syndrome of overgrowth, skeletal and other abnormalities, and neonatal death. Since none of the four genes showed coding-region variations that might explain the differences between the large and small mice, the authors examined expression levels. In Gpc3, but not the others, size correlated with a significant difference in gene expression: larger mice had low levels of the messenger RNA, and smaller mice had high levels—the same pattern seen in the human loss-of-function disorder. The authors identified several non-coding polymorphisms in Gpc3 that differed between the two forms, although it remains to be seen whether the differential effect on growth is due to these or other DNA differences nearby.[](#pbio-0030163-g001){ref-type="fig"}
::: {#pbio-0030163-g001 .fig}
::: {.caption}
###### Variation in the expression of a protein involved in a rare human overgrowth syndrome contributes to size variation in mice (Photo: Lutz Bünger, Scottish Agricultural College)
:::
:::
The results from this study point out an important feature of inheritance, namely, that a gene implicated in a Mendelian trait can also contribute to quantitative variation. While catastrophic expression failure, such as a loss of function, can cause disease, smaller changes in expression of the same gene may simply help fill out the bell curve of normal variation.
|
PubMed Central
|
2024-06-05T03:55:55.061163
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1073698/",
"journal": "PLoS Biol. 2005 May 5; 3(5):e163",
"authors": []
}
|
PMC1073699
|
The birth of combinatorial chemistry in the early 1990s held out the promise that scientists would soon synthesize trillions of compounds at a time and screen up to a million a day, revolutionizing the process of drug discovery. But synthesizing a vast library of compounds is just the first step in the historically painstaking process of determining whether a compound has the desired effect on a target. In addition to an ever-growing library of candidate therapeutic compounds, advances in genome analysis have produced a growing list of potential drug targets---drowning drug researchers in an excess of riches.
In a new study, Kevan Shokat and colleagues report a high-throughput screening method that substantially narrows the field of candidate therapeutic agents. Their approach takes advantage of a recently developed automated system (called Cytometrix) that combines advanced imaging and bioinformatics approaches to classify cells according to small-molecule-induced changes in cell size, shape, and structure (morphology). Their analysis identified a novel compound with promising potential as an anticancer agent.
The Cytometrix system offers a high-throughput, unbiased (that is, machine-rendered) approach to identifying molecules that induce changes in cell processes, molecules that could be used to probe cells or to test for therapeutic effect. High-tech imaging equipment, combined with statistical analysis, extracts the biological effects of small molecules as "phenotypic readouts" based on the physical and structural characteristics of the cells. Using this system, the authors tested 107 small-molecule compounds with structural similarities to four types of protein kinase inhibitors---used in anticancer therapies---by injecting them into human cancer cell lines (and one noncancerous cell line). The phenotypic readouts produced by each compound were classified based on a statistical analysis of cell morphology, staining intensity (staining aids visualization), and the spatial distribution of subcellular structures like nuclei, microtubules, and the Golgi compartments. This analysis could also identify inhibitors of cell components not targeted by known kinase inhibitors.
From the library of screened compounds, Shokat and colleagues identified a molecule (hydroxy-PP) that, though structurally related to a known kinase inhibitor, induced morphological changes distinct from any known kinase inhibitor. What does hydroxy-PP target? An enzyme, called carbonyl reductase 1 (CBR1), that acts on xenobiotics like anticancer drugs and is thought to cause the heart damage associated with daunorubicin chemotherapy.
To better understand how compound and enzyme interact, the authors solved the structure of hydroxy-PP and CBR1 bound together. Knowing their respective structures also suggests ways of enhancing a molecule\'s effect on a target. In this case, Shokat and colleagues used their structural analysis to increase hydroxy-PP\'s inhibition of CBR1 in cell culture so they could further explore the enzyme\'s biological function. These experiments revealed a previously uncharacterized role for CBR1 in programmed cell death.
Given the enzyme\'s suspected role in chemotherapy-related cardiotoxicity, inhibiting CBR1 activity might enhance the efficacy of chemotherapy treatments by reducing their debilitating side effects---a possibility that future studies can explore. But for now, Shokat and colleagues have demonstrated the power of using high-throughput image-based screening to identify small molecules both for probing cell biology and for identifying promising drug candidates.[](#pbio-0030165-g001){ref-type="fig"}
::: {#pbio-0030165-g001 .fig}
::: {.caption}
###### Two closely related compounds produced the morphological differences evident in these lung cancer cells and reveal biological activity that could be important for drug discovery (Photo: Cytokinetics, Inc)
:::
:::
|
PubMed Central
|
2024-06-05T03:55:55.061692
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1073699/",
"journal": "PLoS Biol. 2005 May 5; 3(5):e165",
"authors": []
}
|
PMC1073700
|
With exponential advances in computing power over the past ten years, data-generating capacity has far outpaced anyone\'s ability to mine the rich seams of information. This is especially true in the field of genomics. So far, over 222 prokaryote (bacteria) genomes have been sequenced, 21 archaea (primitive bacteria-like extremophiles), and 17 eukaryotes (from yeast to fly and rat to human), according to the Center for Biological Sequence Analysis in Denmark (<http://www.cbs.dtu.dk/services/GenomeAtlas/>). All these genomes promise to provide powerful insights into the biological processes of life, but such insights come with painstaking analysis by trained experts. Matching genotype to phenotype---the visible or measurable characteristics of species---is a major challenge in what Francis Collins, Director of the United States National Human Genome Research Institute, has called the "post-genomic era."
In a new study, Peer Bork and a team of bioinformatics-savvy molecular biologists tested a new approach to extracting biologically meaningful information from the massive MEDLINE database. The US National Library of Medicine\'s MEDLINE contains over 12 million abstracts from thousands of publications dating back to 1965. Combining automated literature mining with comparative genomics---which compares genome sequences of different organisms to discern differences and similarities in gene content---the authors conducted a systematic search for associations between genes and phenotypic traits. Their approach automates tasks that typically require human curation.
Recognizing that the best source of information on species phenotypic traits is the scientific literature where biologists describe them, the authors first ran a search to identify associations between species and traits in MEDLINE abstracts. Words that tended to occur with subsets of species, the authors reasoned, were more likely to reflect particular traits. From a total of 255,249 MEDLINE abstracts showing any connection to 92 prokaryotic species with sequenced genomes, 172,967 nouns showed meaningful associations related to the species\' traits. "Flagellum" and "motility" showed up more often in self-propelling species, for example, and "endosymbiont" aptly appeared with the intracellular bacteria (Buchnera aphidicola) that inhabits aphids.
Next, Bork and colleagues detected the presence or absence of over 200,000 evolutionarily conserved genes across the 92 species and sorted the results into species--word and species--gene groups. The analysis revealed a number of words and genes with similar distribution in related species, leading to over 2,700 significant associations between trait-descriptive words and orthologous (evolved from a common ancestor) groups of genes. These genes encode over 28,000 proteins. Many were already known---including genes involved in pathogenicity, biodegradation and biosynthesis, and photosynthesis---but many, the authors note, are "novel" or of "unexpected character and complexity."
And it is the ability to uncover unexpected relationships across numerous genes and genomes---patterns likely to escape human analysis---that makes this approach so powerful. Among these unexpected match-ups, Bork and colleagues linked a number of food and food-poisoning-related terms with metabolic-enzyme-coding genes. All 37 genes predicted to play a role in food spoilage and toxicity are present in food-borne pathogens but not in most other prokaryotes. By assigning functions to these previously uncharacterized genes, the authors could also assign new roles for pathways that use the genes. For example, by linking two genes with pathways that metabolize propanediol and ethanolamine---compounds found almost exclusively in highly hazardous food-borne pathogens---the authors predict that propanediol and ethanolamine pathways are "crucial genomic determinants of pathogenicity associated with food poisoning."[](#pbio-0030166-g001){ref-type="fig"}
::: {#pbio-0030166-g001 .fig}
::: {.caption}
###### Many predicted genes were tied to food poisoning and bacterial pathogens, such as Salmonella typhimurium (Photo: Volker Brinkmann, Max Planck Institute for Infection Biology, Berlin, Germany)
:::
:::
That their analysis linked so many predicted genes with bacterial pathogenicity might be expected, the authors note, since both genome sequencing and biological research are heavily focused on human health. Given the weekly increase in the number of genomes sequenced and in MEDLINE entries, the method outlined here should provide a valuable tool to help researchers narrow the gap between the promise and payoff of the genomic revolution.
|
PubMed Central
|
2024-06-05T03:55:55.062312
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1073700/",
"journal": "PLoS Biol. 2005 May 5; 3(5):e166",
"authors": []
}
|
PMC1073701
|
An important source of genetic novelty is the introduction of new genes. Since most genes in an organism\'s genome are under selective constraint, opportunities for the evolution of new gene functions---which in turn might confer selective advantage---most often arise when new genes enter the genome. In eukaryotes---a category that includes humans and rice---novel genes typically arise when existing genes undergo duplication. Extra copies of genes can be created when normal DNA replication hiccups and erroneously duplicates entire regions of DNA. These extra gene copies reside in species\' genomes for generations and might eventually mutate to code for novel proteins, adding new genes to the species\' repertoire. The new genes, along with the rest of the genome, are passed down from one generation to the next in a process known as vertical transmission.
In prokaryotes---which include unicellular organisms in the bacteria and archaea domains---novel genes can appear through multiple routes. In addition to gene duplication, prokaryote genomes can change when DNA fragments are taken up directly by cells, passed from cell to cell, or transferred to new cells with the help of viruses. All three scenarios provide a means for whole genes to move directly from one bacterial genome to another, a process called lateral gene transfer (LGT) or horizontal transmission.
Until now, the importance of vertical versus horizontal transmission in the evolution of any large prokaryote group was unknown. In a new study, Emmanuelle Lerat et al. capitalized on the availability of complete genome sequences within the diverse γ-Proteobacteria, a group of prokaryotes that includes Escherichia coli, Salmonella spp., and some nitrogen-fixing bacteria, to pursue that question.
Sorting out the issue is no simple task. If the same gene is present in more than one species, it could have been inherited from a common ancestor or it could have jumped from one lineage to another by LGT. Even if the same gene appears twice in one species\' genome, the copies could have different histories---one copy could have been acquired vertically from its ancestors, while the other could have come from a different species.
Though previous studies have looked at the distributions of genes across species phylogenies, information about gene origin appeared contradictory. To create a clearer picture, Lerat et al. accounted for the possibility of widespread LGT by statistically comparing the phylogenies of many different gene families with a benchmark phylogenetic tree that reflected the accepted evolutionary history of γ-Proteobacteria.
The authors found that LGT plays a substantial role in generating the diversity of genes found in γ-Proteobacteria genomes. Members of the group are constantly acquiring and losing genes, although the extent of LGT can vary greatly among species. In contrast, gene duplications play a much smaller role in explaining γ-Proteobacteria genome diversity, although duplications have been shown to be important for short-term adaptation.
Genes that have arrived by LGT within a single genome do not necessarily share a common history with each other. Many of the genes that are found only in a single genome and are not widely distributed across the γ-Proteobacteria were recently acquired from distant sources. Most of these acquired genes will likely be lost soon after joining a genome; those that persist are then inherited vertically. This helps to reconcile why gene trees tend to provide valid phylogenetic inferences about the relationships among different bacterial lineages, despite the potential mixing that could result from LGT. Phylogeneticists aiming to reconstruct a phylogeny for a group look at variations in genes distributed in the species, and these are largely vertically transmitted.
Lerat et al. propose that LGT is a common source of genes in γ-Proteobacteria because it has the potential to introduce functionally different genes into the genome with immediate contributions to fitness. Gradual evolution of gene duplicates doesn\'t provide the same type of immediate reward.
|
PubMed Central
|
2024-06-05T03:55:55.062876
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1073701/",
"journal": "PLoS Biol. 2005 May 5; 3(5):e169",
"authors": []
}
|
PMC1074340
|
Background
==========
HIV/AIDS continues to be a major public health problem worldwide with millions of people currently infected and new infections being on the rise. As no effective vaccines are currently available for prevention, new and innovative therapies need to be developed. Although combinatorial therapies such as HAART have proven to be effective in prolonging life, they do not afford a complete cure. Other constraints with HAART therapy are the development of drug resistant viral mutants and toxicity after prolonged therapy. Intracellular immunization by gene therapy strategies offers a promising alternative approach for controlling and managing HIV disease. A number of previous approaches that involved the use of transdominant proteins \[[@B1]-[@B3]\], decoys \[[@B3]-[@B7]\], and ribozymes \[[@B5],[@B8]-[@B12]\] had shown initial promise but fell short of practical utility in providing adequate protection. With the discovery that the RNA interference phenomenon operates in mammalian cells and is highly effective in selective gene silencing, new potent small interfering RNA (siRNA) molecules have become available to add to the anti-HIV arsenal \[[@B13]\].
RNAi is a highly potent mechanism of post-transcriptional gene silencing. Mediated by sequence specific siRNAs, it can effectively down regulate expression of either viral or cellular RNA target molecules by selective degradation of mRNAs \[[@B13]-[@B16]\]. Mechanism of destruction involves an endonuclease present in the RISC complex which is guided by the antisense component of the siRNA for target recognition. A number of reports have shown that delivery of siRNAs by transfection of presynthesized or plasmids encoding siRNAs into cultured cells can effectively inhibit HIV-1 infections \[[@B17]-[@B26]\]. Antiviral effects of these delivery methods are only transient due to eventual degradation and dilution of siRNAs during cell division. For HIV gene therapy strategies to succeed in long range, it is necessary that siRNA coding transgenes be maintained and expressed long term in a virus susceptible target cell. In this regard, lentiviral vectors have proven to be highly effective in high efficiency gene transduction and sustained gene expression.
A number of previous approaches using either synthetic siRNAs or plasmid expressed constructs have successfully targeted viral transcripts and achieved effective viral inhibition. Of these, some anti-HIV-1 siRNAs, such as siRNAs against tat, tat-rev had been introduced into lentiviral vectors and their efficacy was demonstrated both in cell lines and primary T cells and macrophages \[[@B27],[@B28]\]. Promising data was also obtained in experiments showing that anti-rev siRNAs against HIV-1 were functional in conferring viral resistance in differentiated T cells and macrophages derived from lentiviral transduced CD34+ hematopoietic progenitor cells \[[@B29]\].
In addition to targeting viral transcripts, many studies including ours also investigated the efficacy of siRNAs in down regulating host cell molecules necessary for HIV-1 infection \[[@B18],[@B21],[@B23],[@B24],[@B30],[@B31]\]. An advantage in targeting cellular molecules is that efficacy will be more broad spectrum against all the clades of the virus and the frequency of escape mutants will be lower. Down regulation of the primary cell surface receptor CD4 and consequent inhibition of HIV-1 infection was shown using synthetic siRNAs. However, since CD4 is an essential cell surface molecule for immunological function, it is not a practical target for HIV gene therapy. Chemokine receptors CCR5 and CXCR4 play critical roles as coreceptors for viral entry during infection with macrophage tropic R5 and T cell tropic X4 HIV-1 viral strains respectively \[[@B32],[@B33]\]. Thus they are suitable targets for siRNA mediated down regulation. Since both R5 and X4 viral strains are involved in disease pathogenesis, it is important to consider blocking of both respective coreceptors when developing effective therapeutics. In a segment of the human population, a naturally occurring 32-bp deletion in the CCR5 gene results in the loss of this coreceptor thus conferring significant resistance to HIV infection \[[@B34]-[@B36]\]. Homozygous or heterozygous individuals for this mutation remain physiologically normal. With regard to the CXCR4 coreceptor, it was found to be dispensable for T cell development and maturation in murine studies \[[@B37]\]. These findings suggest that CCR5 and CXCR4 are promising targets for HIV therapies.
Based on this rationale, recent work with synthetic siRNAs demonstrated that down regulating either CXCR4 or CCR5 will protect cells from X4 or R5 HIV-1 strains respectively at the level of viral entry \[[@B18],[@B21],[@B23],[@B24]\]. Although stable expression of an anti-CCR5 siRNA was achieved using a lentiviral vector in one study, down regulating CCR5 alone in the face of an HIV-1 infection is insufficient \[[@B31]\]. Therefore, we recently experimented with synthetic bispecific combinatorial constructs targeted to both CXCR4 and CCR5 and have shown their efficacy in cultured cells \[[@B24]\]. To make further progress, our present studies are directed towards constructing a single bispecific lentiviral vector expressing both CXCR4 and CCR5 siRNAs. Using this combinatorial construct, here we show high efficiency transduction, simultaneous down regulation of both coreceptors resulting in HIV-1 resistance.
Results and Discussion
======================
Coreceptor down regulation by a bispecific lentiviral vector
------------------------------------------------------------
Our major goal in these studies is to introduce both CXCR4 and CCR5 siRNAs into a single lentiviral construct to achieve their stable expression in transduced cells. Lentiviral vectors offer advantages over conventional retroviral vector systems since they can transduce dividing as well as nondividing cells and are less prone to transgene silencing \[[@B44]-[@B47]\]. The transfer vector HIV-7-GFP-XHR (referred to as XHR) contained a short hairpin type anti-CXCR4 siRNA driven by a Pol-III U6 promoter followed by a short hairpin anti-CCR5 siRNA driven by a different Pol-III promoter, H1. Downstream, the reporter gene, EGFP is driven by a CMV promoter. The control GFP-alone vector, HIV-7-GFP, contained only the reporter gene EGFP (Fig [1](#F1){ref-type="fig"}).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
[Bispecific lentiviral vector (XHR) encoding anti-CXCR4 and CCR5 siRNAs.]{.underline} A) Control transfer vector pHIV-7-GFP encoding a CMV promoter driven EGFP reporter gene. B) To derive the bispecific vector pHIV-XHR-GFP, a U6 promoter driven short hairpin CXCR4 siRNA cassette was cloned into the *BamH*I site upstream to the CMV-EGFP cassette. The H1-CCR5 siRNA cassette was inserted into an *Mlu*I site downstream to the U6-CXCR4 siRNA cassette.
:::
:::
Magi-CXCR4 cells constitutively expressing CXCR4 on the cell surface when transduced with the control vector or XHR vector had shown 97% and 83% EGFP expression respectively as measured by FACS analysis indicating high efficiency of transduction (Fig [2A](#F2){ref-type="fig"} and [2C](#F2){ref-type="fig"}). To determine if CXCR4 was down regulated by the respective siRNA in the XHR construct, the transduced cells were analyzed for CXCR4 surface expression. The surface levels of CXCR4 were reduced significantly in XHR transduced cells (73% lower) compared to the cells transduced with control vector (Fig [2B](#F2){ref-type="fig"} and [2D](#F2){ref-type="fig"}) indicating the efficacy of the CXCR4 siRNA on its target. Similarly, to determine the activity of the anti-CCR5 siRNA in the XHR vector, transduced Ghost R5 cells that constitutively express CCR5 were evaluated. As seen in Fig [3A](#F3){ref-type="fig"} and [3C](#F3){ref-type="fig"}, high levels of transduction (84% and 83%) were seen in Ghost-R5 cells with either the control vector or XHR vector, respectively. When the transduced cells were analyzed for CCR5 expression, a dramatic decrease in CCR5 expression was seen in XHR cells (72%) compared to control vector transduced cells (Fig [3B](#F3){ref-type="fig"} and [3D](#F3){ref-type="fig"}). These results had shown that the bispecific lentiviral vector XHR efficiently down regulates both CXCR4 and CCR5 targets in respective cells.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
[Cell surface down regulation of CXCR4 in XHR transduced Magi-CXCR4 cells.]{.underline} Magi-CXCR4 cells that constitutively express CXCR4 were transduced with control GFP or XHR vectors. Cells were stained with PECy5-conjugated antibodies to CXCR4 and analyzed by FACS 72 hours post-transduction. Levels of CXCR4 in non-transduced cells are superimposed (unshaded areas). Transduction efficiency was determined by FACS for EGFP expression. Levels of EGFP in control GFP-alone vector (A) and XHR vector (C) transduced cells. Levels of CXCR4 expression in GFP-alone (B) and XHR (D) vector transduced cells. Percent positive cells are indicated.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
[Cell surface down regulation of CCR5 in XHR transduced Ghost-R5 cells.]{.underline} Ghost-R5 cells that constitutively express CCR5 were transduced with GFP-alone or XHR vectors. Cells were stained with PECy5-conjugated antibodies to CCR5 and analyzed by FACS 72 hours post-transduction. Levels of CCR5 in non-transduced cells are superimposed (unshaded areas). Transduction efficiency was measured by FACS for EGFP expression. Levels of EGFP in control GFP-alone vector (A) and XHR vector (C) transduced cells. Levels of CCR5 expression in GFP-alone (B) and XHR (D) vector transduced cells. Percent positive cells are indicated.
:::
:::
Expression of siRNAs and down regulation of CXCR4 and CCR5 transcripts
----------------------------------------------------------------------
To confirm that the down regulation of both CXCR4 and CCR5 coreceptors as seen by FACS analysis is due to reduced levels of the corresponding mRNAs, vector transduced cells were analyzed by RT-PCR. As an internal control, GAPDH mRNA was also analyzed. XHR vector transduced cells showed considerable reduction in transcript levels for both CXCR4 and CCR5 as compared to non-transduced and control GFP vector transduced cells. The levels of GAPDH control mRNA remained unchanged in all samples (Fig [4](#F4){ref-type="fig"}). To validate the expression of individual siRNAs in transduced Magi-CXCR4 and Ghost R5 cells, cellular RNA was analyzed by northern analysis for their presence. As internal controls, the presence of constitutively expressed miRNA-16 RNAs were also analyzed in parallel. As expected, comparable levels of miRNA-16 RNAs (22 bp in length)were detected in GFP control vector transduced as well as in XHR vector transduced cells (Fig [5A](#F5){ref-type="fig"}). RNAs corresponding to CXCR4 and CCR5 shRNAs (representing the 21nt antisense strand of each shRNA) were seen in XHR transduced but not in GFP control vector transduced cells (Fig [5B](#F5){ref-type="fig"}).
::: {#F4 .fig}
Figure 4
::: {.caption}
######
[RT-PCR detection of CXCR4 and CCR5 mRNA down regulation.]{.underline} Total RNA was extracted from vector transduced cells and one-step RT-PCR was performed. PCR products of 450 bp were amplified to detect the coreceptor transcripts. A) Levels of CXCR4 mRNA in non-transduced (lane 1), GFP-alone (lane 2), and XHR (lane 3) vector transduced Magi-X4 cells. B) CCR5 transcript levels in non-transduced (lane 1), GFP-alone (lane 2), and XHR vector transduced Ghost-R5 cells. GAPDH transcript levels were used as internal controls (PCR product size \~550 bp).
:::
:::
::: {#F5 .fig}
Figure 5
::: {.caption}
######
[Northern analysis to detect siRNA expression in transduced cells.]{.underline} Small RNAs (\<200 nt) were extracted from transduced cells and probed with specific primers to detect the expression of siRNAs as described in materials and methods. A) Northern blot to detect the presence of miRNA-16 (\~22 bp) as an internal control in GFP-alone vector transduced (lane 2) and XHR transduced Magi-X4 (lane 3) and Ghost-R5 (lane 4) cells. B) siRNA (\~21 bp) detection in GFP-alone vector transduced (lane 2) and XHR transduced Magi-X4 (lane 3) and Ghost-R5 (lane 4) cells. Decade markers (lanes A1 and B1).
:::
:::
Bispecific siRNA vector does not induce interferon
--------------------------------------------------
Double stranded RNA molecules longer than \~30 bp are known to induce the interferon pathway in response to viral infections. As siRNAs are generally comprised of 19--24 bp in length, they are not expected to activate such a response that mediates a non-specific down regulation of cellular or viral mRNAs. However, recent data had shown that in some circumstances, certain siRNAs might induce variable levels of interferon activation \[[@B48]-[@B50]\]. To rule out such a possibility with the present siRNAs, we looked for upregulation of phosphorylated-PKR by western blot analysis. PKR is a protein kinase that becomes activated through phosphorylation in the presence of dsRNA and is involved during the interferon response. Our results have shown that the levels of phosphorylated PKR remain unchanged in XHR transduced cells similar to mock and GFP vector transduced cells. In contrast, elevated levels of phosphorylated PKR could be seen in poly I:C transfected cells used as positive controls (Fig [6](#F6){ref-type="fig"}). These data exclude the possibility of non-specific interferon activation by the combinatorial lentiviral construct.
::: {#F6 .fig}
Figure 6
::: {.caption}
######
[Lack of interferon induction in siRNA transduced cells.]{.underline} To detect interferon induction in siRNA vector transduced cells, western blot analysis was performed to detect elevated levels of phophorylated PKR. Poly I:C was used to induce interferon as a positive control. Transduced cell extracts were run on 10% SDS-PAGE gels, transferred, and probed with an anti-phospho-PKR antibody. Positive control poly I:C transfected (lanes 1 and 2), non-transduced (lane 3), GFP-alone vector (lane 4), and XHR transduced (lane 5) Magi-X4 cells and XHR transduced Ghost-R5 cells (lane 6). An anti-actin antibody was used as an internal control.
:::
:::
Resistance of siRNA transduced cells to HIV-1 infection
-------------------------------------------------------
To determine if down regulation of the essential coreceptors, CXCR4 and CCR5, translated to virus resistance, transduced Magi-CXCR4 and Ghost R5 cells were challenged with X4 (NL4-3) and R5 (BaL1)-tropic strains of HIV-1 respectively. Viral p24 antigen levels at different days post-challenge were determined by ELISA to quantify levels of HIV-1 resistance. Over a 10-fold reduction in viral antigen levels was seen with both XHR transduced Magi-CXCR4 and Ghost-R5 cells as compared to non-transduced and GFP-alone vector transduced cells (Fig [7](#F7){ref-type="fig"}). There was a slight increase in viral production in XHR transduced cells on days 5 to 7. This could be due to nontransduced and/or low siRNA expressing cells producing the virus. We next wanted to determine if the XHR vector expressing CXCR4 and CCR5 siRNAs is effective in physiologically relevant cells for gene therapy. Accordingly, PBMCs transduced with vectors were challenged in the same manner as above. A 3-fold level of inhibition was seen on days 3, 5, and 7 (Fig [8](#F8){ref-type="fig"}). These results established that the XHR vector is also effective in primary cells in inhibiting HIV-1. Although clearly significant, the levels of virus inhibition were not as dramatic as seen with Magi and Ghost cell lines. The observed levels of viral inhibition in primary PBMC are similar to those observed in a recent report \[[@B31]\]. Lower levels of protection in PBMCs were likely due to the lower levels of transduction. Future studies that are aimed at increasing transduction efficiencies into primary lymphocytes and macrophages are likely to overcome this hurdle.
::: {#F7 .fig}
Figure 7
::: {.caption}
######
[HIV-1 challenge of XHR transduced Magi-X4 and Ghost-R5 cells.]{.underline} Vector transduced cells were challenged with either X4 tropic or R5 tropic viruses at an m.o.i of 0.01. Culture supernatants were collected at different days post challenge and p24 antigen was assayed by ELISA. A) Transduced Magi-X4 cells challenged with X4 tropic HIV-1 NL4-3. B) Transduced Ghost-R5 cells challenged with R5 tropic HIV-1 BaL-1. Data presented is from triplicate experiments.
:::
:::
::: {#F8 .fig}
Figure 8
::: {.caption}
######
[HIV-1 challenge of XHR transduced PBMCs.]{.underline} Vector transduced PBMCs were challenged with either X4 tropic or R5 tropic viruses. Culture supernatants were collected at different days post challenge and p24 antigen was assayed by ELISA. Transduced PBMCs challenged with either HIV-1 NL4-3 (A) or BaL-1 (B). Data presented is from triplicate experiments.
:::
:::
In summary, our studies have shown for the first time that a single lentiviral vector could be used to stably deliver two different siRNAs targeted to two different cell surface co-receptor molecules and achieve protection against both X4 and R5 tropic HIV-1 viral strains. The short hairpin design permitted use of a single promoter to transcribe both the sense and anti-sense strands of each of the siRNAs. No promoter interference was observed between the U6 promoter driving the transcription of CXCR4 siRNA and the H1 promoter driving the CCR5 siRNA since comparable amounts of both the siRNAs could be seen in transduced cells. Furthermore, possible interferon induction by the combinatorial construct was also ruled out.
A major advantage in using a combinatorial lentiviral construct targeted to both the coreceptors is that infection with either of the viral strains could be prevented at the entry step thus eliminating the possibility of proviral integration and viral latency. Given the success with the current bispecific construct, other novel constructs could be designed and experimented with that incorporate siRNAs targeted to both the cellular as well as viral targets. Based on the design employed here, it is possible to introduce more than two siRNAs in a single construct in the future. However caution should be exercised while incorporating multiple siRNAs in a single construct because the possibility exists that over expression of foreign siRNAs in a cell may have undesirable effects such as saturating the endogenous RISC complex and consequent toxicity. Such a possibility needs to be tested in long range experiments *in vivo*. We previously have introduced a monospecific siRNA targeted to HIV-1 *rev*into CD34 hematopoietic progenitor cells via lentiviral vectors and derived transgenic macrophages *in vitro*and T cells *in vivo*\[[@B29]\]. The transgenic cells were found to be apparently normal while markedly resistant to HIV-1 infection.
No deleterious effects are expected by the stable knock down of the CCR5 coreceptor *in vivo*since individuals harboring a 32 bp deletion in the corresponding gene are physiologically normal \[[@B34],[@B35]\]. Although CXCR4 down regulation in circulating mature T cells in the periphery may not have any insurmountable ill effects, this may have possible drawbacks in a stem cell setting due to its role in cell homing into bone marrow \[[@B51],[@B52]\]. Additionally, recent gene expression profiling studies indicated some off-target effects by siRNAs \[[@B53]\]. Therefore, the present combinatorial construct targeted to both CXCR4 and CCR5 coreceptor molecules need to be thoroughly tested in an *in vivo*system such as the SCID-hu mouse model to evaluate its efficacy and possible toxicity in differentiated cells before it can be used for gene therapy in human subjects. Such experiments are currently underway.
Conclusions
===========
For HIV/AIDS gene therapy strategies to succeed, novel molecules need to be harnessed. In this regard, siRNAs offer great potential. Exploitation of these promising candidates to down regulate essential cellular coreceptors via the use of lentiviral vectors facilitates long term derivation of resistant T cells and macrophages which are the main targets for the virus. Our results showed for the first time that expression of both CXCR4 and CCR5 siRNAs in combination is possible by the use of lentiviral vectors. Coreceptor specific siRNAs stably transduced with the bispecific lentiviral vector showed marked resistance against both T cell tropic and monocyte tropic HIV-1 infection in cell lines and primary PBMCs. The newly developed bispecific vector shows promise for potential *in vivo*application.
Materials and Methods
=====================
Plasmid and lentiviral vector construction
------------------------------------------
Previously characterized siRNAs against CXCR4 and CCR5 were used in generating the bispecific lentiviral vector \[[@B23],[@B24],[@B30]\]. A third generation lentiviral vector backbone was employed to derive the bispecific constructs. The two *cis*-acting elements, namely, the central DNA flap consisting of cPPT and CTS (to facilitate the nuclear import of the viral preintegration complex) and the WPRE (to promote nuclear export of transcripts and/or increase the efficiency of polyadenylation of transcripts), are engineered to enhance the performance of the vector \[[@B38],[@B39]\]. An siRNA expression cassette targeting CXCR4 under the control of the Pol-III U6 promoter was PCR amplified from the plasmid pTZ-U6+1 as described by Castanotto *et al*\[[@B40]\]. This cassette was cloned into pHIV-7-GFP transfer vector in the *BamH*I site immediately upstream of the CMV-EGFP gene. This cassette contained a *Mlu*I restriction site downstream from the CXCR4 siRNA sequence for subsequent cloning of the H1 promoter driven CCR5 siRNA cassette. The H1-CCR5 siRNA expression cassette was also generated as described above using the plasmid pSUPER (Oligoengine, Seattle, WA). Sequencing and confirmation of candidate clones was performed by Laragen Inc. (Los Angeles, CA). The transfer vector containing the inserts U6-X4 siRNA and H1-CCR5 siRNA is termed pHIV-XHR-GFP.
Cell culture and vector production
----------------------------------
293T cells and PBMCs were maintained in DMEM media supplemented with 10% FBS. Magi-CXCR4 cells obtained from the AIDS Reference and Reagent Program were maintained in media as previously described \[[@B41],[@B42]\]. Ghost-R5 cells obtained from the AIDS Reference and Reagent Program were maintained in media as previously described \[[@B43]\]. To generate lentiviral vectors, fifteen micrograms of transfer vector with either GFP-alone or XHR were transfected along with 15 ug pCHGP-2, 5 ug pCMV-Rev, and 5 ug pCMV-VSVG into 293T cells at 60% confluency in 100 mm culture dishes using a calcium phosphate transfection kit (Sigma-Aldrich, St. Louis, MO). Six hours after transfection, fresh medium was exchanged. Cell culture supernatants containing the vector were collected at 24, 36, 48, and 60 hours post transfection and pooled. Vector supernatants were concentrated by ultracentrifugation and later titrated on 293T cells using FACS analysis for GFP expression.
Lentiviral vector transduction and FACS analysis
------------------------------------------------
Magi-CXCR4 and Ghost-CCR5 cells were seeded in 6-well plates 24 hours prior to transduction, 5 × 10^5^cells per well. Cells were transduced with lentiviral vectors at an m.o.i. of 10 in the presence of 4 ug/ml polybrene for 2 hours. For transduction of PBMCs, cells were first isolated from whole blood by Histopaque^®^-1077 (Sigma-Aldrich), and then cultured in CD3 and CD28 antibody coated plates. Three days after stimulation, PBMCs were transduced at an m.o.i of 20 in the presence of 4 ug/ml polybrene. PBMC transduction was repeated the following day. Seventy-two hours post transduction with siRNA containing lentiviral vectors, FACS analysis was performed to determine the levels of cell surface expression of CXCR4 and CCR5. Non-transduced and transduced cells were stained with appropriate antibodies conjugated with PE-Cy 5 (Pharmingen, San Diego, CA) namely, anti-CXCR4 for Magi-CXCR4 cells and anti-CCR5 for Ghost-CCR5 cells. Transduction efficiency was determined by assaying for EGFP expression. FACS analysis was performed on the Beckman Coulter Epics XL using ADC software for analysis.
Northern analysis for shRNA expression
--------------------------------------
Total RNA was extracted from non-transduced and transduced Magi-CXCR4 and Ghost-CCR5 cells using the RNA-STAT-60 reagent (Tel-Test, Friendswood, TX). Small RNAs, \<200 nt, were separated and concentrated using the *mir*Vana™ miRNA Isolation Kit (Ambion, Austin, TX). Twenty micrograms of small RNAs were hybridized overnight at 37°C using the *mir*Vana™ miRNA Detection Kit (Ambion) with γ-^32^P labeled probes made using the *mir*Vana™ Probe & Marker Kit (Ambion). Probes were complementary to the antisense strands of CXCR4 and CCR5 siRNAs. Hybridization reactions were processed according to the manufacturer\'s protocol and run on 15% polyacrylamide TBE-Urea gels. Gels were then exposed to X-ray film. A probe complementary to miRNA-16 supplied with the miRNA detection kit was used as an internal control.
Western Blot analysis of phosphorylated PKR
-------------------------------------------
Cell lysates of non-transduced and transduced cells were run on 10%-polyacrylamide-SDS TBE gels. Proteins were immunoblotted onto Immobilon™-P membranes (Millipore, Bedford, MA) and incubated with antibody specific for phosphorylated-PKR (Sigma-Aldrich), while anti-actin antibody (Sigma-Aldrich) was used to detect cellular actin as an internal control. A secondary antibody, goat anti-rabbit IgG conjugated with alkaline phophatase (Promega, Madison, WI), was then added. An alkaline phophatase substrate reagent, Western Blue (Promega), was used to visualize the bands.
RT-PCR
------
Total RNA was extracted from non-transduced and transduced cells. Primers specific for CXCR4 (forward: 5\'-ggaggggatcagtatatacacttc and reverse: 5\'-cgccaacatagaccaccttttc) and CCR5 (forward: 5\'-caaaaagaaggtcttcattacacc and reverse: 5\'-cttgctcgctcgggagcctc) (IDT, Coralsville, IA) were used to determine transcript levels while GAPDH (forward: 5\'-ctgagaacgggaagcttgtcatcaa and reverse: 5\'-gcctgcttcaccaccttcttgatg) primers were used as an internal control. One-step RT-PCR reactions were performed using the Superscript™ III One-Step RT-PCR kit (Invitrogen, Carlsbad, CA). Reactions were run on 1% agarose gels and appropriate bands were visualized with UV light.
HIV-1 Challenge
---------------
To determine if down-regulation of CXCR4 and CCR5 transcript levels and cell surface expression inhibited HIV-1 infection, non-transduced and transduced cells were challenged with NL4-3 (X4-tropic) and BaL-1 (R5-tropic) strains of HIV-1, at an m.o.i of 0.01, as previously described \[[@B24]\]. Viral supernatants were collected daily from infected Magi-CXCR4 and Ghost-CCR5 cells for p24 assay. ELISA was used to determine p24 values employing a Coulter-p24 kit (Beckman Coulter, Fullerton, CA). For PBMC challenge experiments, non-transduced and transduced cells were infected with NL4-3 and Bal-1 strains and cell culture supernatants were collected on days 1, 3, 5, and 7 post-infection to measure p24 levels.
Competing interests
===================
The author(s) declare that they have no competing interests.
Author\'s contributions
=======================
JA carried out all of the experiments. RA was responsible for the overall experimental design and implementation of the project.
Acknowledgements
================
Work reported here was supported by NIH grants AI50492 and AI057066 to R.A. This work has also been facilitated by the infrastructure and resources provided by the Colorado Center for AIDS Research Grant P30 AI054907. We thank Karen Helms for help with FACS analysis and William Wheat for critically reading the manuscript. We thank NIH AIDS Research and Reference Reagents Program for providing many reagents and cell lines used in this work.
|
PubMed Central
|
2024-06-05T03:55:55.063746
|
2005-1-13
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074340/",
"journal": "AIDS Res Ther. 2005 Jan 13; 2:1",
"authors": [
{
"first": "Joseph",
"last": "Anderson"
},
{
"first": "Ramesh",
"last": "Akkina"
}
]
}
|
PMC1074341
|
\" The open society, the unrestricted access to knowledge, the unplanned and uninhibited association of men for its furtherance -- these are what may make a vast, complex, ever growing, ever changing, ever more specialized and expert technological world, nevertheless a world of human community.\"
Julius Robert Oppenheimer, *Science and the Common Understanding*, Simon & Schuster, 1954 \[[@B1]\]
Open Access, online publication of scientific journals is a growing trend. In essence, Open Access is free and unrestricted availability of literature. This allows authors control over the integrity of their work and the right to be properly acknowledged and cited \[[@B2]\]. The Directory of Open Access Journals (DOAJ) <http://www.doaj.org/> lists 1358 journals, with in excess of 61129 articles; these numbers are increasing on a daily basis. BioMed Central and the Public Library of Science (PLoS) have literally revolutionized the business of science publication and popularization. The Open Access, online publications were initially spearheaded by BioMed Central. It is interesting to reflect that the concept of Open Access scientific publications was met with disbelief and opposition by both the scientific community and the publishing houses \[[@B3]\], however the success of BioMed Central has shown that increasing numbers of scientific publications will occur in the Open Access area. Indeed, most scientists would enjoy disseminating their science and ideas without being encumbered by the copyright laws and restrictions imposed by traditional publishing houses. Scientists are finding that Open Access is the fastest way of communicating their research with the rest of the world without any hindrance.
AIDS is one of the biggest threats to humankind. At the end of 2003 there were 40 million HIV-infected patients worldwide, of which 3 million were estimated to have died in 2003 \[[@B4]\]. The number of deaths is expected to rise year after year if the spread of disease is unchecked, with an estimated 5 million new cases being added every year. In countries where the HIV infection rates are as high as 30%, it is not a matter of mere human survival, but the survival of the countries. Humankind needs to make every possible effort to rid of this scourge. It is being approached from two traditional means: prevention and treatment. We need to greatly enhance our efforts in both these approaches. It has been predicted by several mathematical models that even modest prevention efforts in the spread of HIV-1 can tame the scourge considerably.
Dissemination of the latest knowledge in basic science and its application to AIDS therapy and prevention is of paramount importance. To date, this has been accomplished with several excellent print journals dedicated to HIV-1 and the disease caused by it, AIDS. Many of them are now accessible online. However, we are launching the first Open Access journal in the AIDS arena -- ***AIDS Research and Therapy***. The purpose of this journal is not to compete with or replace any of them, but to provide a unique platform where both basic research and clinical science can be studied side by side to contemplate, design, and develop applications of basic science in preclinical or clinical research. The overarching goal of the journal is to communicate emerging knowledge at a faster pace with an aim to accelerate bed-side research by taking into consideration the latest bench-side accomplishments. An additional benefit of an Open Access, online journal is enormous cost savings for libraries and institutions in terms of the high subscription rates and archival of the old journals.
*AIDS Research and Therapy*will encompass all aspects of basic and clinical science that impact on abating the spread of AIDS; it is a multidisciplinary journal that aims to keep scientists and clinicians abreast of the latest research on HIV-1. *AIDS Research and Therapy*will be particularly interested in publishing articles on novel and developing treatment strategies for AIDS, as well as articles on the outcome of treatment strategies. The ultimate goal of the journal is to lead research from bench to bedside in the combat of AIDS.
Each manuscript submitted to the journal will be sent to the appropriate [Section Editor]{.underline}, who will assign at least 3 reviewers (members of the Editorial Board or outside experts). Reviewers are expected to return reports within two-three weeks, and the Section Editor will then make a decision based on these reports. This is to insure timely publication of findings that are important in the prevention and cure of AIDS. Indeed, we anticipate a turn around time (from submission to acceptance) of about two months. Authors, together with the Editor/reviewer handling the manuscript will be required to declare any competing interests. *AIDS Research and Therapy*will consider following types of articles: Research, Hypotheses, Methodology, Short reports, Study protocols, short as well as comprehensive Reviews. Efforts have been made to select Editors from as diverse areas of HIV and AIDS research as possible. However, Editors from areas currently not represented in the Editorial Board will be included. Readers have the option of recommending scientists who may be suitable Editors and/or reviewers.
*AIDS Research and Therapy*is published by [BioMed Central]{.underline}, an independent publishing house, committed to ensuring peer-reviewed biomedical research is Open Access -- immediately and permanently available online without charge or any other barriers to access. In an Open Access journal, the communication has to be bi-directional for the journal\'s success and improving the quality of the journal. Consequently, the Editorial Board of the journal will constantly seek new ideas to improve the journal to serve as an excellent resource in AIDS research and therapy to the scientific community.
Rapid dissemination of science is crucial for the progress and survival of human kind. Let the wheel of communication roll at the fastest pace and to the widest possible world.
|
PubMed Central
|
2024-06-05T03:55:55.065913
|
2004-12-14
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074341/",
"journal": "AIDS Res Ther. 2004 Dec 14; 1:1",
"authors": [
{
"first": "Kailash C",
"last": "Gupta"
}
]
}
|
PMC1074342
|
Background
==========
Human T lymphocytes and macrophages are the major host cells for HIV-1 replication. The initial infection is established by macrophage tropic viruses (R5) that use the chemokine receptor CCR5 and CD4 to gain entry into a susceptible host cell. During the later stages of the disease, T-cell tropic viruses (X4) that use CXCR4 as a coreceptor predominate \[[@B1],[@B2]\]. Since HIV-1 coreceptors play a key role during the early viral-cell interactions, they are attractive targets for many antiviral approaches. A 32-base pair deletion in the CCR5 gene found in a segment of the normal European and North-American population rendered their macrophages resistant to infection by R5-tropic HIV-1 \[[@B3]\]. Since these individuals lacking a functional CCR5 are apparently normal, this gene has been targeted by many investigators to confer HIV-1 resistance. Using MuLV vectors for gene delivery, ribozymes or DNA-enzymes targeted against CCR5 were previously shown to inhibit HIV-1 entry both *in vitro*and *in vivo*in a SCID-hu mouse model \[[@B4]-[@B6]\]. Efficacy of siRNAs in down regulating the CCR5 coreceptor and thereby preventing HIV-1 entry was also described recently \[[@B7],[@B8]\].
The regulatory proteins Tat and Rev encoded by the viral genome are indispensable for HIV-1 gene expression and replication. The Tat protein interacts with the bulged RNA region of the transactivation response element (Tar), present at the 5\'-end of all HIV-1 transcripts \[[@B1]\]. In the absence of Tat, only short ineffective transcripts are generated. Tat is also known to interact with cellular factors like cyclin T1 and cyclin dependent kinase (Cdk9). Since Tat plays a critical role in virus replication, it is an ideal target. Effective inhibition of HIV-1 replication was shown earlier by the use of Tar specific RNA decoys and ribozymes \[[@B9]-[@B11]\]. Moreover, siRNAs directed against Tat were also found to be highly potent in inhibiting HIV-1 replication in cultured cell lines and in PBMCs \[[@B12],[@B13]\]. However, development of viral resistance and generation of escape mutants are possible obstacles for long range efficacy of these constructs as exemplified by the recent findings of Boden et al \[[@B14]\]. These obstacles can be overcome by the use of combinatorial constructs against multiple targets in the viral genome as well as cellular targets that assist in viral infection and replication.
CD34+ hematopoietic progenitor stem cells (HPCs) are ideal targets for transducing anti-HIV genes as they give rise to both T cells and macrophages which are the main viral targets. Most of the previous work with anti-HIV ribozymes and RNA decoys employed conventional MuLV derived retroviral vectors to transduce these cells \[[@B5],[@B6],[@B11]\]. However, the efficiency of gene transduction by these vectors is relatively low as they are unable to transduce non-dividing cells. In addition, the transgenes carried by these vectors are prone to gene silencing during the differentiation of end stage cells such as T cells and macrophages \[[@B15]\]. On the contrary, lentiviral vectors appear not to have these limitations \[[@B16],[@B17]\]. Based on these advantages, we used the new generation lentiviral vectors to achieve high level gene transfer and sustained gene expression. A ribozyme against CCR5 and a Tar aptamer decoy were previously shown to inhibit HIV-1 in transduced cells \[[@B6],[@B18],[@B19]\]. In previous *in vivo*studies, MuLV based conventional retroviral vectors were used. In addition, it is not known if Tar decoy is effective in *in vivo*differentiated thymocytes. In these studies, our goal is to determine the utility of these constructs when introduced into human CD34+ progenitor cells via an HIV-1 based lentiviral vector to derive HIV resistant differentiated target cells both *in vitro*and *in vivo*. Using an *in vitro*cell differentiation system and *in vivo*SCID-hu mouse model, we show that expression of Tar decoy alone or in combination with an anti-CCR5 ribozyme has no adverse effect on lineage specific differentiation of CD34+ cells into macrophages and T lymphocytes. We also show that the transgenic cells display resistance to HIV-1 challenge.
Results
=======
High efficiency transduction of CD34+ cells with lentiviral constructs
----------------------------------------------------------------------
Early studies using MuLV based retrovirus vectors have shown the efficacy of aptamers and ribozymes against Tat, Rev, or envelope in interfering with HIV-1 infection \[[@B5],[@B6],[@B10],[@B20]\]. Down regulation of CCR5 by the ribozyme used here and its corresponding inhibitory effect on HIV-1 infection was described previously \[[@B5],[@B6],[@B19]\]. To further expand the utility of such inhibitory RNAs, we used a third generation HIV-1 based self-inactivating vector (Fig. [1](#F1){ref-type="fig"}) \[[@B21]\] A highly enriched population of human CD34+ cells (\>90% pure) were used for vector transductions. A representative FACS profile of purified CD34+ cells is shown in Fig. [2A](#F2){ref-type="fig"}. Transduction efficiency, as determined by FACS for EGFP at 48 hrs post-transduction, showed high levels of gene transfer and exceeded 90% for both U16Tar decoy and Tar-CCR5 vector constructs (Fig. [2](#F2){ref-type="fig"}, panel B & C).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
HIV-1 based lentivirus transfer vectors: A, control vector pHIV-7-EGFP with an EGFP reporter gene driven by the CMV promoter. B, HIV-U16Tar (F)-GFP vector with U6 driven Tar decoy. C, HIV-Tar-CCR5 ribozyme vector with U6 driven Tar and VA1 driven CCR5 ribozyme \[18\]. To generate vector viruses, a four-plasmid transfection system was used as described in methods.
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
CD34+ cell purity and transduction efficiency: CD34+ cells derived from human fetal liver were purified by immunomagnetic beads and assayed by FACS. Percentage purity is indicated in A. Purified CD34+ cells were transduced with HIV-1 Tar (B) and Tar-CCR5 ribozyme (C) containing lentiviral vector. Percentages of EGFP positive cells at 48 hrs post-transduction are indicated in panel B and C. Isotype antibody control is shown in each panel (unshaded areas).
:::
:::
Tar and Tar-CCR5Rz vector transduced CD34+ cells differentiate normally into mature macrophages
-----------------------------------------------------------------------------------------------
It is not known if lentivirus transduced Tar decoy and Tar-CCR5Rz, will have any adverse effects on the lineage specific differentiation of CD34+ cells into different end stage cells. Our results showed that both control and vector transduced cells matured normally into erythroid and myeloid colonies and no significant differences were observed between their colony forming abilities (data not shown). To determine if Tar and Tar-CCR5Rz RNA expressing CD34+ cells can give rise to mature macrophages, myeloid colonies were pooled and allowed to differentiate into adherent cells in cultures supplemented with M-CSF and GM-CSF for a period of 7 days. Results showed that cells derived from control, vector alone, or vector expressing transgenes (Tar and Tar-CCR5Rz) showed similar pattern of CD14 expression (Fig. [3](#F3){ref-type="fig"}, panel A1 to A4). The transgenic macrophages were also analyzed for the levels of EGFP reporter expression. As expected, the nontransduced cells did not show any EGFP expression (Fig. [3](#F3){ref-type="fig"}, panel B1) but cells transduced with either EGFP vector alone (panel B2) or vector with transgenes (panel B3 and B4) were strongly positive (\>80%) for EGFP production. RT-PCR analysis confirmed the expression of the transgene Tar. Tar specific products of expected size (125 bp) were detected in both Tar (Fig. [4](#F4){ref-type="fig"}, panel A, lane 2) and Tar-CCR5Rz (lane 3) vector transduced macrophages. Control nontransduced macrophages, as expected, did not show any specific product (lane 1). No difference in the amounts of control *β*-actin RNA could be seen in the corresponding lanes (Fig. [4](#F4){ref-type="fig"}, panel B).
::: {#F3 .fig}
Figure 3
::: {.caption}
######
FACS analysis of transgenic macrophages for the CD14 surface marker and EGFP expression: Control and HIV-1 Tar & Tar-CCR5Rz vector transduced CD34+ cells were differentiated into macrophages *in vitro*in cytokine medium. Cells were stained for the macrophage surface marker CD14 using CD14-PE antibodies. Cells were analyzed by FACS for CD14 and EGFP. Panel A: CD14 staining of macrophages. 1, Nontransduced cells; 2, Control EGFP vector transduced; 3, HIV-1-Tar vector transduced; 4, HIV-1-Tar-CCR5Rz transduced. Percent CD14 positive cells are indicated together with isotype staining controls (unshaded areas). Panel B: EGFP expression by transduced macrophages. 1, Nontransduced cells; 2, Control EGFP vector transduced; 3, HIV-1-Tar vector transduced; 4, HIV-1-Tar-CCR5Rz vector transduced. Percent EGFP positive macrophages are indicated.
:::
:::
::: {#F4 .fig}
Figure 4
::: {.caption}
######
RT-PCR detection of Tar RNA in differentiated macrophages: Total cellular RNA was extracted from control and transgenic macrophages and subjected to RT-PCR using primers specific for HIV-1 Tar decoy. A. HIV-1 Tar specific amplification (125 bp). 1, control macrophages; 2, HIV-1-Tar vector transduced macrophages; 3, HIV-1-Tar-CCR5Rz vector transduced macrophages. B. *β*-actin RNA amplified in corresponding lanes as controls.
:::
:::
Tar and Tar-CCR5Rz transgenic macrophages resist HIV-1 challenge
----------------------------------------------------------------
To determine if Tar and Tar-CCR5Rz transduced *in vitro*differentiated macrophages resist HIV-1 challenge, they were infected with R5-tropic HIV-1 Bal strain. Culture supernatants collected on different days post challenge were assayed for p24 antigen by ELISA. Compared to unmanipulated control or EGFP control, both Tar and Tar-CCR5Rz expressing macrophages showed remarkable resistance against HIV-1 challenge (Fig. [5](#F5){ref-type="fig"}). Small amounts of p24 could be detected on day seven and none at nine days post infection in Tar and Tar-CCR5Rz transduced cells.
::: {#F5 .fig}
Figure 5
::: {.caption}
######
HIV-1 challenge of differentiated macrophages: HIV-1 Tar, Tar-CCR5Rz and control EGFP vector transduced and unmanipulated control CD34+ cells were allowed to differentiate into macrophages *in vitro*. Later, they were challenged with a macrophage tropic HIV-1 strain BaL. Viral supernatants were collected at different times post-infection and assayed for p24 antigen by ELISA. Values represent averages of duplicate cultures.
:::
:::
Tar and Tar-CCR5Rz transduced CD34+ cells can give rise to thymocytes in SCID-hu thy/liv grafts
-----------------------------------------------------------------------------------------------
Human thy/liv grafts in SCID-hu mice provide an ideal environment for CD34+ cells to mature into thymocytes. To determine if the lentivirally expressed transgenes Tar and Tar-CCR5Rz would have any adverse effect on this differentiation process, thymocytes obtained from SCID-hu grafts 60--70 days post reconstitution were analyzed for EGFP expression. All of the four mice (two each with Tar and Tar-CCR5Rz) that were injected with transduced CD34+ cells were positive for the presence EGFP expressing thymocytes. Of the two mice injected with Tar construct one showed 85% reconstitution levels with the other being 33%. For the two Tar-CCR5Rz construct injected mice, the reconstitution levels were 75% and 30% (Fig. [6](#F6){ref-type="fig"}, panels A and B). Reconstitution levels are known to vary considerably between mouse to mouse based on the variable sizes of the grafts injected and possibly due to the varying numbers of true stem cells present in the samples injected \[[@B22],[@B23]\]. Since vector transduced CD34+ cells gave rise to EGFP expressing thymocytes in SCID-hu grafts, these results suggested that expression of either Tar or Tar-CCR5Rz RNA did not have any detectable deleterious effects on the thymopoiesis steps *in vivo*. FACS analysis was carried out on biopsied thymocytes by staining for CD4 and CD8 antigens to evaluate the presence of different cell subsets. The majority of the thymocytes (75 -- 80%) stained positive for CD4 and CD8 (double positive) consistent with normal thymopoiesis (Fig. [7](#F7){ref-type="fig"}, panels B to D). Similar to the control (B), both CD4 and CD8 single positive mature thymocytes are also seen in Tar and Tar-CCR5Rz transduced thymocytes derived in SCID-hu grafts (C and D). These data indicated normal development of all three thymocyte subpopulations from Tar and Tar-CCR5Rz transduced CD34+ cells. When these cells were cultured *in vitro*for an additional 7 days, a rapid decline in the number of CD4/CD8 double positive cells was observed which coincided with a corresponding increase in single positive mature thymocytes (data not shown). To determine if the transgenic thymocytes derived in the SCID-hu mice retained their ability for nonspecific mitogenic stimulation in the presence of IL-2, they were cultured in the presence of PHA-P for 3 days. Approximately, a 3-fold increase in the number of thymocytes was observed for both control as well as transduced cells (data not shown). These cells expressed the chemokine receptor CXCR4, as expected (data not shown).
::: {#F6 .fig}
Figure 6
::: {.caption}
######
EGFP expression by in vivo derived thymocytes: HIV-1-Tar and Tar-CCR5Rz vector transduced CD34+ cells were injected into the SCID-hu mice thy/liv grafts and allowed to differentiate into thymocytes. At \~60 days post-engraftment, the cells were harvested and analyzed by FACS for EGFP expression. A. thymocytes from HIV-1 Tar transduced cells. B. thymocytes from Tar-CCR5Rz transduced cells. Percent positive cells are indicated. Representative samples from one mouse each are shown.
:::
:::
::: {#F7 .fig}
Figure 7
::: {.caption}
######
FACS profiles of thymocyte subsets derived in SCID-hu grafts: To determine the presence of different thymocyte subsets, *in vivo*differentiated cells from SCID-hu grafts were collected and stained for thymocyte markers CD4 and CD8 by using PE and FITC conjugated antibodies respectively. The stained cells were analyzed by two color FACS. A, Isotype control. B, thymocytes from a control animal; C, thymocytes from a HIV-1 Tar animal; D, thymocytes from a HIV-1-Tar-CCR5Rz animal.
:::
:::
*In vivo*derived transgenic thymocytes resist HIV-1 challenge
-------------------------------------------------------------
To determine if Tar and Tar-CCR5Rz RNA expressing thymocytes display resistance to HIV-1 replication, FACS sorted EGFP positive cells were challenged with a T-tropic HIV-1 NL4.3 virus *in vitro*(Fig. [8](#F8){ref-type="fig"}). Thymocytes isolated from both groups of mice (Tar and Tar-CCR5Rz transduced) showed remarkable resistance to HIV-1. In contrast, control unmanipulated thymocytes produced large amounts of virus (\~8 fold higher p24 antigen on day 6). These control cells continued to produce detectable virus until 25 days post infection (data not shown)
::: {#F8 .fig}
Figure 8
::: {.caption}
######
HIV-1 challenge of *in vivo*differentiated thymocytes: Vector transduced CD34+ cells were injected into SCID-hu thymic grafts and allowed to differentiate. Thymocytes were harvested from two different mice at 60 days post-engraftment. To enrich for the EGFP positive cells, they were sorted by FACS (\>90% purity). The cells were expanded by culturing *in vitro*and challenged with the T cell tropic HIV-1 NL4-3. On different days post-infection, samples were collected and assayed for p24 antigen. Tar 1 & 2 and Tar-CCR5Rz 1 & 2 represent data from two different mice.
:::
:::
Discussion
==========
The future success of stem cell based gene therapy strategies against HIV-1 infection depends on harnessing novel interfering genes for *in vivo*application in humans. This requires collective utilization of stem cell gene transduction with novel vectors followed by preclinical evaluation of gene therapeutic constructs in an *in vivo*setting. To achieve this goal, we used lentiviral vectors to transduce CD34+ cells with a Tar decoy alone or in combination with an anti-CCR5 ribozyme that down regulates an essential HIV-1 coreceptor. Tar decoy interferes with an essential HIV-1 regulatory gene thus inhibiting post-entry steps of viral replication, whereas the anti-CCR5 ribozyme helps prevent viral entry. In view of the previously demonstrated *in vitro*high efficacy of aptamers targeted to different HIV-1 proteins and the high likelihood of these being exploited for clinical use, it is essential that they be tested thoroughly *in vivo*. These experiments mark the first simultaneous evaluation of these constructs in lentivirally transduced CD34+ cells both *in vitro*and *in vivo*.
The use of lentiviral vectors permitted higher levels of gene transfer (\>90%) into CD34+ cells with both of the constructs as assayed by EGFP expression. Two rounds of transduction with a highly concentrated VSV-G pseudotyped vector helped achieve high levels of gene transfer. Cells continued to express EGFP throughout the experimental period when cultured *in vitro*in the presence of cytokines and growth factors, and as long as 70 days *in vivo*in SCID-hu mice. The self-inactivating lentiviral vector employed here incorporated two important *cis*elements, namely a flap region and a WPRE element, to achieve high levels of EGFP expression \[[@B21]\]. Additionally, to minimize promoter interference, EGFP reporter, Tar, and CCR5 ribozyme genes were placed under the control of three different promoters namely, CMV, U6 and VA1 respectively \[[@B18]\]. Based on the high levels of gene transfer and sustained expression obtained, lentiviral vectors are highly suitable for gene transfer of RNA decoys and ribozymes.
CD34+ cells can be differentiated into myeloid, erythroid, and macrophage cell progeny in the presence of appropriate growth factors *in vitro*, and into mature T lymphocytes *in vivo*in SCID-hu mice \[[@B6],[@B23]\]. *In vitro*CFU assays yielded similar levels of differentiated erythroid and myeloid colonies, and in long term culture with cytokines, more than 90% of cells matured into macrophages and expressed normal levels of CD14. Remarkably, EGFP production also remained very high (\>80%) in transgenic macrophages. Thus lentivirus mediated Tar and Tar-CCR5Rz transgene expression did not adversely interfere with the differentiation of CD34+ cells into different lineages including macrophages. In *in vivo*experiments with SCID-hu mice, cell biopsies analyzed 60 to 70 days post engraftment showed that both Tar and Tar-CCR5Rz transduced progenitor cells matured into T-lymphocytes. During the normal course of thymopoiesis, the T cell precursors initially give rise to CD4 and CD8 double positive immature cells followed by subsequent end stage maturation into single positive CD4 and CD8 cells \[[@B24]\]. It is possible that transgene expression may selectively alter maturation of different cell subsets. Our results showed the presence of all three thymocyte subsets in grafts reconstituted with transduced cells when compared to control cells. Additionally, when the transgenic thymocytes were sorted and cultured *in vitro*, the levels of immature thymocytes declined rapidly with a corresponding increase in single positive CD4 and CD8 cells demonstrating their capacity to mature. Collectively, this data established that transduced CD34+ progenitor cells can differentiate normally into mature macrophages and thymocytes thus indicating no apparent toxicity of these constructs on lineage specific differentiation.
Viral challenge experiments demonstrated that both Tar and Tar-CCR5Rz RNA expressing mature T-lymphocytes and macrophages are remarkably resistant to HIV-1 infection. No synergistic effect could be observed with the combinatorial construct most likely due to the predominant effect of the Tar decoy itself at the low m.o.i. used here. However, synergistic effect of the combinatorial construct was demonstrated in previous studies that used a higher challenge dose \[[@B18]\]. In early studies with similar constructs using MuLV based retrovirus vectors, \[[@B6],[@B20]\] viral inhibition was seen up to 2 weeks post challenge in differentiated thymocytes. However, there was a significant viral breakthrough by the third week. In contrast, with the lentiviral vector delivered constructs employed here, virus production in both challenged macrophages and T-lymphocytes remained significantly lower throughout the three week observation period. This improved level of protection is likely due to higher levels of gene transduction and expression, lower levels of transgene silencing during cell differentiation steps, or a combination of both. Although the levels of viral inhibition achieved in transgenic macrophages and T-lymphocytes are highly significant, small amounts of viral production is still detectable. This could be due to sub-optimal levels of transgene expression in a subpopulation of cells, or alternatively due to the presence of a small number of non-transduced cells in culture. Nevertheless, the above results demonstrated the efficacy of these transgenes in a combinatorial setting in a stem cell-based gene therapy context. The above results paved the way for exploiting this approach in human clinical trials.
Conclusions
===========
High efficiency transduction and sustained expression of HIV-1 interfering genes, anti-CCR5 ribozyme and HIV-1 Tar aptamer, could be achieved in CD34+ hematopoietic progenitor cells by using lentiviral vectors. The transduced progenitor cells differentiated normally into mature thymocytes *in vivo*in thy/liv grafts of SCID-hu mice and into normal macrophages *in vitro*. When challenged with HIV-1, transgenic cells showed marked resistance against HIV-1 infection. These results showed for the first time that expression of these transgenes in combination do not interfere with normal thymopoiesis and thus have set the stage for their application in stem cell based gene therapy for HIV/AIDS.
Methods
=======
Tar decoy and Tar-CCR5Rz containing lentiviral vectors
------------------------------------------------------
The design, structure, and *in vitro*efficacy in cultured cells of anti-CCR5 ribozyme, Tar decoy, and Tar-CCR5Rz constructs were described previously \[[@B5],[@B10],[@B18],[@B19]\]. These inhibitory RNAs introduced into a third generation self-inactivating lentiviral vector were used in the present study \[[@B21]\]. The transfer vector pHIV-7-GFP containing a CMV driven EGFP reporter gene is depicted in Fig. [1](#F1){ref-type="fig"}, panel 1A. Two important unique features are the *cis*-acting elements, the HIV-1 central flap sequence and the woodchuck post-transcriptional regulatory element (WPRE) for optimal EGFP expression. In the transfer vector pHIV-U16-Tar-GFP, the Tar decoy under the control of the U6 promoter was positioned upstream of the EGFP reporter (Fig [1B](#F1){ref-type="fig"}). In the combinatorial construct pHIV-U16Tar-CCR5Rz-GFP, the Tar decoy is driven by U6 whereas the CCR5 ribozyme is under the control of the VA1 promoter (Fig. [1C](#F1){ref-type="fig"}).
Production of high titered retroviral vectors
---------------------------------------------
To generate vector stocks, 293T cells were transfected with 15 μg of pCHGP-2 (encodes HIV-1 gag/pol), 15 μg of transfer vector (pHIV-7 GFP or pHIV-U16 Tar-GFP or pHIV-U16Tar-CCR5Rz-GFP), 5 μg of pCMV-rev and pCMV-G each as described previously \[[@B23]\]. Viral supernatants were collected at 24, 48 and, 72 hrs post transfection, pooled, and concentrated by ultracentrifugation \[[@B25]\]. Concentrated virus was resuspended in a small volume (500 μl) of DMEM containing 10% fetal bovine serum. The titer of the vector preparation was determined in 293T cells as described previously and ranged from 1 to 3 × 10^8^TU/ml. Multiple aliquots were made and stored at -70°C.
Isolation of CD34+ hematopoietic progenitor cells and high efficiency vector transduction
-----------------------------------------------------------------------------------------
Human fetal liver CD34+ hematopoietic progenitor cells (HPC) were purified by positive selection on a magnetic column using the Direct CD34 Progenitor Cell Isolation Kit from Miltenyi Biotech, Gladbach, Germany, as described in detail earlier \[[@B23]\]. Purified cells were suspended in Iscove\'s medium supplemented with IL3, IL6, and human stem cell factor (SCF), each at a concentration of 100 ng/ml (R & D Systems, Minneapolis, MN) and cultured for 15 hours at 37°C. Vector transductions were carried out in a 12-well tissue culture plate using 2 × 10^6^cells at an m.o.i. of 10 to 20 in a final volume of 100 μl of medium containing 4 μg/ml polybrene. Following transduction, cell aliquots were used for carrying out *in vitro*colony forming unit (CFU) assays, generation of macrophages, and for reconstitution of human thy/liv grafts in SCID-hu mice to generate T cells.
CFU assays and generation of macrophages
----------------------------------------
Control, or vector transduced CD34+ cells were allowed to differentiate into multiple lineages of erythroid and myeloid lineages in a semi-solid medium (MethocultTM GF H4434, Stem cell Technologies, Vancouver, BC, Canada). This medium contains the following components: 1% Methylcellulose in Iscove\'s MDM, 30% fetal bovine serum, 1% bovine serum albumnin, 0.1 mM 2-mercaptoethanol, 2 mM glutamine, 50 ng/ml rh stem cell factor, 10 ng/ml rh GM-CSF, 10 ng/ml rh IL-3, and 3 units/ml rh erythropoietin. A colony forming unit (CFU) was defined as having at least 50 cells after 14 days in the above selective medium. Individual myelomonocytic colonies were pooled and cultured in DMEM supplemented with 10% fetal bovine serum, 50 ng/ml M-CSF, and 20 ng/ml GM-CSF, for a period of 14 days for differentiation into macrophages. Cells were stained for CD14 antigen and analyzed by FACS to determine macrophage yield.
Reconstitution of SCID-hu grafts with transduced CD34+ cells and derivation of T cells
--------------------------------------------------------------------------------------
Human fetal thymus and liver tissues were implanted under the kidney capsule of SCID mice to generate SCID-hu mice as described earlier \[[@B26]\]. Control and lentivirus vector transduced CD34+ progenitor cells (1 × 10^6^) were injected directly into thy/liv grafts for reconstitution. Eight to ten weeks post reconstitution, thus allowing for T cell differentiation, the animals were sacrificed and thymocytes were isolated from the grafts. The differentiated thymocytes were cultured *in vitro*and checked for their ability to respond to mitogen, PHA-P and interleukin-2. Briefly, thymocytes were washed once with medium containing serum and resuspended into Iscove\'s medium supplemented with 10% fetal bovine serum. Approximately 2 × 10^6^cells were plated in a 12 well tissue culture plate and stimulated with PHA-P (4 μg/ml) and IL-2 (10 U/ml) for three days. Cells were counted after 3 days to determine expansion.
PCR Detection of Tar RNA
------------------------
Total RNA was isolated from approximately 1 × 10^6^control and transgenic macrophages using Qiagen RNA/DNA mini kit (Qiagen, Germany) and subjected to RT-PCR as described before \[[@B20]\]. In each case, a 125 bp DNA fragment is expected. The following primers were used: 1, Forward Tar: 5\'-GCAATGATGTCGTAATTTGC and 2, Reverse Tar: 5\'-CTTGCTCAGTAAGAATTTTCGTC.
HIV-1 infection of thymocytes
-----------------------------
Thymocytes derived from thy/liv grafts of SCID-hu mice were sorted by FACS to enrich for EGFP expressing cells (\>90% purity). They were later expanded by stimulation by PHA-P in medium containing serum and IL-2 as described earlier \[[@B6],[@B23]\]. Approximately 10^6^cells were infected with HIV-1 NL4-3 at an m.o.i. of 0.001 in a final volume of 100 μl for 3 hrs at 37°C. Infected cells were washed twice with DMEM with 10% fetal bovine serum and cultured in a 12 well plate for 3 weeks. Supernatants (0.5 ml) were collected on alternative days with media replenished in each well. Amounts of virus produced in cell culture supernatants was measured by HIV-1 p24 ELISA.
HIV-1 challenge of CD34+ cell derived macrophages
-------------------------------------------------
Infection with a macrophage-tropic Bal-1 strain of HIV-1 was carried out in a 6 well plate. Approximately 2 × 10^6^adherent macrophages differentiated *in vitro*were infected with Bal-1 virus at an m.o.i. of 0.001 in the presence of 4 μg/ml of polybrene for 6 hours. Thereafter, 3 ml of DMEM supplemented with 10% serum was added. Supernatants (0.5 ml) were collected every other day from each well for 3 weeks and stored at -70°C. The levels of virus released were determined by p24 antigen ELISA.
Competing interests
===================
The author(s) declare that they have no competing interests.
Author\'s contributions
=======================
AB carried out most of the experiments. M Li and JR were responsible for vector design and preparation. LR assisted in SCID-hu mice generation, CD34 cell reconstitutions into mice, PCR and FACS analysis. RA was responsible for the overall experimental design and implementation of the project.
Acknowledgements
================
Work reported here was supported by NIH grants AI50492 and AI057066 to R.A and AI 42552 and AI2932 to JR. This work has also been facilitated by the infrastructure and resources provided by the Colorado Center for AIDS Research Grant P30 AI054907. We thank Jeanette Hayes-Klug for assistance with SCID-hu mouse surgeries, Karen Helms for help with FACS and Joe Anderson for critically reading the manuscript. We thank NIH AIDS Research and Reference Reagents Program for providing many reagents and cell lines used in this work.
|
PubMed Central
|
2024-06-05T03:55:55.066819
|
2004-12-17
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074342/",
"journal": "AIDS Res Ther. 2004 Dec 17; 1:2",
"authors": [
{
"first": "Akhil",
"last": "Banerjea"
},
{
"first": "Ming-Jie",
"last": "Li"
},
{
"first": "Leila",
"last": "Remling"
},
{
"first": "John",
"last": "Rossi"
},
{
"first": "Ramesh",
"last": "Akkina"
}
]
}
|
PMC1074343
|
Introduction
============
Based on visual observation, the ancients characterised inflammation by five cardinal signs, namely redness (*rubor*), swelling (*tumour*), heat (*calor*; only applicable to the body\' extremities), pain (*dolor*) and loss of function (*functio laesa*). The first four of these signs were named by Celsus in ancient Rome (30--38 B.C.) and the last by Galen (A.D 130--200) \[[@B1]\]. More recently, inflammation was described as \"the succession of changes which occurs in a living tissue when it is injured provided that the injury is not of such a degree as to at once destroy its structure and vitality\" \[[@B2]\], or \"the reaction to injury of the living microcirculation and related tissues \[[@B3]\]. Although, in ancient times inflammation was recognised as being part of the healing process, up to the end of the 19^th^century, inflammation was viewed as being an undesirable response that was harmful to the host. However, beginning with the work of Metchnikoff and others in the 19^th^century, the contribution of inflammation to the body\'s defensive and healing process was recognised \[[@B1]\]. Furthermore, inflammation is considered the cornerstone of pathology in that the changes observed are indicative of injury and disease.
The classical description of inflammation accounts for the visual changes seen. Thus, the sensation of heat is caused by the increased movement of blood through dilated vessels into the environmentally cooled extremities, also resulting on the increased redness (due to the additional number of erythrocytes passing through the area). The swelling (oedema) is the result of increased passage of fluid from dilated and permeable blood vessels into the surrounding tissues, infiltration of cells into the damaged area, and in prolonged inflammatory responses deposition of connective tissue. Pain is due to the direct effects of mediators, either from initial damage or that resulting from the inflammatory response itself, and the stretching of sensory nerves due to oedema. The loss of function refers to either simple loss of mobility in a joint, due to the oedema and pain, or to the replacement of functional cells with scar tissue.
Today it is recognised that inflammation is far more complex than might first appear from the simple description given above and is a major response of the immune system to tissue damage and infection, although not all infection gives rise to inflammation. Inflammation is also diverse, ranging from the acute inflammation associated with *S. aureus*infection of the skin (the humble boil), through to chronic inflammatory processes resulting in remodeling of the artery wall in atherosclerosis; the bronchial wall in asthma and chronic bronchitis, and the debilitating destruction of the joints associated with rheumatoid arthritis.
These processes involve the major cells of the immune system, including neutrophils, basophils, mast cells, T-cells, B-cells, etc. However, examination of a range of inflammatory lesions demonstrates the presence of specific leukocytes in any given lesion. That is, the inflammatory process is regulated in such a way as to ensure the appropriate leukocytes are recruited. These events are controlled by a host of extracellular mediators and regulators, including cytokines, growth factors, eicosanoids (prostaglandins, leukotrines, etc), complement and peptides. In fact, it is the discovery of many of these mediators over the past 20 years that has increased our understanding of the regulation of the inflammatory process whilst, at the same time, revealing its complexity. These extracellular events are matched by equally complex intracellular signalling control mechanisms, with the ability of cells to assemble and disassemble an almost bewildering array of signalling pathways as they move from inactive to dedicated roles within the inflammatory response and site.
Which cells and mediators come into play depends on wide range of factors. These include: what stage the process of inflation is at; the initiating event, i.e. type of pathogen, auto-immune, chemical or physical injury, etc.; the tissue or organ involved; whether the inflammation is of an acute, resolving form or chronic, non resolving or long-lasting type; whether formation of granuloma is involved, or whether scarring results.
The role of inflammation as a healing, restorative process, as well as its aggressive role, is also more widely recognised today. Inflammation is now considered as the full circle of events, from initiation of a response, through the development of the cardinal signs above, to healing and restoration of normal appearance and function of the tissue or organ. However, in certain conditions there appears to be no resolution and a chronic state of inflammation develops that may last the life of the individual. Such conditions include the inflammatory disorders rheumatoid arthritis, osteoarthritis, inflammatory bowel diseases, retinitis, multiple sclerosis, psoriasis and atherosclerosis.
In order to study inflammation a multidisciplinary approach is necessary. Classically, it has required the study of the immune system, in order to understand the events involved in initiating and maintaining inflammatory conditions. Today it is recognised that the underlying genetics and molecular biology basis to cellular responses are also important in order to identify genetic predisposition to inflammatory diseases, while pharmacological studies are necessary to identify targets and develop novel treatments to bring relief from chronic life-threatening inflammatory conditions. Thus research into inflammation includes not only the study of immunological and cellular responses involved but also the pharmacological process involved in drug development.
Many of the drugs used in the treatment of inflammatory conditions, predate our current understanding of the biochemical processes involved in the disease. Traditionally, the standard treatments for rheumatoid arthritis has been to use a non-steroidal anti-inflammatory drug (NSAID), such as aspirin, for pain relief and to use corticosteroids or even disease-modifying anti-rheumatic drugs in an attempt to reduce other symptoms of the disease.
For many years the pharmaceutical industry attempted to develop NSAIDs which shared the therapeutic action of aspirin but which did not cause the main adverse event, namely gastric ulceration. This research led to the development of indomethacin, the fenamates, ibuprofen and many others. However, while all these drugs had clinical utility they also eroded the gastric mucosa. In addition, this research also led to the development of some of the animal models still used in arthritis research today, such as carrageenin oedema \[[@B4]\] and adjuvant arthritis \[[@B5],[@B6]\]).
The development of NSAIDs, with reduced potential to cause gastric ulcers, was finally realised with the demonstration that clinically useful NSAIDs inhibited the enzyme cyclo-oxygenase \[[@B7]\], which was also present in the gastric mucosa. The finding that cyclo-oxygenase present in inflammatory lesions (COX2) was distinct from that found in the stomach (COX1) led to the development of selective COX2 inhibitors, such as celecoxib. These drugs provide relief from many of the symptoms of arthritis but have a reduced potential to cause gastric ulceration \[[@B8]\]. The differential responsiveness to these, and other, therapeutic agents and, indeed, the induction of the inflammatory response in some patients with asthma by aspirin, has led to the concept of pharmacogenomics to understand individual drug sensitivities with a view to producing therapy tailored to the individual.
Similarly, glucocorticoids are widely used in the treatment of inflammation. Unlike the NSAIDs these agents do not relieve pain but reduce inflammation by inhibiting leukocyte function. The active ingredient responsible for the anti-inflammatory activity of adrenal cortex extracts was discovered in the 1940s. This led to the use of cortisol as an anti-inflammatory and the development of potent synthetic agents typified by dexamethasone. However, because cortisol, and synthetic glucocorticoids, produce their therapeutic action at supra-physiological concentrations, adverse effects, such as suppression of the HPA-axis and Cushingoid changes are inevitable. Many of these adverse effects can be avoided by giving glucocorticoids topically. This has led to the development of inhaled glucocorticoids for the treatment of inflammatory diseases of the respiratory tract and steroid containing creams for the treatment of skin inflammation. However, applying this approach to the treatment of rheumatoid arthritis necessitates the use of intra-articular injection. Thus, there is a clear unmet medical need for a drug that provides relief from the symptoms of inflammation but can be given systemically.
The fact that a large number of patients with severe chronic inflammatory disease fail to respond to conventional systemic or topical therapy resulting in a huge clinical and socio-economic burdon underlies the need to develop novel therapies.
Thus, modern research has used molecular techniques to identify which genes are regulated by glucocorticoid receptors in an attempt to identify novel therapeutic targets. This work has attempted to fine tune the immune system through use of agents that inhibit specific pathways and mediators rather than to suppress immune cell activity. Examples of such approaches include the development of anti-TNFa therapies, anti adhesion molecule therapies and inhibitors of cytokines believed to be pivotal in a given pathology \[[@B9]\]. Furthermore, inhibitors of selective pro-inflammatory intracellular signalling pathways are currently in use e.g. cyclsporin or under development e.g. NF-κB, p38 MAPK and PDE4 inhibitors \[[@B10]-[@B17]\]. As we understand more about the complexity of the inflammatory response and the actions of the currently available drugs the value of particular clusters of targets becomes apparent. However, the success of anti-TNFα therapy in RA underlines the importance of understanding/discovering the initial driver(s) of the inflammatory response in individual diseases and patients.
While research into inflammation has resulted in great progress in the latter half of the 20th century, we recognise that the rate of progress is accelerating. Furthermore, it is our perception that there is a need for a vehicle through which this very diverse research can readily be made available to the scientific community. Thus, we have initiated the creation of the *Journal of Inflammation*, a peer reviewed journal which will enable such information to be rapidly disseminated.
What is the Journal of Inflammation?
====================================
*Journal of Inflammation*will consider for publication all forms of original research articles, reviews, commentaries, hypothesis, meeting abstracts (by special arrangement) and comments on all aspects of inflammation. The *Journal of Inflammation*considers the term inflammation today to include the full range of underlying cellular and molecular mechanisms involved, not only in the production of the inflammatory responses but, more importantly in clinical terms, in the healing process as well. Thus the Journal covers molecular, cellular, animal and clinical studies, and related aspects of pharmacology, such as anti-inflammatory drug development, trials and therapeutic developments, etc. It will also consider publication of negative findings.
*Journal of Inflammation*aims to become the leading Internet journal (I-journal) on inflammation and, as online journals eventually replace traditionally published print journals over the next decade, the main archived journal on inflammation. The *Journal of Inflammation*has an open peer-review process, aimed at improving the accountability of peer review and giving reviewers credit for the work they do. This means that we ask reviewers to agree to their named report being passed on to the authors. Each article submitted is reviewed by at least two independent reviewers, with the aim of reaching a decision on publication within 14 days of initial receipt.
*Journal of Inflammation*is edited by Drs Neville Punchard and Cliff Whelan, and is supported by an international Advisory Board of Associate Editors and an Editorial Board drawn from the Academic and Industrial research community.
Open Access
===========
*Journal of Inflammation*is an Open Access, peer-reviewed online journal offering rapid world-wide access to research into all aspects inflammation. Open Access policy changes the way in which articles are published. First, all articles become freely and universally accessible online, and so an author\'s work can be read by anyone at no cost. This rapid and immediate access to research findings in inflammation will aid in promoting the dynamic and productive dialogue between industrial and academic members of the inflammation research community that plays such an important part in the development of future generations of anti-inflammatory therapies.
Second, the authors hold copyright for their work and grant anyone the right to reproduce and disseminate the article, provided that it is correctly cited and no errors are introduced \[[@B18]\]. Third, a copy of the full text of each Open Access article is permanently archived in an online repository separate from the journal. *Journal of Inflammation\'s*articles are archived in PubMed Central \[[@B19]\], the US National Library of Medicine\'s full-text repository of life science literature, and also in repositories at the University of Potsdam \[[@B20]\] in Germany, at INIST \[[@B21]\] in France and in e-Depot \[[@B22]\], the National Library of the Netherlands\' digital archive of all electronic publications.
Open Access has four broad benefits for science and the general public. First, authors are assured that their work is disseminated to the widest possible audience, given that there are no barriers to access their work. This is accentuated by the authors being free to reproduce and distribute their work, for example by placing it on their institution\'s website. It has been suggested that free online articles are more highly cited because of their easier availability \[[@B23]\]. Second, the information available to researchers will not be limited by their library\'s budget, and the widespread availability of articles will enhance literature searching \[[@B24]\]. Third, the results of publicly funded research will be accessible to all taxpayers and not just those with access to a library with a subscription. As such, Open Access could help to increase public interest in, and support of, research. Note that this public accessibility may become a legal requirement in the USA if the proposed Public Access to Science Act is made law \[[@B25]\]. Fourth, a country\'s economy will not influence its scientists\' ability to access articles because resource-poor countries (and institutions) will be able to read the same material as wealthier ones (although creating access to the internet is another matter \[[@B26]\]).
Competing interests
===================
Dr Neville A. Punchard is also a Section Editor for Current Opinion in Investigational Drugs.
|
PubMed Central
|
2024-06-05T03:55:55.069569
|
2004-9-27
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074343/",
"journal": "J Inflamm (Lond). 2004 Sep 27; 1:1",
"authors": [
{
"first": "Neville A",
"last": "Punchard"
},
{
"first": "Cliff J",
"last": "Whelan"
},
{
"first": "Ian",
"last": "Adcock"
}
]
}
|
PMC1074344
|
Background
==========
Ischemia reperfusion injury is a widespread phenomenon that affects all muscle tissues \[[@B1],[@B2]\]. It is a significant cause of morbidity following injury especially to limb blood vessels with resultant muscle necrosis, fibrosis and joint contracture (Volkmann\'s contracture). Of the muscle involved some are slow-twitch \'red\' fibre type predominately designed for sustained isometric contraction to stabilise joints while the other fast-twitch \'white\' fibre type muscles act with speed and dexterity such as the lumbricals and flexor digitorum profundi. In the leg the soleus is predominately slow-twitch, the extensor digitorum longus (EDL) fast-twitch and the gastrocnemius is mixed slow/fast-twitch fibre type.
Mast cells were first implicated in IR injury of skeletal muscle in studies from our laboratory \[[@B3]\]. Initial experiments in the gastrocnemius muscle, resulted in resistance to IR injury in mast cell depleted (W^f^/W^f^) mice \[[@B3],[@B4]\]. More recently we demonstrated that re-engraftment of mast cells into W^f^/W^f^mice restores susceptibility to IR injury, thus proving that mast cells play a pivotal role in IR injury to skeletal muscle \[[@B5]\]. Our IR injury model consists of 70 minutes tourniquet hind limb ischaemia followed by 24 h reperfusion. Unlike other models \[[@B6]\], the extended reperfusion period permits full manifestation of the reperfusion injury. In order to determine the usefulness of therapies against mast cells, it is important to know the degree to which mast cells are involved in IR injury of other skeletal muscle fibre types. In this study we selected skeletal muscles, representative of slow-twitch (soleus), slow/fast-twitch (gastrocnemius) and fast-twitch (EDL) types, and compared their susceptibility to IR injury in four genotypically different sets of mice that harbour different mast cell densities in their skeletal muscle. These were the C57/DBA mast cell depleted (W^f^/W^f^) mice, their heterozygous (W^f^/+) and normal littermates (+/+) and control C57BL/6 mice.
Methods
=======
Animals
-------
Mast cell depleted mice, W^f^/W^f^, (C57BL/6W^f^× DBA/2W^f^)F~1~; wild-type littermate, +/+, (C57BL/6 × DBA/2)F~1~; and heterozygous littermate, W^f^/+, (C57BL/6 × DBA/2W^f^or C57BL/6W^f^× DBA/2)F~1~hybrids were purchased from Flinders Medical Centre (Bedford Park, South Australia) aged 6--10 wk (18--25 g). Each genotype was clearly identified by coat colour. C57BL/6 × C57BL/6 (C57BL/6) mice were purchased from Animal Resources Centre, Perth, Western Australia aged 10 wk (25--30 g). A C57BL/6 strain mouse group was included as an additional control to the C57/DBA strain, to test for strain differences in susceptibility to IR injury. This was important, as the C57BL/6 is the most commonly used mouse strain. There was no significant difference in the data between male and females so the results were pooled. Mice were given food and water ad libitum, and housed with a 12 h day/night cycle. The Animal Ethics Guidelines outlined by St. Vincent\'s Hospital and National Health & Medical Research Council were adhered to in all experiments.
Ischemia-Reperfusion injury
---------------------------
Mice were anaesthetized by intraperitoneal (i.p.) injection of 4% chloral hydrate (0.1 ml/10 g body weight), followed by i.p. injection of the analgesic carprofen to minimise postoperative pain. Tourniquet warm ischemia was induced by using 2 × size 8 rubber bands as previously described \[[@B5]\]. During the 70 min ischemia, a needle thermistor probe was inserted subcutaneously in the right leg, and the hind limb temperature was monitored and maintained at 36 ± 1°C. After the ischemia period, the bands were removed and the mice allowed to recover. After 24 h reperfusion the mice were re-anaesthetized and the gastrocnemius, soleus and EDL muscles were carefully removed from both treated and contralateral sides and weighed, before the mouse was sacrificed.
Age matched (12--24 wk, 25--35 g) male and female mice from each genotype (W^f^/W^f^, W^f^/+, +/+ and C57BL/6) were grouped into the same sex, age and strain and underwent warm ischemia at the same time. A minimum of four mice (n = 4) from each of the four genotypes was used.
NBT Assay
---------
Nitro Blue Tetrazolium (NBT) assay was used to determine muscle viability in fresh whole mounts slices as previously described \[[@B5]\]. Both sides of each muscle slice were post-fixed in 10% buffered formal saline (BFS) and analysed under a dissecting microscope for viable tissue, which was identified by its blue reaction product. The percentage of viable tissue in treated muscle was determined by standard point counting technique \[[@B7]\] and was expressed as a percentage of viable tissue in the contralateral control. Finally, the mean percent viable tissue of treated versus contralateral muscles was calculated for each group where n = number of mice.
Histology
---------
For histological analysis, muscle slices were immersion fixed for 24 h in 10% BFS, washed in 0.1 M phosphate buffered saline (PBS) and processed into paraffin. Five micron sections were cut, dewaxed and stained with Haematoxylin and Eosin (H&E) for general analysis \[[@B5]\].
Mast Cell Staining
------------------
Mast cells were selectively stained by routine toluidine blue \[[@B8]\] and chloroacetate esterase (CAE) methods \[[@B9]\].
Mast Cell Density
-----------------
An overview of mast cell content in the four groups of mice tested was obtained by counting a minimum of 100 mast cell profiles in tongue, skin and heart as well as skeletal muscle. Using 100 X magnification all mast cell profiles that fell within a grid area of 1.35 mm^2^but did not touch the right hand and bottom side boundaries were counted and the data expressed as mast cells/mm^2^. There was no difference in the toluidine blue or CAE labelled mast cell profile numbers in comparable tissue sections (data not shown), hence the CAE technique was subsequently used in preference to toluidine blue.
Statistics
----------
Statistical analyses were performed using SPSS software (Statistical Package for the Social Sciences, version 11.5). All results are expressed as means ± standard error of the mean (SEM) of grouped data where n = number of mice/group. For comparison between groups, means were analysed using univariate analysis of variance. Pearson\'s correlation was used to test the correlation between tissue viability and mast cell profile counts. A probability level of p \< 0.05 was taken to indicate statistical significance.
Results
=======
Morphologic appearance of labelled mast cells
---------------------------------------------
After CAE or toluidine blue staining, mast cells were easily distinguished from other cells, by their red or purple stained cytoplasmic granules respectively. In general, they were intact but varied in size, and were predominantly located near nerves and blood vessels as reported by others \[[@B10]\].
Histological appearance of skeletal muscle before and after IR injury
---------------------------------------------------------------------
Prior to injury, the morphologic appearance of transverse muscle sections from littermate controls and W^f^/W^f^mice was similar (see Figure [1A](#F1){ref-type="fig"} and [1C](#F1){ref-type="fig"}). As expected, the muscle fibres were nucleated, intact and arranged in groups. In longitudinal sections (not shown), the muscle striations were clearly evident, thus demonstrating viable fibres. After IR injury, muscles from normal littermate mice (Figure [1B](#F1){ref-type="fig"}) were infiltrated by numerous inflammatory cells that were often observed invading the muscle fibres. Many fibres were fragmented and appeared moth eaten, while other fibres were condensed and shrunken. At high magnification (not shown), the sarcomeric pattern was not visible in the majority of the fibres and many fibres did not contain nuclei. In contrast, a large proportion of fibres from W^f^/W^f^mice were intact, nucleated, stained amorphously and comparable in size to controls. Surprisingly, this tissue also contained a cellular inflammatory exudate that was largely confined to the interstitial area (Fig. [1D](#F1){ref-type="fig"}).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Gastrocnemius muscle stained with Haematoxylin and Eosin. Control (A & C) and IR treated hind limbs (B & D) from littermate control (A & B) and W^f^/W^f^(C & D) mice. Scale bar = 50µm.
:::
:::
Mast cell density of different organs
-------------------------------------
The density of CAE stained mast cells varied between tissues and between mouse strains (see Table [1](#T1){ref-type="table"}). In mast cell replete mice (C57BL/6, +/+, W^f^/+), the tongue and skin had a consistently high mast cell density ranging from 31.3 to 49.9 mast cell profiles/mm^2^. Cardiac muscle had significantly fewer mast cells ranging from 0.3 to 2.3 mast cell profiles/ mm^2^. In W^f^/W^f^mice, the mast cell density was markedly reduced in both tongue (1.95 ± 0.42) (Figure [2](#F2){ref-type="fig"}) and skin (6.73 ± 3.27). No mast cells were observed in W^f^/W^f^cardiac muscle following screening of a large number of sections (20 fields at X 200 magnification for each mouse).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Mast cell profile number of tissues from different mouse strains.
:::
Animal strain Heart Tongue Skin
--------------- ------------- -------------- ---------------
C57BL/6 1.52 ± 0.42 40.31 ± 3.25 31.28 ± 1.89
+/+ 2.27 ± 0.39 43.36 ± 3.79 38.73 ± 9.60
W^f^/+ 0.29 ± 0.10 37.15 ± 2.23 49.90 ± 10.10
W^f^/W^f^ ND 1.95 ± 0.42 6.73 ± 3.27
Mean ± SEM (n = 4); ND-not detected
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
An example of histological sections of tongue stained with chloroacetate esterase for identification of mast cells in littermate controls (A) and W^f^/W^f^(B) mice. Note mast cells appear as brilliant red colour. Scale bar = 100µm.
:::
:::
Mast cell density of skeletal muscles
-------------------------------------
Table [2](#T2){ref-type="table"} shows the mast cell density of each skeletal muscle type in four different groups of mice. In general the slow twitch soleus muscle had almost twice the mast cell density of both the EDL and gastrocnemius muscles (p \< 0.05). There were no mast cells in the C57/DBA W^f^/W^f^skeletal muscles as expected. However, there was twice the density of mast cells in their normal compared to their heterozygous littermates, suggesting a c-kit gene dosage effect on mast cell density (p \< 0.05). There were significantly fewer mast cells in the C57/C57 compared with C57/DBA littermate control mice (p \< 0.05) but no difference in the mast cell density of C57/C57 compared with C57/DBA heterozygous mice.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Mast cell profile number of skeletal muscles from different mouse strains.
:::
Animal strain Soleus Gastrocnemius EDL
--------------- ----------- --------------- -----------
C57BL/6 2.1 ± 0.4 1.0 ± 0.2 1.3 ± 0.2
+/+ 3.7 ± 0.6 1.8 ± 0.3 2.3 ± 0.3
W^f^/+ 2.2 ± 0.4 1.0 ± 0.1 1.6 ± 0.2
W^f^/W^f^ ND 0.0 ± 0.0\* ND
Mean ± SEM (n = 4); ND-not detected
\*only one mast cell was detected from all tissue examined
:::
Skeletal Muscle Viability after IR
----------------------------------
Skeletal muscles from W^f^/W^f^mice were significantly more resistant to tourniquet induced warm ischaemia/reperfusion injury, as assessed by NBT assay, compared to the other 3 mouse strains irrespective of muscle type (Fig [3](#F3){ref-type="fig"}). In the W^f^/W^f^mice the soleus muscle was a remarkable 95% viable after IR. With respect to the muscle types, the slow-twitch soleus muscle sustained significantly less (P \< 0.05) IR injury compared to the slow/fast gastrocnemius and the fast EDL muscles in each group. The gastrocnemius and EDL muscles showed a similar degree of injury in all groups.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Muscle viability (% contralateral control) as assessed by NBT assay in the soleus (white bars), gastrocnemius (hatched bars) and EDL (black bars) muscles. All values are mean ± SEM, n = 4. \* P \< 0.05 Significantly different to C57BL/6, +/+ and W^f^/+ mice. \# P \< 0.001 Significantly different to C57BL/6, +/+ and W^f^/+ mice. @ P \< 0.001 Significantly different to C57BL/6, +/+ and W^f^/+ mice.
:::
:::
Correlation between mast cell density and viability
---------------------------------------------------
There was an inverse correlation (Pearson\'s correlation factor 0.043) between mast cell density and muscle viability for each muscle type in C57/DBA mice.
Mouse strain susceptibility
---------------------------
The viability of muscles from C57/DBA heterozygous mice was significantly greater than the C57/C57 controls (p \< 0.05) even though the mast cell density was the same.
There was no difference in the viability of muscles from the C57/C57 controls compared with C57/DBA littermate controls even though there was double the mast cell density in the latter muscles.
Discussion
==========
In this study a tourniquet was placed high up on the thigh to induce a short warm (36°C) ischaemia of 70 minutes duration followed by a long reperfusion period of 24 h, in order to assess the impact of reperfusion injury on different skeletal muscles and the degree to which mast cells mediate this injury. This differs from other studies where an extended ischaemia and short reperfusion period is used to study the early effects of ischaemia on skeletal muscle. Using this approach, we show that mast cells contribute to ischaemia reperfusion injury of fast-, mixed fast/slow- and slow-twitch muscle types. Viability of these muscles was inversely correlated with mast cell density and all muscles exhibited a remarkable resistance to IR injury in mast cell depleted mice. In the absence of mast cells, the predominately slow-twitch oxidative soleus muscle was more resistant to IR than the fast/slow-twitch gastrocnemius and the fast-twitch EDL muscles. We also demonstrate that muscle fibre type, and mouse strain independently, determined susceptibility to IR injury.
The susceptibility of different skeletal muscle types to ischaemia is hypothesized to relate to their different metabolic disposition, but data regarding this is conflicting. Skeletal muscles in mice are composed of two main distinct fibre types. In general, slow-twitch fibres have a high oxidative enzyme activity, high capillary density and increased numbers of mitochondria; in contrast fast-twitch fibres have high glycolytic enzyme activity, low capillary density and few mitochondria. It has been suggested that fast-twitch muscles display a greater resistance to ischaemia than slow-twitch muscles because of their greater potential to maintain ATP levels during ischaemia \[[@B11]\]. Alternatively, it has been proposed that the greater accumulation of anaerobic metabolites during ischaemia in the fast-twitch fibres, compared to the slow-twitch fibres, give rise to oxygen free radicals during reperfusion that makes them more susceptible to injury. In our study, the predominately slow-twitch soleus muscle was consistently more resistant to IR injury than the slow/fast- and fast-twitch muscles. Data from other studies are difficult to compare because of the wide variety of IR models in use. In particular, there is a great deal of variation in the period of ischaemia, the muscle temperature during ischaemia and the period of reperfusion allowed for manifestation of the reperfusion injury. Idstrom \[[@B12]\] utilised a period of 2, 4 and 6 h cold 25°C hind limb ischaemia in rats and one hour reperfusion to measure damage and recovery of adenine nucleotides. Consistant with our data, he showed that the fast-twitch tibialis muscles displayed a faster degradation rate and slower recovery of these molecules than the slow-twitch soleus. This was attributed to differences in the regulation of enzymes during ischaemia and differences in blood flow during reperfusion. Woitaske *et al.*\[[@B13]\] used 3 h of hind limb ischaemia at a unknown temperature and up to 14 days reperfusion in mice. The soleus was less injured and recovered function and mass more quickly than the EDL muscle over this period. In contrast to our data, other groups \[[@B6]\] have shown that after a lengthy ischaemia time of 3 h and a short reperfusion (2 h) fast-twitch muscles are more resistant to injury than slow-twitch muscles. Other workers report variable results. Rácz *et al.*\[[@B14]\] showed that slow-twitch muscles were more severely damaged after 1 hour of ischaemia however the fast-twitch muscle was more damaged after 2 h. Sternbergh \[[@B15]\] used an in vitro model of 120 min ischaemia and 55 min reperfusion at 37°C in rat hind limb. The slow-twitch soleus and fast-twitch plantaris showed similar degrees of injury whereas the fast-twitch tibialis was uninjured. He concluded that muscle fibre type does not predict injury. In Carvalho\'s study \[[@B11]\] the fast-twitch muscle was better able to contract during the first 45--60 minutes of ischaemia but both fast- and slow-twitch muscles contracted to equal degrees thereafter.
We have recently shown conclusively that mast cells play a pivotal role in IR injury of murine soleus, EDL and gastrocnemius muscles \[[@B5]\]. W^f^/W^f^IR resistant mice were engrafted with bone marrow derived mast cells (BMMC) from their normal littermates, and their hind limbs underwent IR injury 12 weeks later. The proportion of viable muscle fibres in engrafted mice was significantly reduced, back to the levels observed in their IR susceptible littermates. Thus, engraftment of BMMC into W^f^/W^f^mice restores the susceptibility of skeletal muscles to IR injury irrespective of the other abnormalities in these mice. The role of mast cells has not been considered when investigating the susceptibility of different muscle fibre types.
In the current study, mast cell density was inversely correlated with survival for all muscle types. In all four strains examined, muscles from the W^f^/W^f^mice had a significantly greater viability. In particular, the soleus muscle viability was 95% in the mast cell depleted mice indicating that a large amount of injury was mast cell mediated. The gastrocnemius and EDL muscle viability was 76% and 65% respectively in the absence of mast cells, indicating that other factors, possibly related to muscle type contribute to the IR injury. The demonstration that muscles from the C57/C57 and C57/DBA strains of mice were equally affected by IR, even though the latter contained twice the density of mast cells, indicate that there is also a genetic component to IR injury.
Our data would support the hypothesis that there is a base line level of susceptibility to ischaemia induced injury that can be attributed to mouse strain and muscle fibre type. Mast cells independently exacerbate IR injury during a clinically relevant extended reperfusion. These findings predict that mast cell therapies would be beneficial across different muscle types and that further protection can be tailored to specific muscle types. It is clear that mast cell depleted mice are the desirable model to study the effects of IR on muscle fibre type. Alternatively, the soleus is the most suitable muscle to study the role of mast cells since it has the least fibre type component effect.
Abbreviations
=============
BFS, buffered formal saline; CAE, chlororacetate esterase; EDL, extensor digitorum longus; IR, ischaemia reperfusion; NBT, nitro blue tetrazolium; NOS II, nitric oxide synthase II; W^f^/W^f^, mast cell depleted mice; W/W^v^, mast cell deficient mice.
Authors\' contributions
=======================
SKB performed animal experimentation, muscle viability studies, mast cell densities and drafted the original manuscript. WAM participated in the design of the study. AM participated in design of the study and performed morphological analyses. All authors provided intellectual input, participated in the manuscript preparation and have approved the final manuscript.
Acknowledgements
================
The authors are grateful to Ms. Xiao-Lian Han for expert technical assistance and Ms Wei Wang for her help with statistical analyses. We would also like to thank Assoc. Prof. Prue Hart and her group allowing us to establish our own mast cell depleted mouse colony. This work was supported by the National Health and Medical Research Council.
|
PubMed Central
|
2024-06-05T03:55:55.071141
|
2004-9-27
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074344/",
"journal": "J Inflamm (Lond). 2004 Sep 27; 1:2",
"authors": [
{
"first": "Susan K",
"last": "Bortolotto"
},
{
"first": "Wayne A",
"last": "Morrison"
},
{
"first": "Aurora",
"last": "Messina"
}
]
}
|
PMC1074345
|
Review
======
Tumour and Inflammation
-----------------------
Tumour has been linked with inflammation since 1863, when Rudolf Virchow discovered leucocytes in neoplastic tissues and made the first connection between inflammation and cancer \[[@B1]\]. Since then, chronic inflammation has been identified as a risk factor for cancer and even as a means to prognose/diagnose cancer at the onset of the disease. Examples of such association include the Human papiloma virus (HPV) and cancer \[[@B2]\], including cervical \[[@B3]\], cancers of the oesophagus \[[@B4]\] and larynx \[[@B5]\], Helicobacter pylori bacterial infection and gastric adenocarcinoma \[[@B6]\], the hepatitis B virus, cirrhosis and hepato-cellular carcinoma \[[@B7]\], Schistosoma haematobium and cancer of the bladder \[[@B8]\], asbestos induced inflammation and bronchogenic carcinoma or mesothelioma in humans \[[@B9]\].
Several reports implicate inflammation as a significant risk factor in cancer development: asbestos, cigarette smoke and inflammation of the bowel and pancreas are but a few well-known examples given \[[@B1],[@B10]\]. These papers demonstrate that the inflammation environment is one that would support tumour development and is consistent with that observed in tumour sites. The relationship of cancer with inflammation is, however, not limited to the onset of the disease due to chronic inflammation. Schwartsburd \[[@B11]\] goes a step further and proposes that chronic inflammation occurs due to tumour environment stress and that this would generate a protective shield from the immune system. It has been recently demonstrated that the tumour microenvironment highly resembles an inflammation site, with significant advantages for the progression of tumour, including the use of cytokines, chemokines, leucocytes, lymphocytes and macrophages to contribute to both vassal dilation and neovascularisation for increased blood flow, the immunosuppression associated with the malignant disease, and tumour metastasis \[[@B1],[@B11]\]. Furthermore, this inflammation-site tumour-generated microenvironment, apart from its significant role in cancer progression and protection from the immune system, has a considerable adverse effect to the success of the various current cancer treatments. It has recently been demonstrated that the inflammatory response in cancer can greatly affect the disposition and compromise the pharmacodynamics of chemotherapeutic agents \[[@B12]\].
It is evident that cancer is using natural inflammatory processes to spread and, unlikely as it seems at first, it is proposed that this is through the use of the angiotensin II type 1 (AT1) receptor.
AT receptors and inflammation
-----------------------------
Angiotensin II (Ang II) is a peptide hormone within the renin-angiotensin system (RAS), generated from the precursor protein angiotensinogen, by the actions of renin, angiotensin converting enzyme, chymases and various carboxy- and amino-peptidases \[[@B13]\]. Ang II is the main effector of the RAS system, which has been shown to play an important role in the regulation of vascular homeostasis, with various implications for both cardiovascular diseases and tumour angiogenesis. It exerts its various actions to the cardiovascular and renal systems via interactions with its two receptor molecules, angiotensin II type 1 receptor (AT1) and angiotensin II type 2 receptor (AT2) \[[@B13]\]. AT1 and AT2 receptors have been identified as seven transmembrane-spanning G protein-coupled receptors \[[@B13]\], comprising an extracellular, glycosylated region connected to the seven transmembrane α-helices linked by three intracellular and three extracellular loops. The carboxy-terminal domain of the protein is cytoplasmic and it is a regulatory site. AT1 is 359 amino acids, while AT2 is 363 amino acids being \~30% homologous to AT1 and are both N-linked glycosylated post-translationally. Various studies have looked at the pharmacological properties of the two receptors and the expression of those receptors on various cell lines. Their affinity for the angiotensin II peptide and their ability to perform their physiological functions has been characterised using radioligand binding analyses and Scatchard plots. The results have indicated that both receptors have high binding affinities for the AngII peptide. The AT1 receptor has demonstrated a Kd of 0.36 nM for the AngII peptide \[[@B14]\], whereas the AT2 receptor has demonstrated a Kd of 0.17 nM respectively, under similar studies \[[@B15]\].
AT1 receptors are expressed in various parts of the body and are associated with their respective functions, such as blood vessels, adrenal cortex, liver, kidney and brain, while AT2 receptors are highest in fetal mesenchymal tissue, adrenal medulla, uterus and ovarian follicles \[[@B13]\]. The opposing roles of the AT1 and AT2 receptors in maintaining blood pressure, water and electrolyte homeostasis are well established. It is, however, becoming recognised that the renin-angiotensin system is a key mediator of inflammation \[[@B16]\], with the AT receptors governing the transcription of pro-inflammatory mediators both in resident tissue and in infiltrating cells such as macrophages.
In addition to the mediators reviewed by Suzuki *et al*(2003) \[[@B16]\], a number of vital molecules in inflammatory processes are induced by the AT1 receptor. These include interleukin-1 beta (IL-1b) in activated monocytes \[[@B17]\], Tumour Necrosis Factor-alpha (TNF-α) \[[@B18]\], Plasminogen Activator Inhibitor Type 1 (PAI-1) \[[@B19]\] and adrenomedullin \[[@B20]\] all of which have been shown to have active participation in various aspects of cancer development. Activation of AT1 also causes the expression of TGF-β \[[@B21],[@B22]\] and a review of literature indicates this may be a unique capability for this receptor. TGF-β is a multifunctional cytokine that is produced by numerous types of tumours and amongst its many functions is the ability to promote angiogenesis, tissue invasion, metastasis and immune suppression \[[@B23]\]. It has been postulated that the low response rates achieved in cancer patients undergoing immunotherapy is in part caused by tumour expression of TGF-β and this is supported by inhibition of the antigen-presenting functions and anti-tumour activity of dentritic cell vaccines \[[@B24]\].
On examination of the tumour environment, it is interesting to note that angiotensin II actually increases vasodilation, a phenomenon that researchers have attempted to utilise for drug delivery \[[@B25]\]. This would imply something unusual about the presentation of angiotensin receptors; however it is predominantly over expression of the vasoconstrictor AT1 that is reported in association with human cancers of the breast \[[@B26]\], pancreas \[[@B27]\], kidney \[[@B28]\], squamous cell carcinoma \[[@B29]\], keratoacanthoma \[[@B29]\], larynx \[[@B30]\], adrenal gland \[[@B30]\], and lung \[[@B31]\]. AT2 has been identified as expressed in preference to AT1 in only one case, in an earlier paper on colorectal cancer \[[@B32]\].
Is it evolution that causes over expression of AT1?
---------------------------------------------------
In the \'Hallmarks of Cancer\', the authoritative work by Douglas Hanahan and Robert A. Weinberg, the evolutionary acquired capabilities necessary for cancer cells to become life-threatening tumours are described. Furthermore, it is suggested that cancer researchers should look not just at the cancer cells, but also at the environment in which they interact, with cancers eliciting the aid of fibroblasts, endothelial cells and immune cells \[[@B33]\].
Sustained angiogenesis, tissue invasion and metastasis are the latter of six necessary steps in tumour progression, as described in the \'Hallmarks of Cancer\' \[[@B33]\]. These envisaged evolutionary steps allow cancers to progress from growths of \<2 mm to full tumors. A single evolutionary step, however, upregulation of AT1 would provide a considerable advantage to cancer cells that have learnt to evade the apoptosis and growth regulatory effects of TGF-β. Supporting this hypothesis is the observed genetic change from non-invasive cancer esophageal cell line T.Tn to invasive cancer cell line T.Tn-AT1. This genetic change concerns 9 genes, all of which are known to influence inflammation signalling (8 down and 1 up regulated) \[[@B34]\].
Is it environment?
------------------
The alternative basis under which induction of AT1 in tumours may occur is by looking at the environment under which the cancer is developing. Stresses and cell damage on the growing tumour boundary could potentially be causing the expression of AT1. Evidence that appears to support this view can be found in a study of AT1 expression in breast cancers \[[@B35]\]. In this case, in situ carcinoma has over-expressed AT1 receptors in addition to expressing proteins for yet more AT1. In the invasive carcinoma, high proportions of AT1 receptors are found on the tumour boundary, but in this case protein generation for AT1 is very noticeably absent. How could this behaviour be explained? Perhaps the answer lies in oxidative stress and hypoxia.
The formation of oxidised LDL by monocytes and macrophages at the sites of tissue damage has been established in a recent report by Jawahar L. Mehta and Dayuan Li \[[@B36]\]. In this study, the ox-LDL LOX-1 receptor is noted to be induced by fluid shear stress (4 hrs), TNF-α (8 hrs) and self-induced by ox-LDL (12 hrs). Of particular interest is that activation of LOX-1 by ox-LDL induces the expression of the AT1 receptor \[[@B36]\]. This key role of ox-LDL regarding AT1 is demonstrated by HMG Co-A reductase inhibitor causing the down-regulation of the AT1 receptor with consequential reduction in inflammatory response \[[@B37]\]. Also of interest is that another marker of many diseases, homocysteine, enhances endothelial LOX-1 expression \[[@B38]\].
Hypoxia has been demonstrated to induce the expression of both AT1b (AT1a and AT1b are subsets of AT1) and AT2 receptors in the rat carotid body and pancreas \[[@B39],[@B40]\]. The expression of AT1 and AT2 receptors has been studied during the development and regression of hypoxic pulmonary hypertension \[[@B41]\]. Hypoxia has been shown to strongly induce the expression of AT1b but not AT1a. The expression of AT2 is believed to protect the cell from apoptosis and this effect has been demonstrated in the brain when AT1 is antagonized \[[@B42]\]. Since HIF-1α governs many hypoxia driven transcriptions \[[@B43]\], its control of AT1b and AT2 expression can be hypothesized. AT1 activation has also been shown to increase the activity of HIF-1α \[[@B43]\], and is consistent with other cases of AT1 providing a positive feedback mechanism. Since hypoxia counts for the expression of AT1b, the speculation that the AT1a subtype is induced by oxidative stress is tempting, although a review of literature appears absent in this regard and further investigation is required to confirm this hypothesis.
A review of hypoxia and oxidative stress in breast cancer cites the chaotic flow of blood in the tumor environment with resultant periods of hypoxia and reperfusion \[[@B44]\]. Reperfusion after myocardial infarction or cerebral ischemia is known to cause the generation of ROS. Hence, summarised in figure [1](#F1){ref-type="fig"}, the tumor environment thus offers both hypoxia and oxidative stress mechanisms for induction of AT1. It would however seem likely that genetic factors speed up the progression of the more aggressive forms of cancer.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
***AT1 expression in cancer.***A cycle of oxidative stress (enhanced by homocysteine and ox-LDL) and hypoxia on the growing tumour boundary co-operatively promotes AT1 expression, leading to inflammation-associated angiogenesis, invasion, metastasis and immune suppression.
:::
:::
A combination therapy for cancer
--------------------------------
The evidence relating to over-expression of AT1 with cancer progression is compelling. To this effect, AT1 blockade has been hypothesised as the mechanism to overcome cancer associated complications in organ graft recipients \[[@B45]\]. Additionally, a study undertaken in 1998 suggested that hypertensive patients taking ACE inhibitors were significantly less at risk of developing cancer than those taking other hypertensive treatments \[[@B46]\].
Tumour progression has been significantly slowed with AT1 receptor antagonists \[[@B47],[@B48]\]. The results appeared to far exceed the expectations of simple inhibition of angiogenesis. Reduction of MCP-1 was noted \[[@B48]\], as was the expression of many pro-inflammatory cytokines. The activity of tumour-associated macrophages was also noticed to be severely impaired \[[@B48]\]. The importance in reducing the action of tumour-associated macrophages in extracellular matrix decomposition is not to be underestimated, since, in this action, they further progress remodelling by releasing stored TGF-β \[[@B49]\]. The similarity of action of tumour associated macrophages with those in the tissue healing and repair environment has been noted \[[@B49]\]. The tumour suppressant action of tranilast, an AT1 antagonist, \[[@B50]\] has been more widely explored \[[@B51]-[@B54]\]. In one study on the inhibition of uterine leiomyoma cells, Tranilast also induced p21 and p53 \[[@B55]\]. Similarly, the AT1 blocker losartan has been shown to antagonise platelets, which are thought to modulate cell plasticity and angiogenesis via the vascular endothelial growth factor (VEGF) \[[@B56]\]. It has been postulated that losartan and other AT1 blockers can act as novel anti-angiogenic, anti-invasive and anti-growth agents against neoplastic tissue \[[@B56]\]. Furthermore, it has been shown that angiotensin II induces the phosphorylations of mitogen-activated protein kinase (MAPK) and signal transducer and activator of transcription 3 (STAT3) in prostate cancer cells. In contrast, AT1 inhibitors have been shown to inhibit the proliferation of prostate cancer cells stimulated with EGF or angiotensin II, through the suppression of MAPK or STAT3 phosphorylation \[[@B57]\]. Angiotensin II also induces (VEGF), which plays a pivotal role in tumour angiogenesis and has been the target of various therapeutics, including antibodies and aptamers \[[@B58]\]. Although the role of angiotensin II in VEGF-mediated tumour development has not yet been elucidated, an ACE inhibitor significantly attenuated VEGF-mediated tumour development, accompanying the suppression of neovascularisation in the tumour and VEGF-induced endothelial cell migration \[[@B59]\]. Perindopril, another ACE inhibitor has also been shown to be a potent inhibitor of tumour development and angiogenesis through suppression of the VEGF and the endothelial cell tubule formation \[[@B60]\].
The powerful direct and indirect suppression effects of TNF-α \[[@B61]\], IL-1β \[[@B62]\] and TGF-β \[[@B63]\] on APC presenting cells, NK, T and B cell have been reviewed \[[@B64]\]. The expression of these mediators makes an effective immune response most unlikely.
Despite this, it has long been established that the body does have the capability to recognise cancer cells and develop antigens. Dentritic cell vaccines for instance have been developed and have demonstrated limited effect in treating established tumours. The effectiveness of one such approach was greatly enhanced leading to complete regression of tumours in 40% of cases when TGF-β was neutralised using TGF-β monoclonal antibodies in synergy with a dentritic cell vaccine \[[@B24]\].
Strong evidence suggests that tumour cells over-express AT1 receptors and compelling evidence has been presented on the implications of AT1 in cancer progression. Although still at a theoretical stage, this evidence leads to the formulation of the hypothesis that effective blockade of AT1 with a tight binding receptor antagonist, in combination with NSAIDs to further control the inflammation, and immunotherapy, such as cancer vaccines, would provide an effective treatment. Most, if not all, solid tumours utilise inflammation processes, which, through the over-expression and activation of AT1 and the subsequent expression of a number of inflammatory cytokines and chemokines, allow for tumour protection from the immune system through immunosuppression, as well as tumour progression and metastasis. Blocking these pathways through inhibition of AT1 using one of the commercially available AT1 inhibitors, whilst lifting the induced protective effect of immunosuppression and further reducing inflammation with the use of NSAIDs will both inhibit tumour progression and allow currently developed immunotherapies, such as cancer vaccines, to promote their therapeutic effect uninhibited. The role of AT1 post-metastasis, given the observation that AT1 protein expression ceases, as demonstrated in the breast cancer study, requires further investigation \[[@B35]\]. However, the premise for the necessity of immunosuppression by cancer is none the less fundamental and this is encouraging for the prospects of regression of cancers that have progressed to metastasis by combinational AT1 blockade/immune therapy.
Learning from Cancer: wound management
--------------------------------------
Cancer is a systemic disease, one that can affect every part and organ in the body and, as presented in this review so far with regards to the role of AT1 in cancer, AT1 upregulation is of the utmost importance in the activation of inflammation. Systemically, therefore, what purpose does this upregulation of AT1 serve? The release of ACE and extended expression of AT1 and AT2 during the healing process following vascular injury helps to answer this question \[[@B65]\]. AngII is demonstrated to promote migration and proliferation of smooth muscle cells, as well as production of extracellular matrix through AT1 activation. In this work \[[@B65]\], the AT1 and AT2 receptors are recognized as having a substantial role in the tissue repair and healing processes of injured arteries. Although further literature in regard to the role of AT1 and AT2 in the healing process appears absent and additional studies are required, it appears rational that a systemic agent for the management of inflammation and healing would be one associated with the vascular system.
The activation of AT1 (shown in figure [2](#F2){ref-type="fig"}) has a powerful pro-inflammatory effect \[[@B16]\], promoting the expression of many pro-inflammatory mediators, such as cytokines, chemokines and adhesion molecules through the activation of signalling pathways. The influx, proliferation and behaviour of immune cells are steered away from an effective immune response to pathogens (thereby achieving immunosuppression) but instead towards activities consistent with a wound environment. Through the activation of these pathways \[[@B16]\], AT1 effectively elicits this response with local effects intended to initiate wound recovery through destruction of damaged cells, remodelling, the laying down of fibrous material and angiogenesis. AT1 acts in three ways, as indicated in figure [3](#F3){ref-type="fig"}. Firstly, via the up-regulation of growth factors that leads to increased vascular permeability. Secondly, through the increase of pro-inflammatory mediators that leads to utilisation of immune cells such as macrophages in their response to wound mode. Thirdly, through the generation of other factors which promote cell growth, angiogenesis and matrix synthesis. The observation that cancer resembles a wound that never heals is therefore substantiated.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
***AT1 signalling.***Activation of AT1 has a powerful pro-inflammatory effect, promoting the expression of many pro-inflammatory mediators, such as cytokines, chemokines and adhesion molecules through the activation of signalling pathways. The influx, proliferation and behaviour of immune cells are steered away from an effective immune response to pathogens (thereby achieving immunosuppression) but instead towards activities consistent with a wound environment.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
***local effects of AT1 activation.***Activation of AT1 leads to growth factors causing increased vascular permeability, pro-inflammatory mediators that lead to utilisation of immune cells such as macrophages in their response to wound mode and other factors that promote cell growth, angiogenesis and matrix synthesis during fibrosis and resolution.
:::
:::
Confirming the systemic role of the AT1 receptor in inflammation and disease
----------------------------------------------------------------------------
With the role of AT1 in cancer established, when the literature of other diseases is reviewed, it is reasonable to anticipate that the role of this receptor is system-wide with regard to inflammation. Interest in the wider implications of the AT1 receptor within disease is gradually increasing and these studies further substantiate a systemic role for AT1 as a key inductor of inflammation and disease. In these studies, a wide variety of pro-disease mediators, such as TNF-α, NFκB, IL-6, TGF-β, surface adhesion molecules and PAI-I are shown to be induced by AT1 (Table [1](#T1){ref-type="table"}).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
AT1 as a key inductor of inflammation and disease. A wide range of pro-inflammatory mediators, cytokines, chemokines and surface adhesion moleculesinvolved in a number of diseases are induced by AT1 and thus inhibited by its blockade.
:::
**Disease** **Mediators inhibited by AT1 blockade** **Reference**
---------------------------------------------- ------------------------------------------------------------------------------ ---------------
Cardiovascular disease NFκB, \'markers of oxidation inflammation and fibrinolysis\' 66
Cardiovascular disease TGF-β 67
Cardiovascular disease TNF-α, IL-6, ICAM-1, VCAM-1 18
Cardiovascular disease PAI-1 19
Cardiovascular disease Surface adhesion molecules 68,69
Cardiovascular disease MCP in Hypercholesterolemia associated endothelial dysfunction 70
Kidney disease None noted in this study. 71
Pancreatitis (Key markers of the disease) 72
Liver fibrosis and cirrhosis \'TGF-β and pro-inflammatory cytokines\' 21
Skin disease None noted in this study. 73
Osteoporosis \'Markers of inflammation\' 74
Alzheimer\'s, Huntington\'s and Parkinson\'s (TGF-β \[75\], over expression of AT1 and AT2 noted in affected brain areas) 75--78
:::
It is clear that a number of diseases, including heart and kidney disease, diseases associated with the liver and pancreas, as well as diseases of the skin, bone, the brain and most of the autoimmune and inflammatory disorders, are all affected by the AT1 blockade. It is worth noting at this stage that many of these diseases are often considered to be associated with ageing and with fibrosis. An investigation of the action of IGF-1 in the regulation of expression of AT2 leads to an explanation of this association.
Role of IGF-1 in regulating AT receptors
----------------------------------------
The majority of studies on AT1 are related to cardiovascular disease, for which AT1 receptor antagonists were generated as treatment. Regarding AT2, although there has been increased research and interest in its role, this area appears little explored. That which has been learnt so far about the interplay and regulation of these receptors lends itself to a potentially useful model for the management of inflammation:
The expression of AT1 and AT2 receptors on fibroblasts present in cardiac fibrosis is investigated \[[@B79]\]. These types of fibroblast are noted for their expression of AT1 and AT2 receptors. The presence of IL-1b, TNF-α and lipopolysaccharides, through induction of NO and cGMP, all serve to down-regulate AT2 with no effect on AT1 leading to a quicker progression of fibrosis. Interestingly, the continuance in the presence of pro-inflammatory signals serves to delay expression of AT2. This is confirmed in a separate study of AT2 expression in proliferating cells. TGF-β1 and bFGF are shown as powerful inhibitors of AT2 expression, whilst IGF-1 is shown to induce the expression of AT2 \[[@B80]\].
IGF-1 is principally produced by the liver from GH (Growth Hormone) and circulates in the blood (decreasing with age) and is important in the regulation of immunity and inflammation \[[@B81]\]: IGF-1 is also capable of being produced by fibroblasts and macrophages on induction by pro-inflammatory cytokines, including TNF-α and IL-1b. In addition to the induction of AT2, IGF-1 can be seen as responsible for mediating the actions of many active cells in the immune/inflammation response \[[@B81]\]. Of note is that TNF-α and IL-1b also affect the circulating expression of IGF-1 by feedback on the release of GH from the anterior pituitary.
The controlling role of AT receptors in inflammation and healing
----------------------------------------------------------------
Significant evidence has been shown that AT1 receptors are upregulated during disease and that AT2 receptor expression follows behind AT1 expression during injury and healing. Given the opposing roles of AT1 and AT2 it can thus be postulated that the interplay of these receptors plays a significant part in judging the current local status of appropriate versus inappropriate inflammation and in providing feedback to the rest of the body. Indeed it is anticipated that prolonged expression of AT1 combined with a lack of AT2 expression results in sustained chronic inflammation and fibrosis.
Overall, the role of the AT receptors in managing and monitoring the healing process is complex, with many positive and negative feedback mechanisms both within the site of inflammation/healing and with the rest of the systems in the body. An attempt to summarise these systemic signalling inter-relationships is given in figure [4](#F4){ref-type="fig"}. Note the glucocorticoid inhibition of AT1 pro-inflammatory activities via NFκB. This model, although hypothetical, provides an explanation of the mechanisms whereby ox-LDL and homocysteine exert their pro-inflammatory effects. Further supporting this model is evidence that a lack of IGF-1 presence contributes to degenerative arthritis \[[@B81]\], septic shock \[[@B81]\], cardiovascular diseases \[[@B82]\] and inflammation of the bowel \[[@B83]\]. The introduction of IGF-1 is also proposed for protection against Huntington\'s \[[@B84]\], Alzheimer\'s \[[@B85]\] and Parkinson disorders \[[@B86]\]. Upregulation of IGF-1 has been noted in patients with chronic heart failure who undertake a programme of stretching exercise, thus providing benefits against cardiac cachexia \[[@B87]\].
::: {#F4 .fig}
Figure 4
::: {.caption}
######
***Systems view of the AT receptor role.***This hypothetical model shows the role of the mutually antagonistic AT receptors in managing levels of inflammation. Extended expression of AT1 in the absence of sufficient expression of AT2 may lead to a failure of inflammation resolution, sustained chronic inflammation and fibrosis. This model further serves to explain the pro-inflammatory role of hypoxia and oxidative stress and their risk factors in disease. Likewise the anti-inflammatory role of IGF-1 is supported and the risk factor of decreasing circulation of IGF-1 as a result of ageing. \'External Systems\' represents non-local feedback i.e. the rest of the body including hypothalamus, pituitary, thyroid, adrenal glands, liver and pancreas.
:::
:::
Conclusions
===========
The invasiveness and immunosuppression of many cancers appears dependent on inflammation and the upregulation of AT1. Two mechanisms for upregulation of AT1 are discussed: 1) evolutionary changes to take advantage of this pro-inflammatory control mechanism, 2) AT1 expression induced by an alternating environment of hypoxia and oxidative stress. Immunosuppression as a common protection mechanism of solid tumours against immune responses has been verified from current literature and experimental procedures, as has the implication of cytokines and chemokines in tumour growth and metastasis. Given the involvement of AT1 in the immunosuppression and inflammatory processes, as well as in the expression of the pro-inflammatory cytokines and chemokines, it becomes evident that the AT1 receptor is essential for tumour protection and progression. A combination therapy consisting of AT1 receptor antagonists, NSAID for further control of the inflammation and immune therapy in the form of tumour vaccines should provide a novel and successful treatment for solid tumours.
In the renin-angiotensin system, the angiotensin II receptors AT1 and AT2 seem to have opposing functions. The actions of AT1 being principally pro-inflammatory whilst AT2 provides protection against hypoxia, draws inflammatory action to a close and promotes healing. The various direct and indirect mechanisms for feedback between the receptors, their induced products and the external hormonal system in the control of inflammation and healing are summarised in a highly simplified model which none the less can be used to explain how many key promoters and inhibitors of disease exert their effects.
From a review of the current disease literature, it has been demonstrated that the role of AT1 and AT2 in inflammation is not limited to cancer-associated inflammation, but is generally consistent and system wide. Potential therapy by manipulation of these receptors, although at an early stage, has been demonstrated for some of these diseases and it is proposed that this approach will provide an effective basis for the treatment of autoimmune, inflammatory and neurodegenerative disorders using existing drugs. AT1 receptor blockade should, in addition, provide a treatment to alleviate the damage caused by bacterial and viral infections, where their destructive action is through chronic inflammation. Given the importance of the immune suppressant effect of inflammation in cancer, it is anticipated that AT1 blockade should also serve to elicit a more effective immune response to other invaders that seek to corrupt the wound recovery process.
Manipulation of the AT1 and AT2 receptors has profound and exciting implications in the control of disease.
List of abbreviations used
==========================
TGF-β Transforming Growth Factor Beta
AT1 Angiotensin II Type 1 receptor
AT2 Angiotensin II Type 2 receptor
IGF-1 Insulin-like Growth Factor 1
LOX-1 Lectin-like Oxidized Low-Density Lipoprotein Receptor 1
HIF-1a Hypoxia Induced Factor 1 Alpha
HMG CoA 3-Hydroxy-3-Methyl-Glutaryl Coenzyme A
bFGF basic Fibroblast Growth Factor
ROS Reactive Oxygen Species (most notably O~2~^-.^)
Competing interests
===================
Gary R Smith is a founding director of Perses Biosystems Ltd. The goals of the company are to drive laboratory and clinical research into the role of angiotensin receptors in disease management. Although we envisage these activities to be humanitarian (non-profit making) in nature, our long-term ambition is to identify additional drug targets and agents that could work in combination with ACE inhibitors and AT1 blockers to treat most diseases.
Sotiris Missailidis is a Lecturer at the Chemistry Department of The Open University, with research focus on cancer and had been the academic supervisor of Gary R Smith. There are no conflicting interests or financial implications related to the publication of this review article.
Authors\' contributions
=======================
Gary R Smith performed the literature review and proposed the hypothesis that cancer utilises the Angiotensin system to trigger chronic inflammation as a means of spreading and avoiding the immune system. In addition to providing significant editorial contributions and literature related comments, Sotiris Missailidis prompted Gary R Smith to undertake additional research with led to clarification of the role of hypoxia and oxidative stress in governing AT receptor expression. This understanding led Gary R Smith to propose the hypothesis that inflammation through the AT receptors is the cause of many of the diseases that affect mankind, including infectious diseases, which utilise inflammation to disrupt the immune system.
Acknowledgements
================
Many thanks to Jim Iley of the Open University not only for S807 Molecules in Medicine but also for suggesting the title of this paper.
|
PubMed Central
|
2024-06-05T03:55:55.073010
|
2004-9-30
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074345/",
"journal": "J Inflamm (Lond). 2004 Sep 30; 1:3",
"authors": [
{
"first": "Gary Robert",
"last": "Smith"
},
{
"first": "Sotiris",
"last": "Missailidis"
}
]
}
|
PMC1074346
|
Background
==========
RNA interference (RNAi) is an evolutionary conserved sequence-specific RNA silencing mechanism found as an anti-viral response in invertebrates, plants and mammalian cells \[[@B1]\]. Although the mechanism of silencing is not completely understood, the basic premise of RNAi rests on the ability of double stranded RNA (dsRNA) to specifically degrade homologous messenger RNA (mRNA). The RNAi pathway is triggered in mammalian cells by the presence of dsRNA or in the presence of short 19--22nt dsRNA fragments termed small interfering RNA molecules (siRNA). siRNA molecules activate an RNA-induced silencing complex (RISC) that unwinds the siRNA duplex \[[@B2]\]. The specificity of locus degradation is guided by the antisense strand of the unwound siRNA, followed by sense strand siRNA binding to the complementary mRNA site for cleavage by RISC. The cleavage of the sense strand siRNA and target mRNA results in the self-amplifying production of new siRNA intermediaries that continue mRNA target degradation in an ATP dependent manner \[[@B3],[@B4]\]. This phenomenon means that low doses of siRNA can be more effective than antisense therapy. Furthermore, this approach is preferable to gene and antisense based therapies, in that siRNA is non-heritable and does not require adenoviral vectors, which limit the effectiveness and acceptability for use in children.
RNAi can be exploited as a tool to provide novel insights into gene function and structure. The capacity to efficiently deliver siRNA to modulate gene expression *in vivo*may provide new therapeutic approaches to currently intractable diseases. Like other new genetic technologies, siRNA gene suppression faces several methodological limitations *in vivo.*Foremost among these are the efficient delivery of siRNA to target cells \[[@B5],[@B6]\], non-specific effects of putative control duplexes \[[@B7]-[@B9]\] and the potential therapeutic problems of viral expression vectors \[[@B10]\]. One approach to overcoming these obstacles is to deliver non-heritable siRNA duplexes in a model system and monitor the influence upon experimentally induced inflammation. This approach would provide a method that allows the rapid screening of what have been termed \"druggable\" targets \[[@B11]\].
Interleukin-12 (IL-12p70) is a cytokine with a well-characterized pro-inflammatory function \[[@B12]\] that has been suggested as a target for therapeutic intervention \[[@B13]-[@B15]\]. Bioactive IL-12p70 is a heterodimer formed by a heavy chain (p40) and a light chain subunit (p35), encoded by two separate genes whose expression is independently regulated at the transcriptional level \[[@B16]\]. The p35 sub-unit is constitutively expressed at low levels in most cell types but is up regulated during cell activation. In contrast, the IL-12p40 gene is under tight transcriptional control only expressed in macrophages or other APC following activation by microbial products \[[@B17]\]. Production of IL-12p70 is enhanced by IFN-γ via the IFN consensus sequence binding protein \[[@B18]\] but reduced by IL-10 \[[@B19]\].
IL-12p70 has pleiotropic effects on target cells but the major role is as a pro-inflammatory cytokine in cell mediated immunity against microbial insult. In particular IL-12p70 acts upon T and NK cells to increases cytokine production, proliferation, and cytotoxicity, functions that become evident several hours after exposure to infections agents \[[@B19]\]. The IFN-γ subsequently produced, potentiates antigen presentation functions important in clearing infectious agents. These functions include increased co-stimulatory molecule expression, phagocytosis, and production of reactive oxygen and nitrogen intermediates \[[@B19],[@B20]\]. However, IL-12p70 is not always protective or beneficial, indeed a variety of pathological conditions, including sepsis, are associated with IL-12 driven pathology \[[@B21],[@B22]\]. In addition to the well-characterized role of IL-12p70, it is now known that the IL-12p40 subunit is also biologically active. This subunit may act to antagonize the heterodimer function \[[@B23]\], or may have a broader direct role, less dependent on IL-12p70 \[[@B24],[@B25]\].
In order to explore the therapeutic feasibility of RNA interference, we used siRNA to specifically ablate IL-12p40 expression *in vitro*and *in vivo*. This approach extends the power of RNA interference to gene expression studies in live animals without the use of genetic engineering, plasmid DNA reporter systems \[[@B2],[@B26]\] retroviral \[[@B27],[@B28]\] or lentiviral siRNA expression vectors \[[@B29]\] and opens the way for exploring the use of siRNA in humans to treat disease. Our results provide a description of siRNA mediated suppression of an endogenous immune gene *in vivo*and describe a novel therapeutic and research approach for gene specific inhibition of an important cellular and immunological response.
Materials and Methods
=====================
Mice & Cell Lines
-----------------
Female BALB/c mice (Harlan Limited, Bicester, UK) and IL-12p40 gene-disrupted mice (IL12p40^-/-^) (Jackson Laboratories, Bar Harbor, Maine) were maintained under the guidelines of the Irish Department of Health and the local bioethics committee. All mice were 12--14 weeks old at the initiation of experiments and sacrificed on completion. The murine macrophage cell line (J774A.1) was used to investigate silencing of IL-12p40 cytokine gene expression.
Preparation of siRNA
--------------------
siRNA oligonucleotides with the following sense and antisense sequences were designed from the GenBank repository: accession number; NM\_008352, *Mus musculus*interleukin 12b (IL12b), mRNA. IL-12p40 siRNA 5\'-C CUC ACC UGU GAC ACG CCU dTdT-3\' (sense) and 3\'-dTdT G GAG UGG ACA CUG UGC GGA-5\' (antisense); Mutant siRNA 5\'-C CUC ACC UUC GAC ACG CCU dTdT-3\' (sense) and 3\'-dTdTG GAG UGG AAG CUG UGC GGA-5\' (antisense); GFPsiRNA 5\'-GGC UAC GUC CAG GAG CGC ACC dTdT-3\' (sense) and 3\'-dTdT CCG AUG CAG GUC CUC GCG UGG-5\' (antisense). The antisense of the IL-12p40 siRNA duplex (As.RNA) was also used as a control for *in vivo*experiments. Each complementary RNA strand was deprotected according to manufacturer\'s instructions. For the production of the IL-12p40 siRNA duplex, sense and antisense siRNA strands were mixed in equimolar ratios and treated by heating to 95°C for 1 min followed by annealing at 37°C for 1 h and allowed to cool slowly overnight to room temperature. All siRNA oligonucleotides were synthesized commercially (Dharmacon, Lafayette, CO) using 2\'ACE protection chemistry.
*In vitro*siRNA interference
----------------------------
Semi-confluent J774A.1 cells were cultured at 1 × 10^5^cells/ml in antibiotic free, 8% (v/v) endotoxin-low fetal-calf serum RPMI (Gibco-Invitrogen, Paisley, UK) containing L-glutamine (Sigma, Poole, UK) 12--16 h before transfection. For siRNA transfections 3 μl of a 20 μM siRNA duplex (target or control) solution was mixed with 47 μl of Opti-mem (Gibco-Invitrogen). In a second tube 3 μl of oligofectamine (Gibco-Invitrogen) was mixed with 12 μl of Opti-mem and incubated at room temperature for 15 min. Solutions were combined for 40 min and brought to a final volume of 100 μl. The expression of IL-12p40 mRNA and IL-12p70 protein was induced by the addition of 1 μg/ml *E. coli*LPS Serotype 0111:B4 (Sigma) and 10 ng/ml rIFN-γ (Pharmingen, San Diego, CA.), for the last 12 h of each culture post siRNA transfection.
RNA isolation and semi-quantitative RT-PCR (sqRT-PCR)
-----------------------------------------------------
Total cellular RNA was isolated from J774A.1 cells from *in vitro*experiments with TRIZOL Reagent (Gibco-Invitrogen) following the manufacturer\'s protocol and quantified by spectrophotometry. RNA was reverse transcribed, and 100 ng of the complementary DNA product amplified by PCR as previously described \[[@B51]\] using 60 ng of gene specific upstream and downstream primers. Murine β-actin product was used to normalize RNA samples. PCR conditions included a pre-incubation at 95°C for 5 min followed by 35 amplification cycles (95°C, 1 min; 1 min at annealing temperature; 2 min at 72°C, and a final 10 min at 72°C). Upstream and downstream primers for IL-12p40 were specifically designed to flank the IL-12p40 siRNA target region; sense, 5\'-AAACAGTGAACCTCACCTGTGACAC-3\' ; antisense, 5\'-TTCATCAGCAAGTTCTTGGGCG-3\'. PCR products were visualized by UV illuminated agarose gel electrophoresis.
*In vivo*siRNA interference
---------------------------
Control mice (BALB/c & IL12p40^-/-^) received 200 μl Opti-mem intra-peritoneal (i.p.) containing oligofectamine alone. In addition LPS positive control mice received 1 μg *E. coli*LPS. For each experimental administration, 10 μl siRNA duplexes (IL-12p40 or controls at equimolar concentration) were premixed with 40 μl of Opti-mem. Separately, 6 μl of oligofectamine was mixed with 24 μl of Opti-mem and incubated at room temperature for 15 min. These solutions were mixed at room temperature for 40 min. For co-injection experiments, these were combined with LPS (1 μg/mouse) and formulated as above. For therapeutic silencing, mice received 1 μg LPS, in the absence of siRNA duplexes, 1 h prior to administration of siRNA (IL-12p40 or controls) as above. At various time points, blood serum, peritoneal cells or lavage fluid were sampled for further analysis.
Peritoneal Lavage & Serum preparation
-------------------------------------
Peritoneal cells were harvested by washing the peritoneal cavity with 1 ml of sterile PBS. This was centrifuged for 5 min at 400 g, lavage supernatant was removed for analysis and cells analysed by flow cytometry. Serum was prepared by cardiac puncture. Sera and lavage supernatants were assayed without delay or storage.
Flow Cytometry
--------------
Phenotypic analysis of siRNA-transfected cells was performed using a FACScalibur™ with associated Cellquest™ software (Becton Dickinson, San Jose, CA). Forward and side scatter were measured from peritoneal lavage preparations at 12, 24 and 48 h in response to simultaneous delivery of IL-12p40 siRNA and LPS, and at 24 h for those mice receiving therapeutic IL-12p40 siRNA post LPS administration. Cell surface marker analysis of CD11b, CD14, CD40, CD80, CD86, F4/80 and MHC class II by J774A.1 cells was performed as previously described \[[@B52]\], control samples included cells incubated with isotype matched, directly conjugated, control antibodies as appropriate.
Analysis of cytokine production
-------------------------------
Cytokine production from *in vitro*experiments was assayed using commercially available immunoassays for mouse IL-12p70, IFN-γ, IFN-β, IL-10, and IL-4 (Pharmingen). Mouse IL-12p40 in blood serum and peritoneal lavage fluid was assayed using murine IL-12p40 ELISA (R&D systems, Abingdon, UK) according to the manufacturer\'s instructions.
Statistical analysis
--------------------
One-way ANOVA was used to determine significance of cytokine production between groups; post test analyses were not performed. The student t-Test was used to determine the significance of different fluorescent intensities obtained by flow cytometry.
Results
=======
IL-12 p40 siRNA knocks down IL-12 expression *in vitro*
-------------------------------------------------------
To investigate silencing of cytokine gene expression *in vitro,*the murine macrophage-like cell line J774A.1 was transiently transfected with siRNA targeting IL-12p40 for the time points shown in Fig. [1](#F1){ref-type="fig"} (24, 48, 72 h). These cells were stimulated for the final 12 h of each experiment, with LPS and IFN-γ (LPS/IFN-γ), a protocol that induces IL-12p70 \[[@B30]\]. Transfection with IL-12p40 siRNA resulted in a significant suppression of p40 mRNA and a consequent loss of detectable IL-12p70 in cell culture supernatant (Fig. [1A,1B,1C](#F1){ref-type="fig"}). A reduction in IL-12p40 mRNA was observed at 24 h, but silencing was more pronounced at 48 h. Transfection for 72 h with IL-12p40 siRNA was inferior to either 24 or 48 h, as IL-12p40 mRNA expression and IL-12p70 protein synthesis began to recover by this time (Fig. [1A](#F1){ref-type="fig"} &[1C](#F1){ref-type="fig"}). Thus siRNA silencing was transient in this system. Control siRNA transfections included siRNA for IL-12p40 without transfection agent (naked siRNA), siRNA for IL-12p40 where the 10^th^and 11^th^bases were reversed (mutant siRNA), and siRNA targeting GFP, a protein that does not naturally occur in J774A.1 cells. These control siRNAs did not induce IL-12p40 mRNA expression (Fig. [1B](#F1){ref-type="fig"}). Our results show sequence-specific siRNA mediated inhibition of IL-12p40 mRNA synthesis *in vitro*at 48 h post siRNA incubation (Fig. [1A](#F1){ref-type="fig"}). ELISA confirmed the siRNA mediated silencing of IL-12p70 protein expression (Fig. [1C](#F1){ref-type="fig"}), reflecting the significant inhibition of IL-12p40 mRNA synthesis (p \< 0.001, compared to LPS/IFN-γ group). Supernatants from unstimulated cells, or cells incubated with control siRNAs, showed no IL-12p70 protein production. Suppression of IL-12p70 was transient, with levels recovering at the remaining time points. mRNA expression profiling for the inflammatory cytokines IFN-β, IL-12p35, IL-23p19, IL-6, IL-10 and IFN-γ in IL-12p40 or control silenced cells, showed no non-specific siRNA silencing at the doses employed (Table [1](#T1){ref-type="table"}). Control siRNA delivered by the same protocol did not induce mRNA for IFN-β, IL-12p35, IL-23p19, IL-6, IL-10 and IFN-γ. Likewise, cells transfected with IL-12p40 siRNA showed no modulation of the protein levels of IL-4, IL-5, IL-6, IL-10, and TNF-α (results not shown). One cytokine did not follow this pattern. Although IL-12p40 siRNA transfection of stimulated macrophages did not result in a detectable reduction of IFN-γ mRNA (Table [1](#T1){ref-type="table"} and Fig. [2A](#F2){ref-type="fig"}), a reduction of detectable IFN-γ protein was observed (Fig. [2B](#F2){ref-type="fig"}). This discrepancy between IFN-γ mRNA and protein detection may reflect the role of IL-12p40 in post-transcriptional regulation of IFN-γ secretion \[[@B31]\] and is consistent with the timing of IFN-γ protein synthesis and secretion previously observed in LPS challenged IL-12p40^-/-^mice *in vivo*\[[@B32]\].
::: {#F1 .fig}
Figure 1
::: {.caption}
######
siRNA interference of IL-12 transcription and translation *in vitro*. J744A.1 cells were transfected with IL-12p40 siRNA for different periods (24--72 h). For the last 12 h of culture, cells were stimulated with LPS/IFN-γ or with PBS (-LPS, hereafter termed unstimulated). Expression of IL-12p40 (A) was determined by sqRT-PCR. Samples were normalized for β-actin expression (lower panel A). Control siRNA transfections included naked siRNA for IL-12p40, mutant siRNA for IL-12p40 and GFP (B). IL-12p70 protein expression was determined by ELISA (C). Data are representative of at least four independent experiments; IL-12p70 protein concentration is expressed as the mean (+/-SEM) from triplicate cultures (n = 3) on each occasion.
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
IL-12p40 siRNA specifically silences mRNA for IL-12p40 and not other cytokines.
:::
Target Silencing by treatment^a^
------------ --------------------------- ----------- ----------
LPS/IFN-γ: \- \+ \+
SiRNA: IL-12p40 Mut.siRNA IL-12p40
β-Actin \- \- \-
IL-12p40 \- \- \+
IL-12p35 \- \- \-
IL-23p19 \- \- \-
IFN-γ \- \- \-
IFN-β \- \- \-
TNF-α \- \- \-
IL-4 \- \- \-
IL-5 \- \- \-
IL-6 \- \- \-
IL-10 \- \- \-
^a^J774 cells were incubated with or without LPS/IFN-γ and specific or control siRNA as described in the materials and methods section. mRNA for different targets were detected by sqRT-PCR. In this table silencing (+) is defined as the loss of a visible band from stimulated cultures; the - symbol indicates either no loss of a visible band from stimulated (siRNA + LPS/IFN-γ) cultures, or no visible alteration (induction or loss) in unstimulated (siRNA -- LPS/IFN-γ) cultures. All results represent at least two experiments performed in triplicate.
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Silencing IL-12p40 influences IFN-γ protein expression. The influence of silencing IL-12p40 on the expression of IFN-γ was determined using the conditions described in Fig. 1, by sqRT-PCR for IFN-γ mRNA (A) or by ELISA for IFN-γ protein present in culture supernatant (B). Data are representative of at least four independent experiments; IFN-γ protein concentration is expressed as the mean (+/-SEM) from triplicate cultures (n = 3) on each occasion. Levels of IFN-γ are significantly reduced in the presence of IL-12p40 siRNA for 24, 48, and 72 h (p \< 0.0001) when compared to stimulated non-silenced cultures.
:::
:::
Silencing IL-12p40 reduces LPS/IFN-γ driven macrophage activation *in vitro*
----------------------------------------------------------------------------
To determine whether silencing IL-12p40 had broader effects on macrophages, the expression of the activation/co-stimulatory markers CD40, CD80, CD86, and MHC class II was examined following simultaneous exposure of J774 cells to LPS/IFN-γ and either control or IL-12p40-specific siRNA. Expression of CD14, a component of the LPS recognition machinery was also examined. LPS/IFN-γ stimulation alone (24 h) resulted in increased CD40, CD86 and MHC class II expression (Table [2](#T2){ref-type="table"}), but had no effect on CD80 or CD14 as expected. IL-12p40-specific siRNA did not activate macrophages in the absence of LPS/IFN-γ (Table [2](#T2){ref-type="table"}). In the presence of LPS/IFN-γ, siRNA targeting IL-12p40 prevented increased expression of CD40, and CD86, suggesting that silencing IL-12 interfered with macrophage activation. The expression of CD80, CD14 and MHC class II were not affected (Table [2](#T2){ref-type="table"}). In contrast, Mut.siRNA did not prevent CD86 upregulation when cells were stimulated with LPS/IFN-γ but rather resulted in increased expression, suggesting that this sequence may contribute to macrophage activation not seen with IL-12p40 specific siRNA. The expression of the macrophage phenotypic markers CD11b and F4/80 were unchanged in all experiments, no significant difference was seen in levels of apoptosis between groups (data not shown).
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
LPS/IFN-γ driven macrophage expression of CD40 and CD86 is modulated by IL-12p40 siRNA.
:::
Treatment Mean Fluorescent Intensity (+/-SEM)
----------- ------------------------------------- ----------- --------- ---------- --------- --------
\- \- 22 (5) 9 (4) 34 (8) 44 (11) 19 (8)
\- IL-12p40 16 (9) 14 (1) 29 (2) 49 (4) 18 (8)
\+ \- 290 (27) 41(5) 102 (21) 86 (11) 46 (5)
\+ IL-12p40 94 (17)\* 7 (1)\* 66 (13) 47 (9) 46 (5)
\+ Mut.siRNA 241 (18) 72 (6) 102 (4) 86 (16) 43 (5)
Data are the mean ± SEM to nearest whole number of MFI of 3 independent experiments, each performed in triplicate (n = 3), \* statistical significance (α = 0.05) compared to cells stimulated with LPS/IFN-γ in the absence of IL-12p40 siRNA.
:::
siRNA targeting IL-12p40 specifically reduces LPS driven inflammation *in vivo*
-------------------------------------------------------------------------------
We investigated the possibility of silencing IL-12 by RNA interference *in vivo*, using a well-established murine model of LPS driven peritoneal inflammation \[[@B33],[@B34]\]. The delivery of LPS i.p. resulted in increased activated phagocytic cells detectable at 12, 24 and 48 h by lavage, compared to controls (Fig. [3A](#F3){ref-type="fig"} groups I & II). This effect was greatly reduced in IL-12p40^-/-^mice. Simultaneous delivery of a control irrelevant siRNA (GFPsiRNA) or a mutant IL-12p40 siRNA (Mut.siRNA) duplex containing two mismatches to the IL-12p40 specific sequence had no influence on LPS driven inflammation. Likewise, a control siRNA that was the antisense of the functional duplex (As.siRNA) did not result in a significant reduction in the level of activated phagocytic cells. However, delivery of IL-12p40 siRNA dramatically reduced the levels of inflammation (Fig. [3A](#F3){ref-type="fig"}) at 12, 24 and 48 h. Delivery of encapsulated siRNA did not result in increased in cell death of peritoneal cells (apoptosis or necrosis) compared to controls at the time points selected (data not shown).
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Silencing of IL-12 by siRNA interferes with the inflammatory response *in vivo*. siRNA was delivered with LPS (A) or therapeutically 1 h post LPS stimulation (B). Mice received transfection reagents only (no siRNA) (group I), LPS alone (group II) or were co-injected with LPS and control siRNAs (group III-V, As.siRNA, GFPsiRNA, and Mut.siRNA respectively) or LPS and IL-12p40 specific siRNA (group VI). Inflammation was characterized by flow cytometry of peritoneal lavage at 12, 24 and 48 h. The typical inflammatory cell response in the peritoneal cavity is shown in the enclosed region. Control IL-12p40^-/-^mice showed the characteristic germ-line knockout response to LPS throughout the experiment (group VII). IL-12p40 siRNA was also delivered therapeutically (B) 1 h post LPS challenge, and inflammation measured at 24 h. Mice receiving control siRNAs (groups III-V) displayed a similar inflammatory response to mice receiving LPS insult alone (group II). Mice receiving IL-12p40 siRNA (group VI) displayed a reduced number of activated phagocytic cells at the same time point (enclosed region). Data are representative of at least three independent experiments (Groups I-VI) or two experiments (Group VII). In each experiment, n = at least 4 mice on each occasion. The absolute numbers of cells present in lavage fluid, represented by the enclosed region in A, are illustrated (C). Data in bar charts represent the mean number of cells (+/-SEM)/ml lavage fluid from four mice at the time points indicated.
:::
:::
Control wildtype and IL-12p40^-/-^mice showed no inflammatory response to siRNA transfection reagents alone (Fig. [3](#F3){ref-type="fig"}). LPS challenged wildtype mice, and mice co-challenged with control siRNAs displayed a typical inflammatory cell response in the peritoneal cavity with increased numbers of activated phagocytic cells, at 24 h (19.8%, 16.8 %, 22.01%) and 48 h (22.53%, 17.95% 16.64%) compared to control unchallenged mice (2.17% and 3.46%). Similar results were observed in mice co-challenged with LPS and As.siRNA at 24 and 48 h (18.9% and 23.64% respectively). However, mice co-administered LPS and specific IL-12p40 siRNA displayed reduced numbers of activated cells (6.3%), mirroring the reduced inflammatory response seen in LPS challenged IL-12p40^-/-^mice (7.3%) at 24 h. However, modulation of the inflammatory response in the LPS-IL-12p40 siRNA challenged mice was not permanent. An increase in activated phagocytic cells (11.10%) was seen at 48 h, although levels were still lower than the LPS challenged BALB/c mice (22.53%), (Fig. [3A](#F3){ref-type="fig"} &[3C](#F3){ref-type="fig"}). Thus, siRNA-mediated silencing of IL-12p40 mRNA in this model has a significant but non-permanent effect on the ability to mediate a response to LPS challenge *in vivo*.
siRNA targeting of the proinflammatory cytokine IL-12p40 can be used as a therapeutic intervention against inflammation driven by microbial products
----------------------------------------------------------------------------------------------------------------------------------------------------
In order to explore the potential use of siRNA in a more therapeutic context and based on the findings above, we delivered IL-12p40 siRNA by direct injection into the peritoneal cavity, 1 h post LPS challenge. Administration of IL-12p40 siRNA post LPS challenge (Fig. [3B](#F3){ref-type="fig"} I-VI) resulted in a decrease in the number of activated phagocytic cells, (4.22%) at 24 h, when compared to mice receiving LPS only (16.64%), control siRNAs (Mut.siRNA and GFPsiRNA) or As.siRNA, (25.03%, 17.07% and 12.63% respectively). These data demonstrate that IL-12p40 siRNA can be used therapeutically to specifically silence a cytokine-driven inflammatory response *in vivo*, if delivered at a suitable moment.
In parallel experiments, the local and systemic effects of siRNA mediated silencing were assessed. IL-12p40 siRNA co-delivered with LPS or administered post LPS insult, had both local and systemic anti-inflammatory effects (Fig. [4](#F4){ref-type="fig"}). Control BALB/c mice given siRNA transfection reagents alone, showed low levels of IL-12p40 protein expression in blood serum and peritoneal lavage samples (103 pg/ml and 75 pg/ml respectively). However, mice challenged with LPS, or co-challenged with LPS and control siRNAs (GFPsiRNA, Mut.siRNA)(Fig. [4](#F4){ref-type="fig"}) showed significant increases in IL-12p40 protein detected in both serum and lavage compared to control (p \< 0.05). Delivery of As.siRNA did result in reduced serum IL-12p40 protein but only when administered therapeutically (Fig. [4B](#F4){ref-type="fig"}). Strikingly, delivery of IL-12p40 siRNA delivered simultaneous to, or 1 h post LPS administration, resulted in a significant reduction in the levels of IL-12p40 protein detected in all serum and peritoneal lavage samples compared to LPS alone (p \< 0.05, in each case) (Fig. [4](#F4){ref-type="fig"}). Delivery of negative control siRNAs showed no such reduction. Our findings demonstrate that well designed sequence specific siRNA can provide a significant therapeutic effect and elicit local and systemic protection against inflammation.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
siRNA silencing IL-12p40 reduces local and systemic levels of IL-12p40 protein. LPS delivery to the peritoneal cavity with simultaneous (white bars), or therapeutic administration (black bars) of IL-12p40 specific siRNA reduced the appearance of IL-12p40 protein in serum (A) and peritoneal lavage (B) as measured by IL-12p40 specific ELISA. Serum and lavage were sampled at 6 h post LPS challenge. IL-12p40 observed in serum or lavage from LPS challenged mice or mice co-challenged with LPS and either GFPsiRNA or Mut.siRNA was significantly greater than control (p \< 0.05, in both cases). IL-12p40 siRNA when delivered simultaneously or therapeutically significantly reduced IL-12p40 protein production compared to LPS challenge alone (p \< 0.05, in each case). Data are representative of three independent experiments where n = 4 on each occasion; IL-12p40 protein concentration is expressed as the mean (+/-SEM) from triplicate samples.
:::
:::
Discussion
==========
The ability to efficiently deliver small interfering RNA to modulate gene expression *in vivo*may provide new therapeutic approaches to currently intractable diseases. We initially demonstrate that *in vitro*IL-12p40 siRNA specifically silenced its mRNA homologue leading to transient silencing of IL-12p40 protein and consequent knockdown of IL-12p70 expression. This approach did not target other proinflammatory cytokines (IL-6, IL-23, IL-10, TNF-α), for RNA-induced gene silencing, nor did control siRNA induce these cytokines or type I interferon at the concentrations employed. Furthermore, we demonstrate that this approach can be extended *in vivo*by showing that silencing of IL-12p40 results in the non-permanent suppression of IL-12 in a murine model of peritoneal inflammation. Such silencing is evident in the reduced levels of IL-12 detectable locally in peritoneal lavage and systemically in blood serum. Finally, we show that siRNA can be used therapeutically after the initiation of an inflammatory response to knockdown IL-12 expression and to reduce the observed inflammatory infiltrate seen in this model.
IL-12 is a key factor in the early inflammatory response and in the subsequent development of type 1 responses \[[@B35]\]. A variety of signals can stimulate macrophages resulting in increased surface expression of CD40 and the B7 family member CD86 as well as activation of the cell\'s antimicrobial machinery \[[@B20],[@B36]\]. Here, we show that silencing of IL-12p40 interferes with endotoxin mediated activation as measured by CD40 and CD86 expression, similar to that seen in IL-12p40^-/-^mice in which macrophages adopt the so-called M2 profile \[[@B37]\]. Taken together, these data support the hypothesis that IL-12p40 has a central role in driving macrophage polarization, and regulating the intrinsic ability to respond to immunological insult \[[@B30],[@B37]\]. The polarization of CD4^+^T-cell cytokine production towards type 1 or type 2 responses following immunological insult is controlled by a number of factors, including the nature of the immunogen, route of immunization, the APC and the regulatory cytokine milieu at the site of T-cell stimulation \[[@B38],[@B39]\]. IL-12 induces the secretion of IFN-γ by NK and CD4^+^T-cells, promoting the differentiation and development of Th1 cells from Th0 precursors \[[@B40],[@B41]\]. Th1 cells play an important role in the resolution of infections with intracellular organisms, IL-12 influences the course of bacterial, viral, and parasitic infections by altering the balance of Th1 and Th2 cells in favour of IFN-γ production \[[@B42],[@B43]\]. The ability to transiently silence IL-12 may therefore be a useful research tool to dissect the development of polarized immune responses in a variety of infectious diseases.
Although IL-12p40 as a component of IL-12p70 is known to have a direct role in macrophage activation \[[@B36],[@B44]\], it has recently become clear that IL-12p40 has a role independent of the heterodimer \[[@B24]\]. IL-12p40 acts as an antagonist of IL-12p70 function \[[@B23]\], but also has direct effector function \[[@B25],[@B45]\]. In particular IL-12p40 plays a role in macrophage, but not NK or T-cell recruitment and chemotaxis to inflammatory sites \[[@B25],[@B45]\]. The silencing of IL-12p40, and subsequent reduced inflammation seen *in vivo*during the present study supports a broader role for IL-12p40 in macrophage recruitment to sites of inflammation induced by microbial stimuli. Silencing IL-12p40 *in vitro*did not result in non-specific silencing of IL-12p35, IL-23p19, IL-10, TNF-α, IL-6 or IFN-γ mRNA. However, silencing of IL-12p40 by siRNA did result in a reduction of IFN-γ production detected by ELISA. Regulation of IFN-γ production by macrophages has not been extensively studied, however it has been shown that in some cell types IL-12 promotes nuclear localization of IFN-γ mRNA and exerts post-transcriptional control over IFN-γ secretion \[[@B31]\]. Our observations are consistent with this finding and suggest that IL-12 may exercise post-transcriptional control on IFN-γ protein production in macrophages.
Non-specific immune stimulation is an undesirable side effect of antisense oligonucleotides and vector based expression approaches *in vivo*\[[@B8],[@B46]\]. Recently Sledz et al., (2003) have found that under some conditions transfection of siRNA results in IFN-mediated activation of the JAK-STAT pathway and global upregulation of IFN-stimulated genes. To demonstrate specificity of gene suppression, and non-activation of the IFN immune response in our study, three siRNA duplexes were designed according to Semizarov et al. \[[@B46]\]. We employed three different control siRNAs; a mutant IL-12p40 siRNA (Mut.siRNA) with two point mutations at the 10^th^and 11^th^nucleotide of the IL-12p40 siRNA duplex, an irrelevant siRNA duplex (GFPsiRNA) \[[@B26]\] and also the antisense of the siRNA duplex (As.siRNA). At the concentrations employed in this study we saw no non-specific silencing from control siRNA and notably no induction of IFN-β.
The ability to silence an inflammatory mediator *in vivo*has implications for the application of siRNA approaches in inflammatory diseases such as sepsis, acute respiratory distress syndrome, and T-cell mediated autoimmune diseases where the transient suppression of inflammatory gene expression would be likely to prove beneficial \[[@B47]\]. We demonstrate that delivery of liposome-encapsulated siRNA targeting IL-12p40 to the murine peritoneal cavity can moderate an inflammatory stimulus *in vivo*. To date there have been very few demonstrations of siRNA efficacy *in vivo*. It has been shown that intravenous injection of Fas siRNA specifically reduced Fas mRNA levels and expression of Fas protein in mouse hepatocytes \[[@B6]\]. More recently Sorensen et al, reported siRNA mediated TNF-α protein ablation *in vivo*\[[@B48]\]. Using a similar delivery technique, our study greatly expands the use of siRNA as a pharmaceutical tool for drug discovery by demonstrating that i.p. delivery inhibits endogenous gene expression affecting detectable cytokine levels both locally and systemically, resulting in altered cell activation and maturation during inflammatory insult. This supports the findings of Song et al, who showed that treatment with Fas siRNA 2 days prior to mitogen challenge abrogated hepatocyte necrosis and inflammatory infiltration resulting in reduced serum concentrations of transaminases \[[@B6]\]. We investigated whether IL-12p40 specific siRNA could be used therapeutically after endotoxin challenge to reduce both the local and systemic inflammatory response. Our results show delivery of IL-12p40 siRNA provides local and systemic anti-inflammatory effects on IL-12p40 protein levels. Thus, the delivery of siRNA can be used to elicit specific, non-permanent, inhibition of endogenous protein expression after exposure to inflammatory insult.
The simplicity of this approach provides a rapid means to elucidate novel druggable targets in formerly intractable inflammatory and immune-mediated diseases. Importantly, the transient and specific silencing of protein products may prove advantageous. Permanent gene silencing through the use of plasmid DNA reporter systems \[[@B2],[@B26]\] retroviral \[[@B27]\] or lentiviral expression vectors \[[@B29]\] have a number of disadvantages. Permanent silencing of immune mediators may leave the host susceptible to subsequent infection, or set up potentially pathological hyper-responsiveness. Furthermore, there are unresolved safety issues relating to gene integration and host cell transformation that render these approaches less acceptable for use in children. The use of non-heritable siRNA delivered by liposomes circumvents these problems and opens the way for exploring the use of siRNA in humans to treat disease.
Our findings show that synthetic siRNA molecules delivered by intra-peritoneal injection are not affected by serum derived exonuclease activities \[[@B49]\] and do not require structural variations or stabilizing modifications \[[@B50]\] in order to have an efficient local and systemic effect on the target gene. This approach is effective, non-permanent, technically simple, and avoids some of the side effects of other delivery and gene silencing approaches.
Conclusions
===========
We have demonstrated *in vitro*that IL-12p40 siRNA specifically silenced its mRNA homologue leading to transient silencing of IL-12p40 and consequent knockdown of IL-12p70 expression. At the doses employed, this was specific and did not result in detectable induction of type I interferons. Silencing of IL-12p40 by siRNA did result in a reduction of IFN-γ production detected by ELISA. These findings were extended *in vivo*. Silencing of IL-12p40 results in the non-permanent suppression of IL-12 in a murine model of peritoneal inflammation. We show that siRNA can be used therapeutically after the initiation of an inflammatory response to silence IL-12 expression and observe a reduction in peritoneal inflammatory infiltrate.
Competing Interests
===================
The author(s) declare that they have no competing interests.
Author\'s Contributions
=======================
BPM directed the study and experimental design. DGC and MAF designed siRNA molecules and primers, and executed the experimental procedures. SMT participated in the experimental design. All authors read and approved the final manuscript.
Acknowledgements
================
David Casey is funded by Enterprise Ireland, Bernard Mahon is a Wellcome Trust/Health Research Board \"New Blood\" fellow (GR054236).
|
PubMed Central
|
2024-06-05T03:55:55.075919
|
2004-10-1
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074346/",
"journal": "J Inflamm (Lond). 2004 Oct 1; 1:4",
"authors": [
{
"first": "Marion A",
"last": "Flynn"
},
{
"first": "David G",
"last": "Casey"
},
{
"first": "Stephen M",
"last": "Todryk"
},
{
"first": "Bernard P",
"last": "Mahon"
}
]
}
|
PMC1074347
|
Background
==========
Aromatherapy is a folk medicine using essential oils. Recently the clinical use of essential oils has expanded worldwide to include therapy against various kinds of inflammatory diseases, such as allergy, rheumatism and arthritis. These activities have mainly been recognized through clinical experience, especially through skin application via massage and ointment, but there have been relatively little scientific study on their biological actions.
Several investigators have suggested that tea tree \[[@B1],[@B2]\] and lavender oils \[[@B3]\] suppressed allergic symptoms through the suppression of histamine release \[[@B4],[@B5]\] and cytokine production \[[@B6]\]*in vitro*and *in vivo*. Moreover, in human, skin application of tea tree oils was reported to suppress the edema induced by intradermal injection of histamine \[[@B7]\]. However, very few reports \[[@B8],[@B9]\] are available on the inhibitory effect of essential oils on the accumulation of inflammatory cells, which is a histological character recognized in chronic inflammatory diseases.
In earlier papers, we reported that the essential oils: lemongrass, geranium and spearmint suppressed the adherence response of human neutrophils *in vitro*\[[@B10]\], and that the intraperitoneal administration of geranium oil to mice lowered neutrophil recruitment into the peritoneal cavity induced by a chemotactic agent, casein injection *in vivo*\[[@B9]\]. Since essential oils are frequently applied through skin in aromatherapy for inflammatory symptoms, we believed that anti-inflammatory effects of the cutaneous application of these oils should be investigated in animal experiments to obtain practically valuable knowledge in this field. In the present study, we investigated the *in vivo*effects of cutaneously applied essential oils, especially geranium oil, to mice on inflammatory reactions including the accumulation of neutrophils in skin, which was induced by curdlan, a linear (1→3)-β-D-glucan known as a stimulating substance common in fungi.
Methods
=======
Essential oils
--------------
The essential oils used are listed in Table [1](#T1){ref-type="table"} with their manufacturer and main constituents. Table [1](#T1){ref-type="table"} also indicates literature references that show clinical use related to inflammatory symptoms \[[@B11]-[@B13]\]. Essential oils were purchased from Hyperplants, Ltd. (Tokyo, Japan). The constitution of the geranium oil was determined by gas chromatography in this laboratory \[[@B14]\] using a GC apparatus (Model 353B, GL Sciences, Tokyo) equipped with a DB-5 column (0.5 mm × 30 m; J&W Scientific, Folsom, LA, USA), and was shown to contain about 24 % β-citronellol, 10 % citronellyl formate and 7 % geraniol and others.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Essential oils, main constituents, their sources and manufacturer
:::
Essential oil Parent plant Main constituents Manufacturer of the oil References for clinical use
------------------ -------------------------- ------------------------- ------------------------- -----------------------------
Eucalyptus *Eucalyptus glogulus* 1,8 -- cineole Sanoflore (France) 11
Tea tree *Melaleuca alternifolia* terpinen-4-ol Sanoflore (France) 12,13
True Lavender *Lavandula angustifolia* linalool Sanoflore (France) 11--13
Geranium Bourbon *Perargonium asperum* geraniol, β-citronellol Sanoflore (France) 11--13
Juniper *Juniperus communis* α-pinene Sanoflore (France) 11--13
:::
For intraperitoneal injection, these essential oils were diluted to 2.5, 5 % solution by 2.5 % dimethyl sulfoxide (DMSO) in saline and 50 μl of Tween 20 was added to 4 ml of the essential oil solution. For cutaneous application, each essential oil was diluted to 5, 10, 20 and 50 % in DMSO.
Agents
------
Curdlan, a (1→3)-β-glucan preparation purified from the culture fluid of *Alcaligenes faecalis*, was purchased from Wako Pure Chemical Industries, Ltd.(Osaka, Japan), and suspended in 10 mg/ml in saline for intradermal injection and in 5 mg/ml for intraperitoneal injection. Hexadecyltrimethylammonium bromide (HTAB), human myeloperoxidase (MPO), and tetramethylbenzidine (TMB) were purchased from Sigma-Aldrich Japan (Tokyo). Polyoxyethylene(20) sorbitan monolaurate (Tween 20) was from Wako Pure Chemical Industries, Ltd.. Prednisolone injection (10 mg/ml) was from Mitaka Pharmaceutical, Ltd. (Tokyo). Dulbecco\'s phosphate-buffered saline (PBS) was from Invitrogen Corp. (Carlsbad, CA, USA) and stored at 4°C. Diff-Quik was from International Reagents Corp. (Hyogo, Japan). Geraniol and linalool were from Wako Pure Chemical Industries, Ltd.. Terpinen-4-ol and β-citronellol were from Tokyo Kasei Kogyo Co., Ltd. (Tokyo). Hair remover, anchone^®^was from Imju Co., Ltd. (Tokyo).
Animals
-------
All animal experiments were performed according to the guidelines for the care and use of animals approved by Teikyo University. Six week-old female ICR mice (Charles River Japan, Inc., Kanagawa, Japan) were used for all animal experiments except the one using 6 week-old female HR-1 hairless mice (Hoshino Laboratory Animals, Saitama, Japan). The photoperiods were adjusted to 12 hr of light and 12 hr darkness daily, and the environmental temperature was constantly maintained at 21°C. The mice were kept in cages housing 4--6 animals and were given ad libitum access to food and water.
Leukocyte accumulation in peritoneal cavity
-------------------------------------------
Fur in the dorsal region of mice (n = 5), approximately 20 × 50 mm square, was removed on day -3. The animals wore neck collars on day -2 to prevent their licking of the essential oils from the skin. On day 0, 200 μl of curdlan suspension (5 mg/ml) was injected intraperitoneally. A negative control group of mice was injected with 200 μl of saline instead of curdlan suspension. Immediately and 3 hr later, 100 μl of 20 % geranium oil in DMSO was dropped on the dorsal skin and gently spread over the fur-removed area using a glass spreader. To determine the number of leukocytes, mice were sacrificed with carbon dioxide 6 hr after curdlan injection. Three ml of PBS was then injected into their peritoneal cavity, and 2 ml of exudates were taken from the cavity to collect leukocytes. After centrifugation at 350 × g at 4°C for 5 min, the precipitate was suspended in 2 ml of PBS containing 10% heat-inactivated fetal calf serum (PBS solution). The numbers of leukocytes were measured by an electric cell counter; Celltac (Nihon Kohden Corporation, Tokyo). This leukocyte suspension was used for Diff-Quik staining and MPO assay as described below.
Diff-Quik staining
------------------
Neutrophils recovered from the peritoneal cavity were fixed on slide glass by cytocentrifugation and stained by Diff-Quik as described previously \[[@B15]\]. Briefly, the leukocyte suspension was diluted to about 1 × 10^6^cells/ml. Two hundred μl of the suspension was poured into a plastic tube attached to a slide glass and cytocentrifuged at 75 × g for 5 min, the slide glasses were then stained by Diff-Quik. Percentage ratio and the number of neutrophils were calculated by counting the neutrophil number of more than 50 leukocytes/sample.
Myeloperoxidase(MPO) assay for leukocyte suspensions
----------------------------------------------------
The MPO assay was based on the method of De Young et al. \[[@B16]\] and partly modified. One ml of the leukocyte suspension was centrifuged at 620 × g at 4°C for 2 min. The precipitate was suspended in 1 ml of 80 mM sodium phosphate buffer, pH5.4, containing 0.5% HTAB (0.5% HTAB solution), freeze-thawed 3 times and centrifuged at 1400 × g at 4°C for 5 min. Triplicate 30 μl samples of resulting supernatant were poured into 96 well microtiter plates. For assay, 200 μl of a mixture containing 100 μl phosphate buffered saline, 85 μl 0.22 M sodium phosphate buffer, pH5.4, and 15 μl of 0.017 % hydrogen peroxide were added to the wells. The reaction was started by the addition of 20 μl of 18.4 mM TMB•2HCl in 8 % aqueous dimethylformamide. Plates were stirred and incubated at 37°C for 3 min and then placed on ice where the reaction was stopped by addition to each well of 30 μl of 1.46 M sodium acetate, pH3.0. The MPO value was evaluated by measuring the absorbance of samples at 620 nm (OD value) and being converted it into MPO values per mouse. The MPO activity was expressed by relative values calculated by the following formula: (MPO value recovered from oil-treated mice)/(MPO value recovered from control mice) × 100 (%)
Skin preparation from mice with intraperitoneal injection of essential oils
---------------------------------------------------------------------------
Fur in the abdominal region of mice (n = 5--6) was removed on day -3. On day 0, 50 μl of curdlan suspension (10 mg/ml) in saline was injected intradermaly in the abdominal skin of mice. Immediately and 3 hr after curdlan injection, 200 μl of the diluted geranium oil solution was injected intraperitoneally. A dose of 2.5 % solution corresponded to 5 μl of pure oil. The control group of mice was received 200 μl of 2.5 % DMSO solution. One hundred μl/flank × 2 of prednisolone was injected subcutaneously to another active control group of mice 1 hr before curdlan injection, instead of essential oil. All mice were sacrificed with carbon dioxide 6 hr after curdlan injection. Skin was cut off in a 6 mm diameter area, weighed and placed in 1.05 ml of 0.5 % HTAB solution, and stored at -20°C until assay. The average weight of the skin was calculated as a parameter of edema.
Skin preparation from mice after cutaneous application of essential oils or its components
------------------------------------------------------------------------------------------
Fur in the abdominal and dorsal regions of mice (n = 15 for 5--20 μl geranium oil in experiment shown in Fig [3(b)](#F3){ref-type="fig"}, n = 3--6 for another experiments) except hairless mice (n = 4--5) was removed on day -3. The animals wore neck collars on day -2 to prevent their licking essential oils from the skin. On day 0, 50 μl of curdlan suspension (10 mg/ml) in saline was injected intradermaly to mice. Immediately and 3 hr later, 100 μl solution of a designated concentration of the essential oil or its components was dropped on the dorsal skin and gently spread over the fur-removed area using a glass spreader. A dose of 20 % solution corresponded to 20 μl of pure oil. The control group of mice was applied 100 μl of DMSO. Their skin preparations were obtained 6 hr after curdlan injection as described above.
Myeloperoxidase(MPO) assay for skin homogenate
----------------------------------------------
Frozen samples were thawed at room temperature and homogenized for 45 sec at 0°C by Polytron (Kinematica AG, Lucerne, Switzerland). The homogenates were poured into sampling tubes and centrifuged at 12000 × g at 4°C for 15 min. The resulting supernatants were used for MPO assay as described above. The MPO value per each skin sample was calculated.
Statistical analysis
--------------------
The results were expressed by the mean ± standard deviation. All statistical analysis was calculated using the StatView software. Statistical analysis was performed as follows; Students t-test for Fig [3(c)](#F3){ref-type="fig"}, Dunnett after ANOVA for Fig [2(a)](#F2){ref-type="fig"}, [3(a,b)](#F3){ref-type="fig"}, [4](#F4){ref-type="fig"} and [5](#F5){ref-type="fig"}, and Tukey-Kramer after ANOVA for Fig [1](#F1){ref-type="fig"}. Pearson\'s correlation coefficient was calculated for Fig [6](#F6){ref-type="fig"}.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Effects of intraperitoneal injection of curdlan against neutrophil accumulation and MPO activity. Curdlan or saline was injected intraperitoneally, and immediately and 3 hr after the injection, geranium oil or DMSO was applied cutaneously. After 6 hr, leukocytes were collected for Diff-Quick staining and MPO assay. (a) The number of leukocytes and cell differentials in peritoneal exudates. (b) The MPO values in peritoneal exudates. Each value represents an average of 5 mice and the standard deviation. \* p \< 0.05, \*\* p \< 0.01
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Effects of intraperitoneal injection of geranium oil on the inflammation by intradermal curdlan injection. Geranium oil was injected immediately and 3 hr after curdlan injection. Predonisolone, as positive control, was injected 1 h before curdlan injection. After 6 hr, skin was cut off for the MPO assay and histological examination. (a) The MPO activity from skin lesion. (b) Histological examination. Each value represents an average of 5--6 mice and the standard deviation. \*\* p \< 0.01 compared with control.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Effects of cutaneous application of geranium oil on MPO activity by intradermal curdlan injection. Geranium oil was applied immediately and 3 h after curdlan injection. After 6 h, skin was cut off for the MPO assay. (a) 20--100 μl of geranium oil was applied to fur-removed mice. (b) 5--20 μl of geranium oil was applied to fur-removed mice. Data represent the results obtained from 3 experiments. (c) 20 μl of geranium oil was applied to hairless mice. Each value represents an average of 4--5 mice for (a),(c) or 15 mice for (b), and the standard deviation. \* p \< 0.05, \*\* p \< 0.01 compared with control.
:::
:::
::: {#F4 .fig}
Figure 4
::: {.caption}
######
Effects of cutaneous application of essential oils on MPO activity by intradermal curdlan injection. 10 μl of essential oils was applied to fur-removed mice (n = 3--4). Each value represents an average of mice, and the standard deviation. \* p \< 0.05, \*\* p \< 0.01 compared with control.
:::
:::
::: {#F5 .fig}
Figure 5
::: {.caption}
######
Effects of cutaneous application of essential oil components on MPO activity by intradermal curdlan injection. 5 μl of essential oil components was applied to fur-removed mice. Each value represents an average of 4--5 mice, and the standard deviation. \* p \< 0.05, \*\* p \< 0.01 compared with control.
:::
:::
::: {#F6 .fig}
Figure 6
::: {.caption}
######
Correlation between the MPO value and skin weight. This figure is composed from data obtained from all independent experiments in which control mice and mice applied with 20 μl geranium oils were used. Open and filled circles represent control and geranium groups, respectively. r = 0.757, p \< 0.0001. (n = 49 for control group, n = 26 for geranium group)
:::
:::
Results
=======
Inflammation of the skin by curdlan intradermal injection
---------------------------------------------------------
Skin inflammatory response induced by intradermaly injected curdlan (0.5 mg/50 μl), was investigated first by using two parameters, the MPO value of skin homogenates and skin weight. The MPO value and skin weight of the skin lesion 6 hr after curdlan injection were 4.54 ± 2.43 units/skin lesion and 22.5 ± 8.3 mg (n = 49), which were significantly higher than those of saline injection, 0.21 ± 0.14 units/skin lesion and 8.9 ± 1.4 mg, respectively (n = 4). This indicates that curdlan injection caused neutrophil accumulation, which was monitored by increase in the MPO activity, and skin edema, which was observed by increase in skin weight.
Correlation between myeloperoxidase(MPO) activity and neutrophil accumulation
-----------------------------------------------------------------------------
We examined the neutrophil accumulation in the peritoneal cavity after curdlan injection microscopically and enzymatically using MPO activity. The effect of cutaneous application of geranium oil on these changes was observed.
Figure [1(a)](#F1){ref-type="fig"} shows the number of leukocytes and cell differentials in peritoneal exudates, which were determined using Diff-Quik staining. About 3.66 ± 0.49 × 10^6^leukocytes were recovered from the peritoneal cavity of saline-induced mice, and the intraperitoneal injection of 200 μl curdlan solution increased this number to 9.48 ± 1.78 × 10^6^cells. Neutrophils were rarely observed in peritoneal cells of saline-injected mice and the increased content of peritoneal leukocytes in curdlan-injected mice was mostly neutrophils. The MPO value of leukocyte preparation obtained 6 hr after curdlan intraperitoneal injection was 5.36 ± 1.86 units/mouse, which was significantly higher than that of the mice without curdlan (0.064 ± 0.026 units/mouse) (Fig. [1(b)](#F1){ref-type="fig"}).
Figure [1](#F1){ref-type="fig"} also shows that compared to the curdlan control, cutaneous application of geranium oil to these mice significantly lowered the number of leukocytes and neutrophils (Fig. [1(a)](#F1){ref-type="fig"}) as well as the MPO value (Fig. [1(b)](#F1){ref-type="fig"}).
These results indicated that intraperitoneal injection of curdlan caused both the accumulation of neutrophils in the peritoneal cavity and increase of the MPO value, and that cutaneous application of geranium oil suppressed both of them. Therefore, we confirmed that the MPO activity corresponds to the number of neutrophils and that MPO activity can be used as a parameter for this number.
Effects of geranium oil on inflammation induced by curdlan intradermal injection
--------------------------------------------------------------------------------
The effects of geranium oil administered intraperitoneally or cutaneously on the inflammation induced by curdlan intradermal injection were examined.
At first, geranium oil was injected intraperitoneally to mice. Prednisolone was used as an active control. As shown in Fig. [2(a)](#F2){ref-type="fig"}, administration of 2 mg per mouse of prednisolone suppressed the MPO activity to 7 ± 3 %. Similarly but to a lesser degree, 5 and 10 μl of geranium oil significantly lowered the MPO activity to 30 ± 15 and 14 ± 10 %, respectively.
From the histological examination (Fig. [2(b)](#F2){ref-type="fig"}), it was observed that prednisolone clearly suppressed neutrophils accumulated following curdlan injection. Geranium oil also suppressed this accumulation, however, the suppression was not as strong as by prednisolone.
In the second experiment, we examined the effect of the cutaneous administration of geranium oil. As shown in Fig. [3(a)](#F3){ref-type="fig"}, 20, 50 and 100 μl per mouse of geranium oil application lowered the MPO activity significantly (24 ± 14, 13 ± 1 and 19 ± 13 %, respectively). In this experiment, we observed that the mice receiving 50 and 100 μl of geranium oil exhibited an unusual behavior (sedated condition with loss of normal active movement) after the second administration. Therefore, the dose of geranium oil tested was reduced to 5, 10 and 20 μl (Fig. [3(b)](#F3){ref-type="fig"}) and they showed significant suppression of the MPO activity (63 ± 40, 56 ± 32, and 37 ± 23 % respectively). These data depicted in Fig [3(a)](#F3){ref-type="fig"} and [3(b)](#F3){ref-type="fig"} suggest that geranium oil suppress MPO activity in a dose-dependent manner.
In the third experiment, the similar effect of geranium oil on the hairless mice was examined. In this experiment, treatment by hair remover was omitted. Figure [3(c)](#F3){ref-type="fig"} shows that the MPO activity in hairless mice was significantly reduced by geranium application (32 ± 8 %) as in the case of fur-removed mice. This indicated that the geranium application suppressed the fur-removed skin and normal hairless skin similarly, and the effect of the remover was negligible.
Effects of cutaneous application of various essential oils
----------------------------------------------------------
We compared the effects of 10 μl of various essential oils (geranium, lavender, tea tree, eucalyptus, and juniper) against the MPO activity. Although all oils except juniper oil lowered the activity significantly (Fig. [4](#F4){ref-type="fig"}), the inhibitory activity of geranium oil was estimated to be strongest (34 ± 27 %). On the other hand, juniper oil did not significantly suppress the activity (66 ± 7 %).
Effects of cutaneous application of components of essential oils
----------------------------------------------------------------
We compared the activities of the main constituents of geranium, lavender and tea tree oils. As shown in Fig. [5](#F5){ref-type="fig"}, geraniol and terpinen-4-ol lowered the MPO activity (65 ± 23 and 68 ± 32 %, respectively), but not significantly, and linalool and β-citronellol did not lower the activity (126 ± 48 and 89 ± 37 %, respectively).
Correlation between MPO activity and skin weight
------------------------------------------------
We measured the skin weight as a parameter of edema for each experiment. The correlation between skin weight and the MPO activity was examined for all control mice and all mice applied with 20 μl of geranium oil. As shown in Fig. [6](#F6){ref-type="fig"}, the skin weight of each mouse closely correlated with the MPO activity (r = 0.757, p \< 0.0001).
The average skin weight and the MPO value were 22.5 ± 8.3 mg and 4.54 ± 2.43 units/skin lesion for control (n = 49), and 12.0 ± 3.7 mg and 1.16 ± 0.75 units/skin lesion for the geranium group (n = 26), respectively. This indicates that geranium oil suppressed both the neutrophil accumulation and edema induced by curdlan.
Discussion
==========
In this study, we showed that cutaneous application of geranium oil (5--100 μl) to mice suppressed cellular inflammation induced by curdlan dose-dependently, as monitored by the MPO activity of peritoneal cavity and skin. This suppressive activity of geranium oil seemed very potent in comparison with those of other essential oils: 10 μl of lavender, tea tree, and eucalyptus oils lowered the activity significantly, but each was weaker than that of geranium oil. Juniper oil did not suppress the activity.
It was reported that MPO, a marker enzyme of neutrophil granules, can be used as a parameter of infiltration of neutrophils in various inflammatory experiments using tissues including skin \[[@B16]-[@B18]\]. We confirmed here that the MPO activity was closely related to the number of neutrophils which infiltrated into the peritoneal cavity after intraperitoneal injection of curdlan with or without administration of geranium oil. Histological examination of the skin, into which curdlan was injected 6 hr earlier, also showed that the degree of infiltration of inflammatory cells (perhaps neutrophils), at least qualitatively, correlated with the MPO values of the skin homogenates. These observations indicate the MPO activity can be used as a marker of neutrophil accumulation in our experiments.
As far as we know, this is the first experimental report indicating that cutaneous application of essential oils, especially geranium oil, effectively inhibited neutrophil accumulation *in vivo*. Although some irritants appeared to have anti-inflammatory activity, the action of geranium oil can not be explained by such a manner, since geranium oil did not induce neutrophil accumulation by itself as reported previously \[[@B9]\]. Recently, Brand et al. reported that tea tree oil inhibited histamine-induced edema \[[@B4]\], but did not change leukocyte infiltration in a murine contact dermatitis model \[[@B1]\]. In our results, cutaneous application of 10 μl of tea tree oil decreased the MPO activity in curdlan-injected skin weakly but significantly (Fig. [4](#F4){ref-type="fig"}), although intraperitoneal administration of the oil did not suppress the neutrophil accumulation in the peritoneal cavity \[[@B9]\]. Moreover, our previous report showed that geranium oil more effectively suppressed neutrophil adherence response induced by TNF-α stimulation than tea tree oil *in vitro*\[[@B10]\]. All these findings may suggest that geranium oil has a different type of suppressive activity for inflammation from that of tea tree oil. In order to obtain conclusive findings for quantitative differences in the anti-inflammatory activities of essential oils, we must examine their activity in a dose-dependent manner and their bioavailability based on their skin absorption.
We used curdlan, a linear (1→3)-β-glucan, as an inflammatory agent. It has already been reported that curdlan causes local inflammation and induces polymorphonuclear leukocyte accumulation \[[@B19]\], and that the number of neutrophils in the peritoneal cavity greatly increases 6 hr after curdlan intraperitoneal injection \[[@B15]\]. (1→3)-β-glucan is known to activate complements to release C5a, a neutrophil chemoattractant \[[@B20]\], and may induce production of chemotactic cytokines through interaction with toll-like receptors 2,6 on macrophages \[[@B21]\]. Therefore, we can assume that curdlan may induce neutrophil accumulation through these polysaccaride-recognition mechanisms. It is possible that geranium oil interferes with these polysaccaride-recognition steps, however, we wish to note another possibility: geranium oil may suppress neutrophil response in the accumulation step, because this oil can suppress neutrophil recruitment by casein injection *in vivo*as reported previously \[[@B9]\], and can strongly suppress neutrophil response by TNF-α stimulation *in vitro*\[[@B10]\]. Details of the mechanisms involved in the suppression of inflammation remain to be clarified.
We tested the suppressive activity for the MPO response of the main constituents (5 μl) of essential oils, geraniol and β-citronellol (geranium), linalool (lavender) and terpinen-4-ol (tea tree). Geraniol and terpinen-4-ol seemingly suppressed the activity, but the others did not. Thus, geraniol, not β-citronellol, is thought to be an active component of geranium oil. On the other hand, linalool showed no activity, although lavender oil lowered it significantly. It is possible that linalool is not an active component of lavender oil. Further examinations on the activity of various other components and their combinations are necessary to evaluate the active principles of essential oils.
The cutaneaous application of geranium oil suppressed the MPO activity dose-dependently. The GC analysis of the blood 5 min after geranium oil application showed peaks from geranium oil such as β-citronellol, which indicated some of components of the oil were absorbed in the blood circulation very quickly (data not shown). We think that the suppression by oils is done through skin absorption, although we also need to take into account the effect of inhalation of essential oil because of its high volatility. The MPO activity using hairless mice was also suppressed to about 30% by geranium oil, indicating that suppression activity was not interfered with hair remover.
We must note that in these experiments, solvent of essential oil treatments is DMSO. It is known to facilitate the permeation of some drugs. DMSO might modulate the effects of essential oils, although we reported that intraperitoneal injection of essential oils with 2.5% DMSO as solvent, which is relatively lower concentration of DMSO, lowered the neutrophil accumulation in previous study \[[@B9]\]. In further experiments, we need to examine the effects of essential oils using other solvents such as carrier oils.
In this study, we also examined the effect of geranium oil on the edema using skin weight as well as the MPO activity. Normal skin weight was about 8.9 ± 1.4 mg and increased to 22.5 ± 8.3 mg by curdlan injection. This difference indicates the edema by inflammation. Twenty μl of geranium oil reduced the weight to 12.0 ± 3.7 mg, indicating that the oil strongly suppresses the edema induced by curdlan injection. It is well known that tea tree and lavender oils suppress the edema induced by histamine \[[@B3],[@B4]\]. As shown in Fig. [6](#F6){ref-type="fig"}, edema is closely correlated with the MPO activity, and geranium oil reduced both of them. The physiological meaning of this correlation should be clarified.
In aromatherapy, skin application of essential oils to limited parts of the body or in a full body massage is popular and several of these oils are used as a therapeutic treatment for inflammatory symptoms with lesional neutrophil accumulation: rheumatoid arthritis, aphthous stomatitis, and lesional bacterial or fungal infections \[[@B22]\]. In these cases, local application of relatively concentrated (more than 5%) oils to the lesion is effective. But full body massage with a relatively lower concentration (around 3%) of essential oils is also used for some local inflammatory conditions. These clinical usages of essential oils were established traditionally, but their pharmacological efficacies have not been fully confirmed by scientific research. Our results presented here suggest that systemic application of essential oils seems reasonable, because neutrophil accumulation and edema were suppressed through systemic application of essential oil, especially geranium oil, even though the concentration of the oil is higher than that used clinically. This suggests that some essential oils such as geranium may suppress local inflammatory symptoms through systemic skin application in human.
The therapeutic benefit of these essential oils and the roles of anti-inflammatory activity in their therapeutic actions is an urgent theme to be investigated.
Conclusion
==========
Cutaneous application of several essential oils, especially geranium oil, to mice suppressed the cellular inflammation induced by curdlan dose-dependently, as monitored by the MPO activity of peritoneal cavity and skin. This suggests that essential oils using in aromatherapy massage may suppresses the inflammatory symptoms related with neutrophil accumulation and edema.
Competing interests
===================
This work was supported in part by a grant from the Kampo Medicine Research Fund (Tokyo) and a grant (No.15590401) from the Ministry of Education. Culture, Sports, Science and Technology of Japan.
Authors\' contributions
=======================
NM participated in the design of the study, carried out the animal study and GC analysis, and wrote the manuscript. YS and HO carried out the animal study and GC analysis, and performed the statistical analysis. HI and HY helped to carry out the animal study. SI helped to carry out the GC analysis and draft the manuscript. SA conceived of the study, participated in its design and coordination, and helped to carry out the study and write the manuscript. All authors read and approved the final manuscript.
|
PubMed Central
|
2024-06-05T03:55:55.079428
|
2005-2-10
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074347/",
"journal": "J Inflamm (Lond). 2005 Feb 10; 2:1",
"authors": [
{
"first": "Naho",
"last": "Maruyama"
},
{
"first": "Yuka",
"last": "Sekimoto"
},
{
"first": "Hiroko",
"last": "Ishibashi"
},
{
"first": "Shigeharu",
"last": "Inouye"
},
{
"first": "Haruyuki",
"last": "Oshima"
},
{
"first": "Hideyo",
"last": "Yamaguchi"
},
{
"first": "Shigeru",
"last": "Abe"
}
]
}
|
PMC1074348
|
Editorial
=========
*Particle and Fibre Toxicology*is a new Open Access, peer-reviewed, online journal published by BioMed Central, which will publish articles pertaining to all aspects of the toxicological effects of particles and fibres. Particles and fibres are toxicologically important in many scenarios. There is exposure, leading to adverse health effects, during manufacture and use of industrial products, accidental exposure during disturbance of the earth\'s crust in mining and quarrying, accidental exposure from general anthropogenic sources in the environment, and exposure to organic particles containing biologically active compounds such as endotoxins and glucans. Traditionally the particles and fibres most intensively studied include, PM10, PM2.5, asbestos, synthetic vitreous fibres, metals, quartz (crystalline silica), and coalmine dust. There are, however, numerous other particle types under study. The diseases and conditions caused or affected by exposure to particles include cancer, fibrosis, asthma, chronic obstructive pulmonary disease (COPD) and cardiovascular disease. Recently there has been increasing interest in the extra-pulmonary effects of particles, such as transport of particles to the blood and brain and the consequences of such transport on neurological function, blood coagulation and cardiovascular health.
Recently a whole new class of particles, those in the nanoparticle size range, have come under intense scrutiny. Among those nanotubes and their toxicological effects are also receiving increased attention. Nanoparticles are increasingly used in \'Nanomedicine\' for diagnostic, imaging and therapeutic purposes and these require special toxicological considerations \[[@B1]\].
Overview of journal scope
=========================
*Particle and Fibre Toxicology*welcomes manuscripts dealing with the following toxicological aspects pertaining to particles and fibres: mechanistic studies, dosimetry, physicochemical factors in particle toxicity, animal studies, the role of size in particle toxicity, the role of composition in particle toxicity, cell studies, studies in acellular systems, biopersistence and durability studies, studies aimed at improving the dose metric, inhalation studies with real and surrogate particles, mathematical modelling, deposition and clearance studies, anatomical and histological studies, regulation and regulatory studies, pathology studies, risk assessment, translocation studies on the fate of particles, and all toxicological studies that throw light on how particles cause harmful effects in the lungs, liver, blood and nervous system. *Particle and Fibre Toxicology*especially welcomes manuscripts relating to the toxicological hazards and risks of nanoparticles and nanotubes especially when used for specific biomedical purposes. The journal will consider the following types of article: research, commentary, debate, hypothesis, methodology articles, reviews and short reports.
Why is *Particle and Fibre Toxicology*needed?
=============================================
The field of particle and fibre toxicology has a long and famous history. During the 15th and 16th centuries Agricola and Paracelsus were the first to document lung diseases caused by inhalation of dust and by the time of the Industrial Revolution in 18th and 19th century, the toll of ill-health from dusty trades was enormous, with industrial hygiene still in its infancy. By the 20th century the global epidemic of lung disease caused by asbestos and crystalline silica demonstrated that the toxicology of dusts and fibres was still \'a work in progress\'. In the 21st century we face the old challenges plus new ones, such as environmental particulate air pollution, the new focus on the extra-pulmonary effects of particles and potential risks from nanoparticles arising from the nanotechnology industry. A journal dedicated to the toxicology of particles and fibres is a new and worthy addition to this long history.
Studies on topics related to particle and fibre toxicology translate into thousands of scientific articles each year but historically these have been scattered across numerous journals. The target toxicology journal is usually chosen on the basis of the endpoint under study, or if humans are exposed, occupational and environmental journals may be chosen. Those experienced in particle and fibre research appreciate the special expertise required to carry out and interpret toxicological studies because of potentially confusing factors such as absorption, aggregation, surface coating, dosimetry and overload. Such details are not always appreciated by non-specialists. *Particle and Fibre Toxicology*will provide a unique single outlet and focus of expertise in this multi-disciplinary subject. Those involved in research into particles and fibres include physicists, chemists, materials scientists, environmental scientists, hygienists, mechanistic toxicologists, inhalation toxicologists, cell biologists, medical specialists, epidemiologists and risk assessors. A single target journal specialising in particle and fibre toxicology will provide a focus for the diverse community employed in this global research enterprise. Focusing on the interface between particles/fibres and humans, *Particle and Fibre Toxicology*should aid in the advance of this important discipline to the ultimate improvement of occupational and public health.
Peer review policy
==================
Publication of research articles in the broad area of particle toxicology is dependent on: relevance to the aims of the journal; scientific validity and excellence; and coherence with the research area as a whole, as judged by two independent reviewers. An International Editorial Board oversee the peer review of manuscripts submitted to *Particle and Fibre Toxicology*. Peer review is anonymous. A recommendation of acceptance or rejection is at the discretion of the Editors based on the reviewers\' recommendations and will be agreed in consultation with the reviewers. Articles will be published immediately upon acceptance (after peer review) and soon after they will be listed in PubMed and archived in PubMed Central.
Open Access
===========
*Particle and Fibre Toxicology*\'*s*Open Access policy changes the way in which articles are published. First, all articles become freely and universally accessible online, and so an author\'s work can be read by anyone at no cost. Second, the authors hold copyright for their work and grant anyone the right to reproduce and disseminate the article, provided that it is correctly cited and no errors are introduced \[[@B2]\]. Third, a copy of the full text of each Open Access article is permanently archived in an online repository separate from the journal. *Particle and Fibre Toxicology*\'s articles are archived in PubMed Central \[[@B3]\], the US National Library of Medicine\'s full-text repository of life science literature, and also in repositories at the University of Potsdam \[[@B4]\] in Germany, at INIST \[[@B5]\] in France and in e-Depot \[[@B6]\], the National Library of the Netherlands\' digital archive of all electronic publications.
Open Access has four broad benefits for science and the general public. First, authors are assured that their work is disseminated to the widest possible audience, given that there are no barriers to access their work. This is accentuated by the authors being free to reproduce and distribute their work, for example by placing it on their institution\'s website. It has been suggested that free online articles are more highly cited because of their easier availability \[[@B7]\]. Second, the information available to researchers will not be limited by their library\'s budget, and the widespread availability of articles will enhance literature searching \[[@B8]\]. Third, the results of publicly funded research will be accessible to all taxpayers and not just those with access to a library with a subscription. As such, Open Access could help to increase public interest in, and support of, research. Note that this public accessibility may become a legal requirement in the USA if the proposed Public Access to Science Act is made law \[[@B9]\]. Fourth, a country\'s economy will not influence its scientists\' ability to access articles because resource-poor countries (and institutions) will be able to read the same material as wealthier ones (although creating access to the internet is another matter \[[@B10]\]).
Funding Open Access
===================
6 months after the launch of *Particle and Fibre Toxicology*all articles accepted by the journal will be subject to an article processing charge of \$525. The fee will be payable by the authors unless their institution is a BioMed Central member, but will be waived in cases of genuine hardship.
Conclusion
==========
The information revolution is a double-edged sword felt as keenly in science as anywhere else. We live in an age where health science is a multimillion pound/euro/dollar industry and its product, scientific articles, are produced in vast numbers. Scientists and regulators struggle to find this information and interpret it in ways that will benefit human health. This is particularly relevant in toxicology of particles and fibres because of its multidisciplinary nature. Open access publishing clearly helps us make our way through the forest of information by publishing rapidly and allowing easy access to research articles. Our new journal *Particle and Fibre Toxicology*should also help by acting as a focus for multidisciplinary research in this area. We therefore look on *Particle and Fibre Toxicology*as a significant opportunity to advance this important area of public and occupational health and invite scientists in this field to submit their work.
Competing interests
===================
The Editor-in-Chief and the Editorial Board of *Particle and Fibre Toxicology*have no financial incentive to accept manuscripts as they are not paid on the basis of the number of manuscripts accepted. In fact, they do not receive any financial remuneration for their involvement with the journal. We insist that decisions about a manuscript are based on the quality of the work, not on whether the authors\' can pay the article-processing charge. Authors will be asked to declare any competing interests that they have.
|
PubMed Central
|
2024-06-05T03:55:55.081782
|
2004-12-3
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074348/",
"journal": "Part Fibre Toxicol. 2004 Dec 3; 1:1",
"authors": [
{
"first": "Ken",
"last": "Donaldson"
},
{
"first": "Paul",
"last": "Borm"
}
]
}
|
PMC1074349
|
Background
==========
Exposure to fine particulate matter (PM~2.5~) in the ambient air increases daily deaths \[[@B1]\] and hospitalization for cardiovascular diseases \[[@B2]\] in the U.S. and throughout the world \[[@B3]\] with most effects within one day after exposure. It is estimated that 800,000 excess deaths worldwide each year may be attributable to particulate matter air pollution \[[@B4]\], possibly secondary to myocardial infarction \[[@B5]\], life-threatening arrhythmias \[[@B6]\] or heart failure, as reviewed in a recent American Heart Association scientific statement \[[@B7]\]. Yet, the underlying pathophysiological mechanisms that link PM~2.5~and cardiopulmonary mortality are poorly understood.
Particles of motor vehicle origin appear to be especially potent with regard to increased mortality \[[@B8],[@B9]\] and hospital admissions due to cardiovascular diseases \[[@B10]\]. Vehicles represent a microenvironment with potentially high exposure to air pollutants from mobile sources. We previously showed that occupational in-vehicle PM~2.5~exposure to North Carolina Highway Patrol troopers was associated with changes in cardiac parameters, blood proteins associated with inflammation, hemostasis and thrombosis, and increased red blood cell volume (MCV) 10 to 15 hours after completing their shift \[[@B11]\]. These findings were little affected by potential confounders. Controlling for estimates of occupational stress even slightly improved the strength of association with some cardiac parameters. In this paper, we investigated how these health endpoints were associated with specific sources of PM~2.5~.
Results
=======
Subjects
--------
Data from nine male non-smoking troopers (8 Caucasian, 1 African-American) were used for the analysis: ten participated, one was excluded due to very high numbers of ectopic beats and high serum cholesterol. This left a total of 36 person-days with valid health data. Their age ranged from 23 to 30 years (mean 27.3 years), their weight from 74 to 102 kg (87 kg), their height from 168 to 191 cm (179 cm), and their body mass index from 24 to 31 kg/m^2^(27 kg/m^2^). All were in excellent physical condition.
Exposure inside the cars and source identification
--------------------------------------------------
Elemental PM~2.5~-components and co-pollutants that were correlated to the PM~2.5~measurements were included in the analysis, if they had over 75% of the data above the reporting limit. Table [1](#T1){ref-type="table"} shows their in-vehicle concentrations and the correlations to the PM~2.5~measurements. Data of 36 individual samples were available after correction of one silicon-outlier, and replacement of three missing benzene and two missing aldehydes values by their respective means. All concentrations measured were below current occupational threshold limits.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Components included in the analysis In-vehicle concentrations of the elemental components of PM~2.5~and gaseous co-pollutants included in the analysis (n = 36 samples): Arithmetic average, standard deviation and correlation (Spearman-Rho) to PM~2.5Mass~and PM~2.5Lightscatter~. \*) p \< 0.05.
:::
Component Average Standard Deviation Correlation to PM~2.5Mass~ Correlation to PM~2.5Lightscatter~
------------------------------ --------- -------------------- ---------------------------- ------------------------------------
Benzene (ppb) 3.73 2.9 0.50 \* 0.31 \*
Aldehydes (μg/m^3^) 34.6 14.9 0.34 \* 0.52 \*
CO (ppm) 2.6 1.1 0.52 \* 0.52 \*
Aluminum (Al, ng/m^3^) 66.0 54.5 0.58 \* 0.31 \*
Silicon (Si, ng/m^3^) 240.0 542.0 0.66 \* 0.23
Sulfur (S, ng/m^3^) 1703.0 812.0 0.58 \* 0.88 \*
Calcium (Ca, ng/m^3^) 48.2 33.5 0.37 \* 0.22
Titanium (Ti, ng/m^3^) 11.7 10.0 0.41 \* 0.15
Chromium (Cr, ng/m^3^) 2.1 1.7 0.51 \* 0.32 \*
Iron (Fe, ng/m^3^) 371.0 352.0 0.71 \* 0.33 \*
Copper (Cu, ng/m^3^) 33.1 18.8 0.16 0.50 \*
Selenium (Se, ng/m^3^) 12.6 1.2 0.38 \* 0.26
Tungsten (W, ng/m^3^) 5.6 5.9 0.37 \* 0.39 \*
PM~2.5Mass~(μg/m^3^) 23.0 8.0 1 0.71 \*
PM~2.5Lightscatter~(μg/m^3^) 24.1 13.5 0.71 \* 1
:::
Source model A consisted of a factor analysis using all elements and co-pollutants listed in Table [1](#T1){ref-type="table"}. Four source factors were identified. Figure [1A](#F1){ref-type="fig"} shows the factor loadings (loadings larger than 0.4 are highlighted) and Table [2](#T2){ref-type="table"} shows the model characteristics. Factor 1 was dominated by silicon and aluminum (named \"crustal\" factor), factor 2 by iron, chromium and titanium (\"steel wear\" factor), factor 3 by benzene and carbon monoxide (\"gasoline\" factor) and Factor 4 by copper, sulfur and aldehydes (\"speed-change\" factor).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Source factors loadings.**Factor loadings of the different components of the two models and the proposed sources for these factors. Loadings large than 0.4 are highlighted in yellow.
:::
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Source model characteristics Characteristics of the two models and their factors of the principal factor analysis and their associations with the two PM~2.5~-measures: Model A includes all components shown in Table 1, Model B excludes Ca, Cr, Se and W.
:::
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Model characteristics Model A Model B
------------------------------------------------- ---------------------- ------------------------- ----------------------- --------------------------- --------------------------- ----------------------- ---------------------------
Number of exposure variables included 13 9
Total variance explained 62.3% 68.9%
Correlation with PM~2.5Mass~(R^2^) 0.73 0.63
Correlation with PM~2.5Lightscatter~(R^2^) 0.71 0.52
Factor characteristics Factor 1 \"crustal\" Factor 2 \"steel wear\" Factor 3 \"gasoline\" Factor 4 \"speed-change\" Factor 1 \"road surface\" Factor 2 \"gasoline\" Factor 3 \"speed-change\"
Sum of squares of factor loadings 2.48 2.11 1.82 1.67 3.01 1.68 1.51
Proportion of total variance 19.1% 16.3% 14.0% 12.9% 33.5% 18.6% 16.8%
Slope^a^of correlation with PM~2.5Mass~ 2.95\ 4.11\ 2.16\ 4.50\ 4.74\ 3.03\ 3.21\
p = 0.0006 p \< 0.0001 p = 0.003 p \< 0.0001 p \< 0.0001 p = 0.002 p = 0.002
Slope^a^of correlation with PM~2.5Lightscatter~ 0.68\ 3.58\ 3.74\ 11.32\ 2.48\ 4.03\ 9.27\
p = 0.6 (n.s.) p = 0.01 p = 0.004 p \< 0.0001 p = 0.15 (n.s.) p = 0.026 p \< 0.0001
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
a\) Slope in μg/m^3^per 1 SD change in the exposure factor score
:::
Source model B was calculated using only elements that were clearly correlated to PM~2.5~and with the majority of data more than 3 sigma above background noise (i.e., without Ca, Cr, Se and W). Three source factors were identified (Figure [1B](#F1){ref-type="fig"} and Table [2](#T2){ref-type="table"}). Factor 1 was dominated by silicon, aluminum, titanium and iron (named \"road surface\"); factor 2 by benzene and carbon monoxide (\"gasoline\"); and factor 3 by copper, sulfur and aldehydes (\"speed-change\").
The factor \"road surface\" of Model B was significantly correlated to the factors \"crustal\" and \"steel wear\" of Model A (R = 0.80 and 0.64, respectively). Factor \"gasoline\" of A was correlated to factor \"gasoline\" of B (R = 0.80); and factor \"speed-change\" of A to \"speed-change\" of B (R = 0.91). In contrast, the source factors within each model were completely uncorrelated (R \< 0.09).
Health endpoints associated with sources
----------------------------------------
The associations between health endpoints and source factors were studied in a multivariate approach. Figure [2](#F2){ref-type="fig"} shows the results for Model A; Figure [3](#F3){ref-type="fig"} those for Model B (only health endpoints associated to one of the sources with p \< 0.05 are displayed). In both models, most of the significant health effect estimates were associated with the \"speed-change\" factor (MCL, SDNN, PNN50, supraventricular ectopic beats, % neutrophils, % lymphocytes, MCV, von Willebrand Factor, and protein C). The association with MCV remained unchanged when controlled for osmolality. Two significant associations were observed for the \"crustal\" factor of Model A (uric acid and MCL), none for the \"steel wear\" and the \"gasoline\" factor of either model.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Associations between Model A and selected health endpoints.**Effect estimates are shown as percent change per one standard deviation change in the source factors. Lines indicate the 95% confidence interval. Symbols represent the different factors: rectangle = \"crustal\", triangle = \"steel wear\", diamond = \"gasoline\" and circle = \"speed-change\". **Fig 2A:**blood endpoints. \*) The estimates for MCV were multiplied by ten to better fit the scale. **Fig 2B:**cardiac endpoints.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Associations between Model B and selected health endpoints.**Effect estimates are shown as percent change per one standard deviation change in the source factors. Lines indicate the 95% confidence interval. Symbols represent the different factors: rectangle = \"road surface\", diamond = \"gasoline\" and circle = \"speed-change\". \*) The estimates for MCV were multiplied by ten to better fit the scale.
:::
:::
Strong heteroscedasticity (i.e., an indication for a violation of the underlying statistical assumptions) was evidenced in the residual analysis of the models for red blood cell count, hematocrit and hemoglobin (which were significantly associated with the \"speed-change\" factor). However, every attempt to remove the heteroscedasticity by adjusting the variance-covariance structure also completely removed the significance of these associations.
Control for potential confounders
---------------------------------
The associations observed between factors and health parameters were tested for the following potential confounders: Temperature, relative humidity, the number of law-enforcement activities during the shift (as estimate of workload) and the average speed during the shift. Controlling for these confounders had no visible effect on most effect estimates of the \"crustal\" and the \"speed-change\" factor, and the associated health parameters. However, in Models A and B, including these confounders altered the effect estimates with blood urea nitrogen and vWF, especially including all confounders together into the models lowered the effect estimates for the source factor \"speed-change\" by about one fifth and the confidence interval included zero. In Model A, the estimate for PNN50 was not altered by any of the confounders, but including all confounders into the same model widened the confidence interval to include zero.
Discussion
==========
We previously reported that in-vehicle exposure to PM~2.5~was associated with increases in markers of inflammation and coagulation, and modulations of heart rate variability in Highway Patrol troopers \[[@B11]\]. Here we demonstrate that most health endpoints were associated to a PM~2.5~source factor that reflects speed-changing traffic conditions (dominated by copper, aldehydes and sulfur). Under such driving conditions, copper reflects wear of brakes, aldehydes reflect emissions from accelerating vehicles and sulfur reflects secondary aerosols and possibly diesel combustion products.
In Model A, four principal factors of PM~2.5~-exposure inside the patrol cars were identified. Their loadings suggest the main in-vehicle sources of PM~2.5~. Factor 1 reflects exposure to crustal material from the soils in the study region and the road surface (\"crustal\" factor). Factor 2 represents wear and tear of mechanical automotive parts, mostly chrome-titanium steels (\"steel wear\" factor). Factor 3 represents components derived from gasoline combustion (\"gasoline factor\"). Finally, factor 4 is characterized by components expected from speed-changing traffic (\"speed-change\" factor): copper from brakes \[[@B12]\] and aldehydes from engine emissions \[[@B13]\]. Note that photochemical processes \[[@B14]\] are an unlikely source for this factor, since urban background and roadside levels near free-flowing traffic were much lower than in-vehicle levels \[[@B15]\]. The source of the high sulfur loading is unclear. Diesel combustion of accelerating trucks would be a plausible source candidate. However, sulfur is ubiquitous on secondary urban aerosols. It was the most concentrated element on PM~2.5~in the study \[[@B15]\]. This prevents the identification of local sources with sulfur as a tracer. A cautious interpretation might be that factor 4 reflects particles from speed-changing traffic mixed with secondary urban particles; a mixture expected on roads in an urban-sprawl area like Raleigh.
Model B proposes only three sources. However, they correspond in principle to the sources from Model A, except that the factors \" crustal\" and \"steel wear\" seem to be collapsed into a single source factor \"road surface\". This notion is supported by the good correlation between the corresponding factors.
The average PM~2.5~concentration of ca. 23 μg/m^3^inside the vehicles was at a moderate level compared to the 24-hour National Ambient Air Quality Standard for PM~2.5~of 65 μg/m^3^. The two methods used to measure PM~2.5~were highly correlated. The differences of their correlations to the components (Table [1](#T1){ref-type="table"}) and to the source factors (Tables [2](#T2){ref-type="table"}) reflect the fact that two different methods were used to assess the particle mass \[[@B15]\]: PM~2.5Lightscatter~reflects mostly accumulation mode particles (0.2 to 2 μm), whereas PM~2.5Mass~includes some coarse dust including fine sand.
The \"speed-change\" factor (Models A and B) was significantly associated with increased percentage of neutrophil leucocytes in the circulating blood, with decreased percentage of lymphocytes, and with changes in markers of endothelial activation and hemostasis. Endothelial cells are a major storage site for von Willebrand factor \[[@B16]\], and plasma levels of vWF serve as markers for endothelial activation \[[@B17]\]. Protein C is an antithrombotic agent, it is activated on the endothelium and reduced in the blood after inflammatory stimulation due to protein C consumption \[[@B18]\]. Consequently, endothelial cells may be involved in both inflammatory and coagulatory responses to traffic particles. Blood urea nitrogen was also associated with the \"speed-change\" factor. This finding would be consistent with the postulated inflammation since blood urea nitrogen increases several hours after an inflammatory stimulus (pig model) \[[@B19]\]. Blood urea nitrogen and vWF lost significance when controlled for all potential confounders together, although the effect estimates were not much changed. It should be noted that including this many confounders into a model with a relatively small number of samples reduces the strength of the statistics considerably.
The \"speed-change\" factor was significantly associated with changes in MCV (independent of osmolality) and similar to the association for PM~2.5Lightscatter~with MCV reported earlier \[[@B11]\]. The present analysis suggests that particles originating from speed-changing traffic are an important source of this association with circulating red blood cell mean volume (while other red blood cell indices were not affected). This is consistent with in-vitro blood experiments, where high concentrations of particles caused dose-dependent hemolysis, which was explained by oxidative damage to the membranes \[[@B20]\]. Note that MCV increases with increasing doses of hemolytic chemicals \[[@B21]\]. Future studies might answer the question whether particle-induced oxidative stress caused the association observed between MCV and the \"speed-change\" source factor.
The heart beat interval MCL increased in association with the \"crustal\" factor and the \"speed-change\" factor. Additionally, the \"speed-change\" factor was associated with significant increases in heart rate variability (SDNN and PNN50) and frequency of supraventricular ectopic beats. This cardiac response suggests increased vagal tone mostly in response to \"speed-change\" traffic particles. Fluctuations in autonomic tone have been associated with the triggering of atrial arrhythmias \[[@B22]\]. Such fluctuations might also help explain the reported association between air pollution exposure and increases in arrhythmias in patients with an implanted cardioverter defibrillator \[[@B6]\].
The concentrations of particles and components in this study were low. Direct systemic effects seem therefore unlikely. However, the proposed endothelial activation could provide a link to pathological processes and the associated increase in cardiovascular morbidity and mortality \[[@B7]\], as follows: Once particles are deposited on the surfaces of the airways or alveoli, toxic products can quickly leach out or be produced on the surface of the particles. Given the small volume of surface liquid, this can result in high local concentrations. Copper and other transition metals can cause oxidative stress \[[@B23]\] and have been associated with inflammatory lung injury in human subjects \[[@B24]\] as well as airway epithelial cell injury in vitro \[[@B25]\]. This oxidative stress might induce responses in the adjacent cells. In the alveolar region, the distance to the capillary endothelium is about 100 nanometers. Liberation of pro-thrombotic and pro-inflammatory mediators are well-described consequences of oxidative stress to endothelial and other cells \[[@B26]\]. Inflammatory stimuli also might induce a vagal response \[[@B27]\]. The components copper, sulfur and aldehydes dominated the \"speed-change\" factor. They seem to merit further attention in future targeted studies on particle toxicology.
Surprisingly, the \"steel wear\" factor of Model A was not associated to any inflammatory markers, although metal content of particles has been reported to be associated with inflammatory processes \[[@B24],[@B25],[@B28]\]. It would be interesting to study such wear particles with regard to size and solubility of metals.
One limitation of this study is the fact that only the association between the mean exposure during the evening shift and the response on the following morning was studied. This design ensured that potential diurnal variations of exposure and health parameters could not mimic a dose-response association, and that the exposure inside the cars was followed by a long unexposed resting period. However, it cannot be excluded that exposures and follow-up at other times of the day could have resulted in different dose-response estimates. Another limitation is the study population, since the troopers were a homogenous group of young, healthy, non-smoking people in excellent physical condition. Consequently, it is possible that the relative response such as the %-increase of inflammatory blood components or ectopic heart beats might be different in the troopers as compared to what could be expected in the general population or in individuals with elevated cardiovascular risks that have higher baseline levels. A final limitation is that the source factor \"speed-changing\" traffic does not represent a single source but rather a combination of closely related sources such as break wear and engine exhaust products. Answering the question, which of these sub-sources was causing the effects, would require a larger number of subjects or targeted toxicological studies.
Conclusions
===========
Fine particulate matter from vehicular traffic may activate one or more signaling pathways that cause pro-inflammatory, pro-thrombotic and hemolytic responses in healthy young men. The changes in the heart rate variability suggest an increased parasympathetic input to the heart with an associated increase in arrhythmic events, possibly in response to mild lung inflammation. These findings suggest the hypothesis that pollutants emitted during speed-changing traffic conditions negatively impact the health risks of professional or otherwise frequent vehicle drivers and passengers, or other people exposed to these particles. A long-term cardiovascular risk to the troopers can not be excluded, especially when considering the reported increase in myocardial infarction among professional drivers \[[@B29]\] and the increase in mortality among people living near major roadways \[[@B9]\]. These findings might be helpful for designing targeted studies in the future that investigate causative pathways for health effects of PM~2.5~.
Methods
=======
The study was conducted in fall 2001 in Wake County, North Carolina, USA. The Institutional Review Board of the UNC School of Medicine approved the study. All subjects gave informed written consent. Data from nine non-smoking male Highway Patrol troopers were analyzed. Each was monitored from Monday to Thursday while working the 3 PM to midnight shift. The troopers refrained from alcohol, caffeine and any medication from 24 hours before the start until the end of their participation. Each patrol car was equipped with air quality monitors to measure their exposure during the shift as described earlier \[[@B15]\]. Particle mass was assessed by two methods: PM~2.5Mass~by weighing filters; and PM~2.5Lightscatter~based on lightscattering. \"Aldehydes\" refers to the sum of formaldehyde, acetaldehyde, acrolein, propionaldehyde, crotonaldehyde, n-butyraldehyde, benzaldehyde, valeraldehyde, tolualdehyde, hexanaldehyde, and 2,5-dimethylbenzaldehyde.
Health parameters were assessed by ambulatory electrocardiography during the work shift and the subsequent sleep phase, and by analyses of peripheral blood samples drawn 15 hours after completion of the shift as described earlier \[[@B11]\]. Heart rate variability (HRV) measures in the time and frequency domain were calculated for resting periods before and after the shift, and in the morning after awakening. For the analysis presented, only data from the morning resting period were used. Parameters included the mean cycle length of normal R-R intervals (MCL), the standard deviation of normal R-R intervals (SDNN) and the percentage of normal R-R interval differences greater than 50 msec (PNN50), low frequency (0.04 to 0.15 Hz), high-frequency power (0.15 to 0.40 Hz) and the ratio of low to high frequency power. The number of ventricular and supraventricular ectopic beats were counted during the shift and the contiguous night.
Blood was collected from an antecubital vein and analyzed \[[@B11]\]. The analyses included uric acid, blood urea nitrogen, gamma glutamyl transpeptidase, white blood cell count, red blood cell count, hematocrit, hemoglobin, mean red blood cell volume (MCV), neutrophils (count and %), lymphocytes (count and %), C-reactive protein, plasminogen, plasminogen activator inhibitor type 1, von Willebrand factor (vWF), endothelin-1, protein C, and interleukin-6.
Statistical methods
-------------------
Spearman correlations were calculated using SYSTAT 10 (Systat Software Inc., Richmond, CA), all other statistics using S-Plus 6.1 for Windows (Mathsoft Inc., Cambridge, MA).
For classification of the exposure by potential sources, a principal factor analysis (factanal procedure) with variance-maximizing rotation \[[@B30]\] was conducted after controlling for outliers and missing data. One silicon value measured inside a patrol car was an outlier, possibly due to a grain of sand. This value was replaced by an estimate based on the aluminum level (Al was highly correlated to Si). Missing data (3 values of benzene and 2 of aldehydes) were replaced by the mean of the component concerned. For data below the propagated detection limit, machine-readouts were used. Factors with sum of squares of factor loadings larger than one were retained. The number of exposure variables included was limited to obtain stable results with this relatively small number of individual samples: only variables with a clear association to PM~2.5~and with reasonable data quality were used. In a first model (\"Model A\") PM~2.5~-components, that were significantly correlated to either PM~2.5Mass~or PM~2.5Lightscatter~(Spearman Rho \> 0.3), and gaseous co-pollutants, that were strongly and significantly correlated (Rho \> 0.5) were included in the source factor analysis if at least 75% of the data were above reporting limit. A second model (\"Model B\") was calculated to assess the robustness of the source factor modeling and the associated health effects. Model B excluded PM~2.5~-components from the analysis with large uncertainties (Cr, Se and W with over 50% of data less than 3 sigma above background noise) or with weak correlation to PM~2.5~(Ca).
Mixed effects regression models with restricted maximum-likelihood estimation, exposure factors as fixed effects and an unconstrained variance-covariance structure with subjects as grouping factors were used to investigate the associations between exposure and health endpoints \[[@B11],[@B31]\]. Potential confounders were controlled for by including them into the models. Model testing included alternative, constrained variance-covariance structures (first-order autoregressive as well as linear and exponential spatial designs), and adding exposure factors to the random effects structure. None of these attempts improved the overall quality of the models judged by the Akaike Information Criterion, analysis of variance and residual analysis (distribution and autocorrelation). Consequently, only results from unconstrained models are reported.
List of abbreviations
=====================
HRV heart rate variability
MCL mean cycle length of normal R-R intervals
MCV mean red blood cell volume
PNN50 percentage of normal R-R interval differences greater than 50 msec
SDNN standard deviation of normal R-R intervals
VWF von Willebrand factor
Competing interests
===================
The authors declare that they have no competing interests.
Authors\' contributions
=======================
MR conceived of and lead the study, collected the samples, performed the statistical analysis and drafted the manuscript. RBD supervised the analyses of blood components and elemental PM~2.5~-components. TRG supervised the on-site health assessment and the coordination of the troopers. MCH analyzed the ambulatory electrocardiograms and calculated the HRV statistics. PAB participated in the study management and in the layout of the manuscript. RWW supervised the assessment of air pollutants. WEC evaluated the volunteering troopers, and supervised the heart data analysis. All authors participated in the study design, and reviewed and approved the final manuscript.
Acknowledgements
================
We thank the North Carolina State Highway Patrol for enabling this study, and the support staff of NCSHP, UNC Chapel Hill, U.S. EPA, all contractors, and the participating troopers.
This work has been funded by The United States Environmental Protection Agency under Cooperative Agreements CR-824195 and CR-829522 to the University of North Carolina at Chapel Hill, by contract 68-D-00-206 to ManTech Environmental Technology and by the Swiss National Science Foundation to MR. It has been subjected to Agency review and approved for publication but does not necessarily reflect EPA policy.
|
PubMed Central
|
2024-06-05T03:55:55.082907
|
2004-12-9
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074349/",
"journal": "Part Fibre Toxicol. 2004 Dec 9; 1:2",
"authors": [
{
"first": "Michael",
"last": "Riediker"
},
{
"first": "Robert B",
"last": "Devlin"
},
{
"first": "Thomas R",
"last": "Griggs"
},
{
"first": "Margaret C",
"last": "Herbst"
},
{
"first": "Philip A",
"last": "Bromberg"
},
{
"first": "Ronald W",
"last": "Williams"
},
{
"first": "Wayne E",
"last": "Cascio"
}
]
}
|
PMC1074350
|
Background
==========
In addition to sino-nasal cancer \[[@B1]\], exposure to wood dust has been shown to be associated with a wide variety of acute and chronic non-malignant respiratory health effects as well as eye irritation and dermatitis \[[@B2],[@B3]\]. However, the underlying mechanisms involved are not well understood and subject of controversial discussion. Although inflammatory markers were found in nasal and bronchoalveolar lavage fluid from wood-dust exposed individuals \[[@B4]-[@B6]\], other studies do not corroborate the hypothesis that inflammation plays a part in wood dust-induced airway obstruction \[[@B7]\]. Moreover, recent studies do not support the assumption that the complaints related to exposure to wood dust are IgE-mediated \[[@B8],[@B9]\].
In wood processing facilities, the proportion of respirable wood dust ranges from 6% to 75% of the total wood aerosol \[[@B2]\]. Respirable wood dust particles may deposit in the pulmonary alveoli and interact with alveolar macrophages, a cell type that plays an important role in phagocytosis and clearance of inhaled particulates. Upon interaction with noxious particles, alveolar macrophages can produce a broad spectrum of pro-inflammatory mediators, such as tumour necrosis factor-alpha (TNF-α) and macrophage inflammatory protein-2 (MIP-2) as well as reactive oxygen (ROS) and nitrogen species \[[@B10]-[@B13]\]. TNF-α is one of the pre-eminent cytokines that acts as an initiator of inflammatory processes in the lung \[[@B14]\]. The chemokine MIP-2 is known to mediate neutrophilic inflammatory responses in the lung \[[@B10],[@B15]\]. ROS have been shown not only to damage cells by peroxidizing lipids and disrupting DNA and proteins, but also to exert signaling functions and modulate gene transcription \[[@B16],[@B17]\]. Moreover, ROS are suggested to mediate the release of TNF-α and MIP-2 in alveolar macrophages exposed to noxious particles \[[@B18]\]. Interestingly enough, a recent study demonstrated that exposure to pine dust induced increased ROS production and caused cell death in both murine RAW 264.7 macrophages and human polymorphonuclear leukocytes \[[@B19]\].
Pine is one of the most extensively used wood species in the wood processing industry and several studies have shown that exposure to pine dust induced respiratory symptoms, reduced lung function, and asthma \[[@B3],[@B20]-[@B22]\]. Moreover, pine is one of the most common wood species used for heat treatment, one of the treatment processes for stabilization and preservation of wood. After heat-treatment, both physical and chemical properties of wood are changed \[[@B23]\].
This study aimed to investigate the effect of dust from untreated as well as from heat-treated pine on the production of TNF-α, MIP-2, and ROS by primary rat alveolar macrophages and to elucidate the role of oxidative stress in pine dust-induced cytokine production.
Methods
=======
Wood dust
---------
Dust from untreated pine (PD) and heat-treated pine (HPD) was obtained from the Kuopio Regional Institute of Occupational Health (Kuopio, Finland). Dusts were produced using a dust collecting face-grinding machine with 400-grit sanding paper. For particle size distribution analyses, wood dust specimens, gold-coated for 170 seconds with BAL-TEC SCD 005 Sputter Coater (BAL-TEC AG, Liechtenstein), were examined on a JEOL JSM-6400 scanning electron microscope (JEOL Inc., Peabody, MA) at an acceleration voltage of 20 kV. More than 1700 particles for each dust were analyzed from electron micrographs. More than 95% of wood dust particles from both pine and heat-treated pine had a diameter less than 5 μm (Table [1](#T1){ref-type="table"}). The endotoxin content in PD and HPD as analyzed with a LAL gel-clot assay (Charles River, Germany) was below the detection limit of 0.06 EU/ml. For experiments, pine dust was suspended in RPMI-1640 medium with 10% fetal calf serum, ultrasonicated, and vortexed.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Size distribution of wood dust particles
:::
Pine Heat-treated Pine
----------------------------- ------ -------------------
Number of particles counted 1796 1928
\< 1 μm (%) 66.9 65.6
1--5 μm (%) 29.6 30.0
5--10 μm (%) 3.1 3.3
10--20 μm (%) 0.4 0.8
20--50 μm (%) 0.1 0.1
\> 50 μm (%) 0 0.1
:::
Collection of Alveolar Macrophages
----------------------------------
Male Sprague-Dawley rats (Charles River, Sulzfeld, Germany) were anesthetized by an intraperitoneal injection of sodium pentobarbital (30 mg/KG body weight) and killed by exsanguination from the abdominal aorta. The lungs were lavaged ten times with 10 ml of sterile, non-pyrogenic phosphate-buffered saline solution (PBS; Serva, Heidelberg, Germany). The pooled samples were centrifuged at 300 g for 10 min, and the cell pellet was washed twice and re-suspended in RPMI 1640 (Seromed, Munich, Germany) supplemented with L-glutamine, gentamycin (0,16 mg/ml), and 10% heat-inactivated fetal bovine serum (FBS; Gibco BRL, Eggenstein, Germany). Total cell counts were assessed with a standard hemocytometer (Coulter Electronics, Krefeld, Germany). Air-dried cytocentrifuged smears served to identify the cellular populations after staining with May-Grünwald-Giemsa. The preparation of bronchoalveolar cells contained about 97--100% alveolar macrophages. Cell viability as determined by trypan blue exclusion was greater than 90%.
Treatment of cells
------------------
Alveolar macrophages were adjusted according to the differential cell counts to 2 × 10^6^cells/ml. Then, 100 μl-samples of cell suspension were plated to 96-well flat-bottomed cell culture plates (Nunclon Delta, Roskilde, Denmark), and incubated at 37°C in 5% CO~2~and 21% O~2~. After 2 h, non-adherent cells were removed by washing twice with RPMI 1640, and the adherent alveolar macrophages were covered with 100 μl of pine dust suspension at concentrations ranging from 5 to 200 μg/ml. As a negative control, 3-μm polystyrene microspheres (Polysciences, Eppelheim, Germany) were used at a concentration of 100 μg/ml. As a positive control, *Escherichia coli*LPS serotype 055:B5 purchased from Sigma Chemie (Taufkirchen, Germany) was used at a concentration of 100 ng/ml. In a separate series of experiments, alveolar macrophages were treated with 6 mM glutathione (GSH; Sigma-Aldrich, Steinheim, Germany) or 20 mM N-acetyl-cysteine (NAC; Sigma-Aldrich, Steinheim, Germany) for 30 min. Subsequently, the culture medium was replaced with 100 μl of pine dust suspension at a final concentration of 200 μg/ml. After 4 h incubation in the absence or presence of GSH (6 mM) or NAC (20 mM), supernatants were removed and stored at -20°C. There was no effect of either treatment on cell viability as measured by a LDH assay kit (Merck, Germany).
RT-PCR
------
Total cellular RNA was extracted from pine dust-exposed alveolar macrophages using a ribonuclease protection kit (Rneasy Kit, QIAGEN, Hilden, Germany). RT-PCR was performed as described previously \[[@B24]\]. The oligonucleotide primers (MWG-Biotech, Ebersberg, Germany) used were 5\'-TGC CTC AGC CTC TTC TCA TT-3\' and 5\'-TGT GGG TGA GGA GCA CAT AG-3\' (EMBL: RNTNFAA, AC: X66539) for TNF, 5\'-CAA TGC CTG ACG ACC CTA C-3\' and 5\'-CAG TTA GCC TTG CCT TTG TTC-3\' \[[@B25]\] for MIP-2, and 5\'-TCC CTC AAG ATT GTC AGC AA-3\' and 5\'AGA TCC ACA ACG GAT ACA TT-3\' \[[@B26]\] for the housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The sizes of the PCR products were 376 bp for TNF, 194 bp for MIP-2, and 309 bp for GAPDH. PCR products were visualized in 2% agarose gels containing 1% ethidium bromide. For densitometric analyses, BIO-1D V 96 software (Vilber Lourmat, Marne La Vallee, France) was used.
TNF-α and MIP-2 ELISA
---------------------
Concentrations of TNF-α and MIP-2 in culture supernatants were determined by enzyme-linked immunosorbent assay (ELISA) using commercially available kits (Biosource, Solingen, Germany).
Detection of intracellular ROS
------------------------------
To detect intracellular ROS, 2\',7\'-dichlorofluorescin diacetate (DCFH-DA) (MoBiTec, Göttingen, Germany) was used. DCFH-DA diffuses into the cell and is hydrolyzed by intracellular esterases to polar 2\',7\'-dichlorofluorescin. This non-fluorescent fluorescin analogue can be oxidized to highly fluorescent 2\',7\'-dichlorofluorescein by intracellular oxidants \[[@B27]\]. Alveolar macrophages were cultured to adhere and incubated with 10 μM DCFH-DA for 30 min. The cultures were washed twice with RPMI 1640 and subsequently treated as described before. Baseline fluorescence was measured with a fluorometer (FLUOstar, BMG LabTechnologies, Offenburg, Germany) immediately after wood dusts were added. After 4 h of incubation under 37°C in 5% CO~2~and 21% O~2~, fluorescence was measured again. The results are shown as percentage change from baseline values. The addition of 1 μM H~2~O~2~served as an internal positive control.
Electron spin resonance spectroscopy
------------------------------------
Hydroxyl radical formation by wood dusts was assessed by electron spin resonance (ESR) spectroscopy, as described previously \[[@B28]\]. Briefly, wood dust suspensions (20 mg/ml) were prepared in pure water. 100 μl of this suspension was mixed with 200 μl of the spin trap 5,5-dimethyl-1-pyrroline-N-oxide (DMPO, 0.05 M in PBS) (Sigma, St. Louis, MO) and 100 μl H~2~O~2~(0.5 M in PBS) (Fluka, Seelze, Germany). The suspension was incubated for 15 min at 37°C in a shaking water bath, and filtered through a 0.2 μm filter (15 mm syringe filter, Satorius AG, Goettingen, Germany) to remove particles from the suspension. The filtrate was immediately transferred to a capillary and measured with a Miniscope ESR spectrometer (Magnettech, Berlin, Germany). The antioxidant activity of wood dust suspensions was measured by using the stable spin label TEMPOL (Sigma, Steinheim, Germany). TEMPOL was added to wood dust suspensions (10 mg/ml) at a final concentration of 5 μM, mixed and incubated at 37°C for 1 hour in a shaking water bath. After filtering the suspension through a 0.2 μm filter, the filtrate was measured as mentioned above. ESR spectra were recorded at room temperature using the following instrumental conditions: Magnetic field: 3360 G, sweep width: 100 G, scan time: 30 sec, number of scans: 3, modulation amplitude: 1.8 G, receiver gain: 1000. Quantification was carried out as the sum of total amplitude on first derivation of ESR signal, and outcomes are expressed as the total amplitude in arbitrary units.
Statistical analysis
--------------------
Results are presented as mean ± SEM. Statistical comparisons were performed by using RM ANOVA with Student-Newman-Keuls method for multiple comparison procedures. A p value \< 0.05 was considered significant.
Results
=======
TNF-α and MIP-2 mRNA expression
-------------------------------
After 4 h exposure of alveolar macrophages to PD and HPD, mRNA was extracted and the supernatants were collected for cytokine and chemokine measurement. A low, basal level of TNF-α and MIP-2 mRNA expression was observed in control macrophages. Compared to control, TNF-α and MIP-2 mRNA expression in alveolar macrophages exposed to PD and HPD was significantly increased. Interestingly, PD induced significantly (p \< 0.05) higher levels of TNF-α and MIP-2 mRNA expression than HPD (Figure [1](#F1){ref-type="fig"}).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
TNF-α and MIP-2 mRNA expression in alveolar macrophages after exposure to PD and HPD (100 μg/ml) for 4 h. (A) RT-PCR products of GAPDH, TNF-α and MIP-2 in an ethidium bromide stained agarose gel. Data shown are from four representative experiments. GAPDH were used as normalization control. (B) Densitometric analysis of 4 gels. \*p \< 0.05 compared with untreated control. \#p \< 0.05 compared with HPD.
:::
:::
TNF-α and MIP-2 release
-----------------------
As shown in Figure [2](#F2){ref-type="fig"}, exposure of alveolar macrophages to both PD and HPD elicited a significantly (p \< 0.05) increased production of TNF-α and MIP-2 when compared to untreated control cells. This effect was concentration-dependent and already observed at the lowest concentration of 5 μg/ml. Moreover, PD induced a 1.3--2.8 fold (p \< 0.05) higher release of TNF-α and MIP-2 than HPD. Polystyrene microspheres at a concentration of 100 μg/ml, which served as a negative control, did not induce an increased production of TNF-α and MIP-2. In contrast, exposure to LPS at a concentration of 100 ng/ml, which served as a positive control, induced a strong release of both mediators (data not shown).
::: {#F2 .fig}
Figure 2
::: {.caption}
######
TNF-α and MIP-2 protein release after exposure of alveolar macrophages to PD, HPD, and polystyrene microspheres (MS) assayed with ELISA kits. Alveolar macrophages were incubated with 5, 25, 100, or 200 μg/ml dust for 4 h. (A) TNF-α release, (B) MIP-2-release. Values represent means ± SEM of six separate experiments performed in duplicate. \*p \< 0.05 compared with medium alone. \#p \< 0.05 compared with HPD.
:::
:::
ROS generation
--------------
To detect ROS production in PD- and HPD-stimulated alveolar macrophages, the oxidant-sensitive dye DCFH-DA was used. After 4 h incubation, wood dusts at a concentration of 200 μg/ml induced a significantly (p \< 0.05) increased ROS generation when compared to untreated control cells (Figure [3](#F3){ref-type="fig"}). However, the level of PD-induced ROS generation in alveolar macrophages was not statistically different from the level of HPD-induced ROS generation. Treatment of the cells with GSH (6 mM) or NAC (20 mM) caused significant suppression of both PD- and HPD-induced ROS generation.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Production of intracellular ROS in alveolar macrophages after 4 h of stimulation with 200 μg/ml pine dust in the absence and presence of GSH (6 mM) or NAC (20 mM). Values are mean ± SEM of 6 experiments performed in duplicate. \*p \< 0.05 compared with untreated control. \#p \< 0.05 compared with PD or HPD without GSH or NAC treatment.
:::
:::
Effect of antioxidants on cytokine and chemokine expression
-----------------------------------------------------------
To elucidate whether oxidative stress participates in the up-regulation of inflammatory cytokine expression, TNF-α and MIP-2 release was examined in PD- and HPD-exposed alveolar macrophages in the presence or absence of the antioxidants GSH and NAC. Treatment with both GSH and NAC significantly (p \< 0.05) reduced the TNF-α and MIP-2 release elicited by the exposure of alveolar macrophages to PD and HPD (Figure [4](#F4){ref-type="fig"}).
::: {#F4 .fig}
Figure 4
::: {.caption}
######
TNF-α and MIP-2 protein release of alveolar macrophages after 4 h of stimulation with 200 μg/ml pine dust in the absence and presence of GSH (6 mM) or NAC (20 mM). (A) TNF-α release, (B) MIP-2 release. Values represent means ± SEM of six separate experiments performed in duplicate. \*p \< 0.05 compared with PD or HPD without GSH or NAC treatment.
:::
:::
Endogenous oxidant and antioxidant activity of pine dust
--------------------------------------------------------
ESR spectroscopy showed that suspensions of both PD and HPD caused formation of •OH in the presence of H~2~O~2~. However, the ability of PD and HPD to generate •OH was not statistically different (Figure [5](#F5){ref-type="fig"}). The antioxidant capacity of pine dust suspensions was measured by the use of the stable spin label TEMPOL. Interestingly, HPD caused a significantly greater reduction of TEMPOL than PD, indicating that HPD has greater antioxidant capacity than PD (Figure [5](#F5){ref-type="fig"}).
::: {#F5 .fig}
Figure 5
::: {.caption}
######
Hydroxyl radical formation (A) and antioxidant capacity (B) of PD and HPD as measured by ESR spectroscopy. Data are from three representative experiments.
:::
:::
Discussion
==========
A higher prevalence of non-malignant respiratory diseases, such as bronchitis, chronic obstructive pulmonary disease, cryptogenic fibrosing alveolitis, and asthma has been reported in workers exposed to a variety of wood dusts \[[@B2]\]. Sensitization to wood dust from some wood species such as red cedar has been shown to be involved in mechanisms generating a work-related asthmatic response \[[@B29]\]. However, more recent studies have shown that sensitization to wood dust from pine, oak, beech and other wood species may not be the only or even the most important mechanism involved in wood dust-induced respiratory symptoms \[[@B8],[@B9]\]. Therefore, our study aimed to investigate the non-specific inflammatory response of primary lung macrophages to wood dust from pine, one of the most extensively used wood species in the wood processing industry.
Here we show that pine dust induces TNF-α and MIP-2 mRNA expression as well as TNF-α and MIP-2 protein release in rat alveolar macrophages. Alveolar macrophages are important in processing airborne particles and play a key role in mediating inflammatory responses of the lung through the release of various proteolytic enzymes, reactive oxygen and nitrogen species, arachidonic acid metabolites, cytokines such as TNF-α, and chemokines such as MIP-2 \[[@B11]\]. TNF-α plays an important role as a mediator of the respiratory tract\'s response to particles. Studies have shown that a variety of agents which elicit marked lung inflammation can activate alveolar macrophages to release TNF-α, while agents with limited inflammatory activity do not stimulate macrophage TNF-α production. MIP-2 plays a major role in mediating the neutrophilic inflammatory response of the rodent lung to particles such as quartz and crocidolite asbestos \[[@B10]\]. TNF-α and MIP-2 gene expression is under the control of redox-sensitive inflammation-related transcription factors such as NF-κB. Activation of NF-κB is regulated via a number of second messengers, including calcium and ROS \[[@B16]\]. In addition to providing evidence that pine dust stimulates both TNF-α and MIP-2 mRNA production and protein release from rat alveolar macrophages, our study clearly demonstrates that pine dust stimulates the generation of ROS in alveolar macrophages, as previously shown in mouse macrophages and human leukocytes by Naarala et al. \[[@B19]\].
To investigated the role of oxidative stress in pine dust-induced cytokine and chemokine response we treated pine dust-exposed alveolar macrophages with the antioxidants GSH and NAC. GSH plays a major role in the antioxidant system by working as a substrate for glutathione peroxidase, and it has been previously shown that extracellular GSH can elevate intracellular GSH levels and protect phagocytes against oxidant damage \[[@B30]\]. NAC is a thiol compound that can act as a cysteine source for the repletion of intracellular glutathione and act as a direct scavenger of ROS. NAC has been shown to attenuate oxidant-mediated toxicity induced by chrysotile fibres in rats \[[@B31]\] and to down-regulate the nitric oxide pathway in alveolar macrophages \[[@B22]\]. We found that treatment with GSH or NAC attenuated pine dust-induced ROS generation as well as TNF-α and MIP-2 protein release. These findings are concordant with previous studies on silica and ultrafine particles \[[@B18],[@B33],[@B34]\] and indicate that pine dust-induced oxidative stress mediates, at least in part, the expression of TNF-α and MIP-2 in alveolar macrophages.
Interestingly enough, dust from untreated pine (PD) induced a significantly stronger inflammatory response in alveolar macrophages than dust from heat-treated pine (HPD). Consequently, we used ESR spectroscopy to assess the endogenous oxidative and antioxidant capacity of the wood dusts under study. Whereas the ability to generate hydroxyl radical did not differ among PD and HPD, HPD exhibited greater antioxidant capacity than PD. As we have demonstrated in this study that oxidative stress may play a role in mediating the expression of TNF-α and MIP-2, we suggest that the greater antioxidant capacity of HPD may neutralize oxidative stress and thus attenuate expression of TNF-α and MIP-2. As mentioned before, both physical and chemical properties of wood are changed when heat-treated for several hours with temperatures up to 230°C. In particular, the pine resin is easily volatilized and almost completely removed from the wood \[[@B23]\]. One of the components of resin, δ-3-carene, has been reported to decrease the viability of alveolar macrophages and affect the engulfment of particles *in vitro*\[[@B35]\]. Another component of pine resin, abietic acid, has been shown to produce lytic damage to alveolar, tracheal, and bronchial epithelial cells \[[@B36]\]. Further studies are warranted to confirm our results and to determine the specific chemical and physical properties of dust from heat-treated pine that might be responsible for the effects seen in this study. Recently, metabolites of pine bark extract have been shown to have antioxidant activity and to inhibit matrix metalloproteinases (37).
In summary, our findings indicate that non-specific inflammatory reactions, mediated *via*ROS production, may play a role in pulmonary effects of wood dust. However, it is not clear from this *in vitro*study whether the oxidative stress driving TNF-α and MIP-2 protein release is due to ROS derived directly from the dust particles or from cell-generated ROS.
Conclusions
===========
Here, we demonstrate that pine dust is able to induce inflammatory responses *in vitro*. Oxidative stress seems to play an important role in the pine dust-induced cytokine and chemokine response, suggesting that wood dust particles may exert pro-inflammatory effects by a mechanism that is, at least in part, mediated by ROS.
Authors\' contributions
=======================
HL performed the isolation of alveolar macrophages, subsequent cytological and biochemical analyses, and writing and preparation of the manuscript. TS and PJB carried out ESR spectroscopy. JM analysed the size distribution of wood dust particles. KHP, KS, and FK participated in the direction of the study as well as in writing and preparation of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
================
We gratefully acknowledge the excellent technical assistance of Anne-Marie Allmeling, University of Munich, and Jaakko Säntti, Finnish Institute of Occupational Health. We thank Irma Welling, Lappeenranta Regional Institute of Occupational Health, Finland, for help in obtaining the two types of pine for production of wood dust. This study was supported by the EU 5th Framework Programme, Key Action 4, Environment and Health, Quality of Life and Management of Living Resources, Project No. QLK4-2000-00573.
|
PubMed Central
|
2024-06-05T03:55:55.085446
|
2004-12-13
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074350/",
"journal": "Part Fibre Toxicol. 2004 Dec 13; 1:3",
"authors": [
{
"first": "Huayan",
"last": "Long"
},
{
"first": "Tingming",
"last": "Shi"
},
{
"first": "Paul J",
"last": "Borm"
},
{
"first": "Juha",
"last": "Määttä"
},
{
"first": "Kirsti",
"last": "Husgafvel-Pursiainen"
},
{
"first": "Kai",
"last": "Savolainen"
},
{
"first": "Fritz",
"last": "Krombach"
}
]
}
|
PMC1074351
|
The word pain is thought to derive from the Latin word poena, meaning punishment. An emotional reaction to a punishment might have been what Aristotle experienced, as he defined pain as an emotional event. René Descartes, the seventeenth-century philosopher and scientist, pictured a pain pathway consisting of a thread with two ends: one end is in a peripheral part of the body, for example a toe, and the other end is a bell in the brain. According to this picture, fire touching a toe pulls the thread, and rings the bell to sound a warning in the brain. Over the past decades, and in the current Decade of Pain Control and Research (2001--2010), pain research has undergone major changes, from a system level to cellular, subcellular and molecular levels. A new era of molecular pain research is now emerging, and the journal *Molecular Pain*is dedicated to this modern phase of pain research.
Recent advances in pain research are in large part due to the rapid progress in neuroscience, molecular biology, and other fields in the life sciences. Breakthroughs in biomedical technologies have allowed us to address many important issues about pain, enriching our knowledge about the mechanisms by which sensory signals including pain are initiated, encoded, conducted, transmitted, modulated, and perceived. For example, sensory molecular biology has led to the molecular cloning and identification of a number of receptors involved in thermal, mechanical, and nociceptive signalling at the periphery, some of which have been targeted for pain management. Modern electrophysiology has been used to demonstrate the critical roles of synaptic plasticity in pain processing in the spinal cord and the brain. Long-term potentiation and long-term depression at synapses of central sensory regions have delineated the \'memory of pain\' by neuronal circuitry along pain transmitting pathways. Functional imaging of supraspinal areas has revealed central areas related to pain processing (for example, areas coding behavioural learning and memory) and, more significantly, it has now become possible to see the alteration of these signalling pathways under chronic pain conditions. Finally, genomics and proteomics have been applied to pain research to help identify the changes in the array of molecules present in cells under chronic pain conditions. Research within all these fields will provide a better understanding of the physiological and pathological mechanisms of pain.
Pain research at the cellular, subcellular, and molecular levels has provided insights that help guide the treatment and management of intractable pain conditions including neuropathic pain, cancer pain and other chronic pain conditions. At the same time, these efforts continue to provide scientific insights into an inherently fascinating biological process. New molecules related to pain continue to be cloned and identified. It should be stressed that a \'non-pain molecule\' under physiological conditions can become a \'pain molecule\' under pathological conditions. This might be an underlying mechanism for spontaneous pain or pain sensation elicited by innocuous stimuli. Thus, the aims of molecular pain research should include \'non-pain molecules\'. One big task facing us is that many of the pain-related genes or proteins that have been identified are also important for other neuronal functions in the spinal cord and the brain. Pain triggers various responses in the spinal cord and the brain, including reflexes, conscious perception, cognitive learning and memory processes, emotional reaction such as depression, and drug addiction. Thus, molecules that are associated with pain are not only those located on the peripheral nerve endings for the sensing and encoding of stimuli, but also molecules that are present along sensory paths from the spinal cord to the brain for integrating and modulating sensory information. Molecular targets at different levels along sensory pathways are key to future identification of new drugs and therapies that effectively manage intractable pain conditions with low side effects. Molecular pain research will offer new opportunities for drug development in the pharmaceutical industry and improved treatment options in the clinical setting.
In our journey to explore pain mechanisms and to identify effective targets for pain management, it is important for scientists around the world to have a rapid and freely accessible forum for exchanging ideas, debating hot topics, developing collaborations, promoting science, and improving pain medicine. As an online, Open Access journal, *Molecular Pain*will help to fulfil these goals. The journal\'s Open Access policy changes the way in which articles are published. Firstly, all articles become freely and universally accessible online, and so an author\'s work can be read by anyone at no cost. Secondly, the authors hold copyright for their work and grant anyone the right to reproduce and disseminate the article, provided that it is correctly cited and no errors are introduced. Thirdly, a copy of the full text of each Open Access article is permanently archived in an online repository separate from the journal. *Molecular Pain*\'*s*articles are archived in PubMed Central (<http://www.pubmedcentral.org/>), the US National Library of Medicine\'s full-text repository of life science literature, and also in repositories at the University of Potsdam (<http://www.uni-potsdam.de/over/homegd.htm>) in Germany, at INIST (<http://www.inist.fr/index_en.php>) in France and in e-Depot (<http://www.kb.nl/%5d>), the National Library of the Netherlands\' digital archive of all electronic publications.
The launch of *Molecular Pain*would not have been possible without the strong support of many neuroscientists who have carried out pain-related research for many years. *Molecular Pain*has a strong editorial board with wide expertise in pain-related research, from peripheral sensory receptors to cortical sensory processing centers. *Molecular Pain*\'s editorial board members have committed to support the journal by helping identify important and interesting research manuscripts, serving as reviewers, and directly contributing their work to the journal. We aim to publish papers in a timely fashion. Each manuscript will be peer-reviewed by two experts, and the review process is anticipated to be completed within three weeks. Once accepted, papers will be published online immediately, and they will be listed in PubMed as soon as possible after publication. We hope that *Molecular Pain*will become a high impact journal, a journal that provides new directions for pain research and medicine, and a home for creative scientists. We welcome researchers and clinicians as readers of and contributors to, *Molecular Pain*.
Competing interests
===================
The author(s) declare that they have no competing interests.
|
PubMed Central
|
2024-06-05T03:55:55.087775
|
2005-1-14
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074351/",
"journal": "Mol Pain. 2005 Jan 14; 1:1",
"authors": [
{
"first": "Jianguo",
"last": "Gu"
},
{
"first": "Min",
"last": "Zhuo"
},
{
"first": "Michael",
"last": "Caterina"
},
{
"first": "Amy B",
"last": "MacDermott"
},
{
"first": "Annika",
"last": "Malmberg"
},
{
"first": "Volker",
"last": "Neugebauer"
},
{
"first": "Megumu",
"last": "Yoshimura"
}
]
}
|
PMC1074352
|
Background
==========
General anesthesia consists of five distinct components: analgesia, amnesia, unconsciousness, immobility, and blunted autonomic responsiveness \[[@B1],[@B2]\]. While the spinal cord is considered to be the anatomic substrate for anesthetic-induced immobility in response to noxious stimulation \[[@B3],[@B4]\], the anatomic foundations for the other components are less well established. The thalamus is a key integrative structure for somatosensory transmission \[[@B5]\] and, in particular, ascending nociceptive information processing \[[@B6],[@B7]\].
Excitatory input regulates the functional state of thalamic neurons, and such input is provided by both ascending activating systems in the brain stem and hypothalamus and the descending (corticothalamic) pathway \[[@B8]\]. Corticothalamic axons outnumber thalamocortical axons by \~10-fold \[[@B9]\], and activation of this massive descending input depolarizes thalamic neurons, including thalamocortical relay neurons in the ventrobasal (VB) complex, facilitates relay spike transfer, and/or alters the response mode of thalamic relay neurons \[[@B10]-[@B18]\]. Inhibitory control of thalamocortical neurons in rodents is provided exclusively by GABAergic neurons in the reticular thalamic nucleus \[[@B8],[@B19]\], and such control is mediated by disynaptic (cortex to RTN to VB) and monosynaptic (RTN to VB) connections.
Propofol (2-6-di-isopropylphenol) is a widely used intravenous anesthetic with a distinct chemical structure, and is a potent allosteric modulator of GABA~A~receptors \[[@B20],[@B21]\]. Recent clinical findings have revealed possible sites of propofol-elicited anesthetic action in the human brain \[[@B22],[@B23]\]. During propofol-induced unconsciousness in humans, somatosensory-evoked neuronal activity in the cortex and the thalamus is markedly decreased \[[@B24],[@B25]\]. *In vivo*extracellular recordings have also demonstrated that propofol suppresses field potentials in the rat thalamus and cortex, with more prominent effects in the cortex \[[@B26]\]. However, the cortical suppression may reflect anesthetic actions on projection neurons located elsewhere, especially in the thalamus \[[@B22],[@B27]\]. A significant limitation to the *in vivo*data from anesthetized animals is the use of \"background anesthesia\" (typically induced by urethane, sodium pentobarbital or a ketamine/xylazine combination) for baseline recordings; such \"background anesthesia\" makes it impossible to interpret the data subsequently obtained with the anesthetic(s) of interest \[[@B28]\].
Propofol modulates GABA-evoked currents in heterologously expressed GABA~A~receptors containing an α1, α2, α4, α5, α6, δ or γ2L subunit \[[@B29]-[@B37]\]. Behavioral studies suggest that the β3 subunit is important in mediating propofol-induced unconsciousness and immobility \[[@B38]\], while the β2 subunit may mediate sedation \[[@B39]\]. Propofol potentiation of GABA-evoked currents in heterologously expressed GABA~A~receptors is independent of the β1 subunit \[[@B29]\].
Cumulative data from a number of studies using a variety of techniques (including electrophysiology, gene knockout, immunohistochemistry, immunoprecipitation, and ligand binding) suggest that VB neurons primarily express synaptic α1β2γ2 and α4β2γ2 and extrasynaptic α4β2δ GABA~A~receptors while RTN neurons are likely to preferentially express synaptic α3β3γ2 GABA~A~receptors, with denser GABA receptor expression in VB than in RTN \[[@B40]-[@B58]\]. These data further support the hypothesis that the thalamus represents an important anatomic target for propofol.
The thalamus is central to the processing and transfer of nearly all sensory information that ultimately reaches the cortex, with the exception of olfaction, whose signals pass to the cortex without thalamic relay. Clinical observations strongly suggest that thalamic neuronal circuits are important targets for propofol. The effects of propofol at the cellular and synaptic levels in the thalamus are largely unknown, however. Therefore, we investigated the effect of propofol on synaptic integration and action potential firing in response to corticothalamic pathway stimulation in thalamocortical relay neurons in brain slices, using both current- and voltage-clamp recording techniques. The results demonstrated that propofol inhibited VB neurons by potentiating GABA~A~-receptor chloride channel-mediated currents. Preliminary results have been published in abstract form \[[@B59]\]
Results
=======
Under low power magnification, the VB was easily discerned in brain slices (Fig. [1A](#F1){ref-type="fig"}), and with the aid of IR-DIC optics, neurons in VB were readily identified (not shown). VB neurons generally showed a large, slow depolarizing sag and slow after-burst depolarization (ADP, Fig. [1B](#F1){ref-type="fig"}) in response to deep hyperpolarizing current steps \[[@B17]\]. Biocytin labeling in a subset of cells identified the recording site relative to anatomical landmarks (not shown).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Synaptic temporal summation in VB relay neurons**. ***A***: a photomicrograph of a live brain slice containing the ventrobasal (VB) complex, reticular thalamic nucleus (RTN), and other regions with which VB has synaptic connections. Thalamocortical slices were cut at 55° (Agmon and Connors, 1991). ***B***: both tonic and rebound burst firing patterns were initiated in a VB neuron with intracellular current pulses (protocol not shown). A distinct membrane voltage response is characterized by a prominent sag and after-depolarization potential (ADP) in response to hyperpolarizing current pulses. The value to the left of the trace indicates membrane potential (mV) here and throughout. ***C***: EPSPs showing temporal summation were evoked by extracellular stimulation of corticothalamic fibers in the white matter (a train of 5 pulses, 33 Hz, 0.15 ms). ***D***: evoked EPSPs could be blocked by CNQX/AP5. ***E***: group data demonstrating that EPSP temporal summation is frequency-dependent (n = 10).
:::
:::
Propofol suppresses temporal summation in VB neurons
----------------------------------------------------
The responses of VB neurons *in vivo*to somatosensory stimuli depend on the state of arousal, and the functional state is linked to neuronal depolarization levels that can be regulated by corticothalamic excitatory (CT) input \[[@B8]\]. CT excitatory synapses in VB exhibit prominent frequency-dependent summation \[[@B13],[@B16],[@B60]-[@B62]\]. We therefore examined the effect of propofol on CT-evoked temporal summation in VB neurons. Repetitive stimulation (33 Hz, 5 pulses) of the white matter gave rise to incremental excitatory postsynaptic potentials (EPSPs) that showed summation without apparent inhibitory postsynaptic potentials (IPSPs) at membrane potentials of -58 to -54 mV (Fig. [1C](#F1){ref-type="fig"}), identical to those seen by others \[[@B60],[@B63]\]. CT-evoked EPSPs had an average latency of 3.2 ± 0.3 ms (n = 55). In some cases, summation could lead to spike firing at the 5^th^EPSP (not shown); for ease of comparison, we only analyzed those responses without spikes. CT-evoked EPSPs could be abolished by CNQX and AP5 (Fig. [1D](#F1){ref-type="fig"}), consistent with frequency-dependent facilitation mediated by both NMDA and non-NMDA receptors \[[@B64]\]. The degree of CT EPSP summation increased with increasing stimulation frequency, with a dramatic increase (400--480%) in summation at 33--40 Hz (Fig. [1E](#F1){ref-type="fig"}).
We next examined the effect of propofol on temporal summation of EPSPs evoked at 33 Hz. Bath application of propofol (3 μM) hyperpolarized the MP by 3--5 mV, and markedly suppressed the magnitude of CT EPSP summation (102.4 ± 18.5%) in 15 relay neurons tested (Fig. [2A](#F2){ref-type="fig"}), and the degree to which propofol decreased summation was concentration-dependent (Fig. [2D](#F2){ref-type="fig"}). Propofol also decreased EPSP amplitude by 30--75% and input resistance by 35.2 ± 4.1% (P \< 0.05, Fig. [2B](#F2){ref-type="fig"}), consistent with shunting inhibition. In addition, small IPSPs were evoked during propofol application (Fig. [2](#F2){ref-type="fig"}, *middle*).
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Propofol suppresses temporal summation via a shunting mechanism**. ***A***: (*left*) control EPSP summation was recorded as described for Fig. 1C. Bath application of propofol (3 μM, 20 min) markedly decreased EPSP amplitude and summation (*middle*). Overlay (*right*) comparing EPSPs in the absence and presence of propofol. ***B***: overlay (*left*) comparing voltage responses elicited by a hyperpolarizing current pulse (-60 pA, 500 ms, not shown) in the absence (control, thin line) and presence of propofol (thick line) in the same neuron as in ***A***. Note that propofol decreased apparent input resistance. Bar graph (*right*) showing that propofol significantly decreases input resistance. \*: P \< 0.05, n = 15. Time scales: 100 ms for ***A***and 300 ms for ***B***. ***C***: control EPSPs were recorded from a different VB neuron. Overlay (*left*) comparing EPSPs in the absence (control) and presence of the GABA~A~antagonist bicuculline (10 μM). Overlay (*right*) comparing EPSPs in the presence of bicuculline alone and bicuculline + propofol (3 μM). In the presence of bicuculline, propofol had no significant effect on temporal summation, indicating propofol-elicited shunting inhibition was mediated by GABA~A~receptors in VB neurons. ***D***: group data showing that propofol decreases temporal summation through potentiation of GABA~A~receptors. \*: P \< 0.05, one-way ANOVA, *vs*. control. n = 15/each group. prop = propofol (0.6, 0.3 μM), bic = bicuculline (10 μM).
:::
:::
To determine whether propofol-modulation of GABA~A~receptors contributed to the decrease in summation and input resistance, the experiments were repeated in the presence of GABA~A~receptor chloride channel blockade. In another group of cells (n = 10), bicuculline (10 μM) alone had no significant effect on CT EPSP summation evoked by the same repetitive stimulation as above (Fig. [2C](#F2){ref-type="fig"} left panel), and a similar response was observed for picrotoxin (100 μM, not shown). Our data were nearly identical to those observed by others \[[@B60]\]. The failure of either GABA~A~-R blocker to markedly increase temporal summation was likely due to the fact that disynaptic inhibition generated within the cortex-RTN-VB circuit was markedly reduced during repetitive stimulation, which resulted in excitatory response only \[[@B60]\]. Propofol, when co-applied with bicuculline, failed to decrease temporal summation (Fig. [2C--D](#F2){ref-type="fig"}), indicating that GABA~A~receptor-mediated shunting inhibition was involved. The effects of propofol on integrative properties of synaptic responses are summarized in Table [1](#T1){ref-type="table"}.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Effects of propofol on integrative properties of evoked EPSPs in thalamic VB neurons.
:::
Slope (mV/ms) 1/2 width (ms) Decay time (ms)
---------------- --------------- ---------------- -----------------
control 1.8 ± 0.4 14.5 ± 2.6 13.2 ± 2.8
propofol 1.6 ± 0.4 6.3 ± 0.4\*\* 16.3 ± 3.2\*
\+ bicuculline 1.9 ± 0.5 13.5 ± 2.1 12.2 ± 2.4
EPSPs are evoked by extracellular stimulation of corticothalamic fibers (a train of 5 pulses, 0.15 ms, 33 Hz). Data are derived from experiments as described in Fig. 2. Slope (dv/dt) representing depolarization rate is measured from the rising phase (20--80%) of the first EPSP. 1/2 width is measured at 50% of peak amplitude; decay time (20--80%) representing repolarization rate is measured from the falling phase of the fifth EPSP. \*, P \< 0.05. \*\*, P \< 0.001, *vs*. control.
:::
Propofol decreases spike firing evoked at corticothalamic synapses in VB
------------------------------------------------------------------------
As shown above, propofol suppressed glutamatergic excitatory subthreshold responses (EPSPs) *via*modulation of GABA~A~receptor chloride channels. Such suppression might reduce spike generation in response to excitatory synaptic input. We therefore investigated whether propofol could affect corticothalamic-evoked spike firing. VB neurons were held at depolarized membrane potentials (-51 to -48 mV); stimulation (10--15 Hz) of CT fibers evoked single spike firing in 25 neurons tested (Fig. [3A](#F3){ref-type="fig"}) with an average latency of 4.2 ± 1.1 ms, and evoked spikes could be blocked by TTX (not shown). These observations are consistent with other reports \[[@B13],[@B60]\]. Bath application of propofol significantly decreased the number of evoked spikes in a concentration-dependent manner (Fig. [3B](#F3){ref-type="fig"} and [3D](#F3){ref-type="fig"}). The effect of propofol could be completely blocked by bicuculline (Fig. [3C](#F3){ref-type="fig"}), confirming the involvement of GABA~A~receptors.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Propofol decreases corticothalamic-evoked spike firing in VB neurons**. ***A***: action potentials were evoked by corticothalamic (CT) stimulation in a VB neuron under control conditions. ***B***: bath-application of propofol (0.6 μM) suppressed spike generation, and this suppression could be blocked by 10 μM bicuculline (***C***). Propofol was applied 20 min before the trace in (**B**) was recorded. ***D***: bar graph indicates that propofol suppressed successful synaptic transmission (% suppression) at CT synapses in VB in a concentration-dependent manner. \*, P \< 0.05, one-way ANOVA, *vs*. control, n = 15.
:::
:::
Propofol inhibits tonic firing by increasing GABAergic input
------------------------------------------------------------
Thalamocortical neurons *in vivo*fire tonic single spikes during the waking state or in response to excitatory synaptic stimuli \[[@B8]\]. In brain slices, however, these neurons generally do not fire spontaneously; a sustained single spike firing pattern can be induced by pharmacological activation of metabotropic glutamate receptors (mGluRs) with *trans*-ACPD \[[@B63]\]. Since propofol at concentrations less than 10 μM does not affect glutamatergic transmission \[[@B65]-[@B70]\], we tested whether propofol could inhibit *trans*-ACPD-induced spike firing through GABAergic mechanisms. Bath application of *trans*-ACPD (100 μM) gradually depolarized the membrane of VB relay neurons by 14 ± 4 mV (n = 12) from resting membrane potentials, and resulted in the generation of sustained, tonic spike firing (Fig. [4A](#F4){ref-type="fig"}*top*), an effect similar to that seen in the dorsal lateral geniculate nucleus of the thalamus \[[@B63]\]. After a stable tonic, firing pattern was obtained during ACPD application (\> 10 min), propofol was added (Fig. [4A](#F4){ref-type="fig"}*middle*). Propofol markedly decreased both firing rate, and apparent input resistance (P \< 0.05, n = 8, one-way ANOVA, vs. control, traces for measuring input resistance not shown).
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Propofol suppresses *trans*-ACPD-evoked tonic spike firing in VB neurons**. ***A***: bath application of the metabotropic glutamate receptor agonist *trans*-ACPD (100 μM), which can mimic corticothalamic excitatory transmission \[63\], induced a sustained, tonic spike firing pattern (*top*). Addition of propofol (3 μM) depressed firing (*middle*), accompanied by spontaneous IPSPs (sIPSPs). The suppression could be blocked by bicuculline (10 μM; *bottom*). ***B***: segments marked with \"*i*, *ii*, *iii*\" in ***A***are expanded to view sIPSPs. Spikes are truncated for clarity. Note that few sIPSPs are seen prior to propofol application (*i*), small sIPSPs (\< 6.5 mV, *ii*) and large sIPSPs (6.5 -- 16.5 mV, *iii*) are observed following propofol application. Scale: 20 mV, 200 ms for *i*, 20 mV, 100 ms for *ii*and *iii*. ***C***: time course histogram for group data (n = 8) showing propofol suppression of *trans*-ACPD-induced tonic spike firing rate. Arrows indicate the onset and duration of propofol application at given concentrations. The suppression of propofol could be blocked by bicuculline. SE bars (0.5 -- 2.2) are omitted for clarity. D: bar graph of cumulative data indicates that propofol (prop) hyperpolarized the membrane potential (MP) and this effect was reversed during addition of bicuculline (+bic). \*: P \< 0.01, one way ANOVA with Tukey test, propofol vs. control, propofol vs. propofol + bicuculline. n = 8 -- 12/each.
:::
:::
Although action potential firing in VB neurons was inhibited, RTN neurons continued to fire spikes, as evidenced by the presence of spontaneous IPSPs (sIPSPs, Fig. [4B](#F4){ref-type="fig"}). The sIPSPs appeared to result from propofol-induced activation of RTN neurons, rather than direct excitation of RTN by *trans*-ACPD, as few, if any, sIPSPs occurred prior to propofol application. The propofol-induced suppression could be blocked by subsequent addition of bicuculline (Fig. [4A](#F4){ref-type="fig"}*bottom*), indicating that the suppression of VB neuron spike firing was mediated by GABA~A~receptors. Group data demonstrated that propofol significantly suppressed tonic spike firing initiated by pharmacological activation of mGluRs in a concentration-dependant manner (Fig. [4C](#F4){ref-type="fig"}, P \< 0.001, n = 8, one-way ANOVA, vs. control). Propofol hyperpolarized the membrane potential, and the hyperpolarization persisted throughout propofol application (Fig. [4A](#F4){ref-type="fig"}, *middle*). This effect could be reversed by addition of bicuculline (Fig. [4A](#F4){ref-type="fig"}, *bottom*) or picrotoxin (not shown). Group data (Fig. [4D](#F4){ref-type="fig"}) demonstrated that propofol significantly hyperpolarized the membrane potential from -53.7 ± 1.6 to -65 ± 2.9 mV (P \< 0.01, same data set as above), strongly suggesting that propofol likely potentiated the tonic GABA~A~receptor-mediated current.
In another subgroup of cells (n = 4, not illustrated), bicuculline alone was added after ACPD induced the tonic firing rate, followed by co-application of propofol (3 μM) and bicuculline. Bicuculline alone increased the firing rate by 16.8 ± 3.2% and the addition of propofol failed to suppress the firing rate.
Propofol enhances GABA~A~receptor-mediated currents
---------------------------------------------------
Our data strongly suggested that propofol-elicited inhibition of VB neurons was primarily mediated through potentiation of the GABA~A~receptor Cl^-^channel current; this possibility was directly investigated using voltage-clamp recordings. Electrical stimulation of RTN evoked fast unitary IPSCs (eIPSCs) in relay neurons (the membrane potential clamped at -60 mV), with an average latency of 2.4 ± 0.8 ms (Fig. [5A](#F5){ref-type="fig"}). The eIPSC amplitude was 685 ± 28 pA (n = 30), and the two-exponential decay time (10--90%) was 16.2 ± 2.2 ms. Bath application of propofol increased current amplitude and prolonged current decay time (Fig. [5A--B](#F5){ref-type="fig"}), and the propofol-potentiated current could be completely abolished by picrotoxin (100 μM), indicating that the response was mediated by GABA~A~receptor chloride channels (Fig. [5C](#F5){ref-type="fig"}). Inhibitory efficacy of IPSCs can be estimated by calculating total Cl^-^charge transfer \[[@B44]\]. The efficacy of the eIPSC in control was 1.2 ± 0.8 pC, equivalent to 7.2 × 10^6^Cl^-^ion transfer. Group data demonstrated that propofol increased eIPSC amplitude, decay time, and charge transfer in a concentration-dependent manner (Fig. [5D--F](#F5){ref-type="fig"}).
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Propofol-potentiated IPSCs are mediated by GABA~A~receptor chloride channels**. ***A***: GABA~A~IPSCs were evoked in a VB relay neuron in a horizontal slice by RTN stimulation (50 μA, 180 μs, every 15 s) in the presence of the GABA~B~antagonist 2-OH saclofen (100 μM). The membrane potential was clamped at -60 mV. Synaptic currents were potentiated by propofol (3 μM, 10 min). Overlay showing current amplitude in the absence (control) and presence of propofol. Each trace is an average of 10 sweeps. ***B***: normalized traces showing that propofol prolonged the decay time of IPSCs. ***C***: propofol-potentiated eIPSCs could be abolished by picrotoxin (100 μM). ***D-F***: bar graphs of pooled data indicate that propofol increased eIPSC amplitude, decay time, and charge transfer in a concentration-dependent manner. \*: P \< 0.05, one-way ANOVA, *vs*. control, n = 20.
:::
:::
Discussion
==========
The present study demonstrated for the first time that the intravenous anesthetic propofol, at clinically relevant concentrations, suppressed corticothalamic-evoked EPSP temporal summation and action potential firing in thalamic somatosensory relay neurons in VB *in vitro*. The importance of corticothalamic excitatory input in the regulation of thalamic information processing and transfer has been stressed by recent evidence from *in vivo*and *in vitro*experiments. For example, activation of corticothalamic input facilitates single spike firing in thalamocortical neurons \[[@B13],[@B64]\], and alters thalamocortical responses to peripheral sensory stimuli \[[@B71]-[@B73]\]. *In vivo*studies in humans have shown that propofol, at plasma levels sufficient to produce unconsciousness, suppressed nociceptive \[[@B25]\] and non-nociceptive stimulus-induced increases in thalamic blood flow \[[@B24]\], indicating that thalamic activity was decreased. Our data provide unambiguous support for the hypothesis that propofol disrupts neuronal activity and synaptic transmission in the thalamus.
The thalamus is an important propofol target-site
-------------------------------------------------
Propofol-induced unconsciousness in humans is accompanied by thalamic inhibition of somatosensory-evoked activity \[[@B25]\] suggesting that such inhibition may play an important role in contributing to general anesthesia. Evidence supporting this assumption is the fact that propofol, at clinically relevant concentrations, consistently suppressed firing activity in all thalamic neurons tested here. Neurons in other brain areas, however, are relatively insensitive to propofol even at very high concentrations. For example, propofol potentiated GABA~A~receptor-elicited synaptic responses at 50 -- 500 μM in the hippocampus \[[@B70],[@B74]\], enhanced GABA-elicited inhibition at 50 μM in the olfactory cortex \[[@B75]\], and suppressed spike firing at 30 -- 100 μM in the locus coeruleus \[[@B76]\]. Therefore, our data demonstrate that the corticothalamic circuit is a highly sensitive target for propofol.
Significance of propofol-induced suppression of thalamic excitatory responses
-----------------------------------------------------------------------------
The transmission of sensory information through thalamic relay neurons to the cerebral cortex is state-dependent: transmission is reduced during slow wave sleep or drowsiness, and is enhanced during the waking state \[[@B8]\]. These changes in thalamic excitability are linked to depolarization of relay neurons, which is primarily regulated by corticothalamic excitatory input, or feedback \[[@B13]\]. Corticothalamic projection neurons in the cortex fire high-frequency single and burst spikes *in vivo*, and such excitatory input can readily lead to temporal summation in thalamic target neurons \[[@B8]\]. Here, we clearly demonstrated that corticothalamic-evoked temporal summation and action potential firing were markedly suppressed during propofol application.
A sustained, tonic, firing pattern in thalamic neurons is prevalent during the waking state \[[@B8],[@B13]\]; such a firing activity is lacking in brain slices. Thus, the metabotropic glutamate receptor agonist *trans*-ACPD was used as a pharmacological means to mimic this firing pattern \[[@B63]\]. We found that propofol also inhibited ACPD-evoked firing through a shunting mechanism, and cessation of spike firing in VB neurons was companied by the appearance of spontaneous IPSPs (Fig. [4](#F4){ref-type="fig"}). The occurrence of sIPSPs strongly suggests that propofol potentiated GABAergic inhibitory input from RTN to VB \[[@B77]\]. Coherent thalamocortical activity during the waking state appears to be essential for conscious experience \[[@B17],[@B78]\], and propofol-elicited shunting inhibition may disrupt such neuronal activity, thereby producing the behavioral changes seen during general anesthesia \[[@B79]\].
IPSPs can contribute to the sculpting of excitatory potentials, and thereby modulate synaptic integration \[[@B80],[@B81]\]. Such an effect is consistent with our observation propofol produced a GABA~A~receptor-mediated decrease in temporal summation in VB neurons (Fig. [2](#F2){ref-type="fig"}). In addition, the decrease in temporal summation in VB neurons in response to corticothalamic stimulation parallels the failure of spike transfer shown in Fig. [3](#F3){ref-type="fig"}. The failure of spike transfer in VB neurons following propofol application supports the observation that feedback inhibition gates spike transmission in hybrid thalamic circuits \[[@B82]\]. It is unlikely that propofol directly suppressed glutamatergic transmission because propofol, at the concentrations used here (\< 10 μM), has no effect on glutamate receptors \[[@B65],[@B66]\] or glutamatergic excitatory transmission \[[@B69]\].
GABA~A~receptors mediate the effect of propofol in VB neurons
-------------------------------------------------------------
Anesthetic suppression of excitatory responses may be mediated by at least two distinct mechanisms: enhancement of GABAergic transmission and direct suppression of glutamatergic transmission. Our results showed that there was strong evidence for shunting inhibition of synaptic temporal summation (Fig. [2](#F2){ref-type="fig"}) and *trans*-ACPD-evoked spike firing rate (Fig. [4](#F4){ref-type="fig"}), as a marked decrease in apparent input resistance was observed during propofol application. In addition, propofol caused a *prolonged*hyperpolarization of the membrane potential while inhibiting ACPD-evoked spike firing (Fig. [4](#F4){ref-type="fig"}), and the hyperpolarization was reversed by bicuculline or picrotoxin. The data strongly suggested that a tonic GABA~A~receptor current may be involved in mediating the inhibition during propofol application, consistent with previous observations \[[@B83],[@B84]\]. The GABA~A~receptor δ subunit is expressed in VB \[[@B57]\], and likely contributes to an extrasynaptic pentameric receptor with an α4βδ configuration \[[@B48],[@B53],[@B54]\]. Propofol potentiates δ subunit-containing GABA~A~receptors when co-expressed with an α4, but not α6, subunit \[[@B30],[@B31]\]. Thus, the propofol-induced hyperpolarization of the cell membrane observed here is consistent with its potentiation of extrasynaptic GABA~A~receptors.
Our data also provide evidence for propofol potentiation of the GABA~A~receptor chloride channel-mediated phasic currents, as propofol markedly increased picrotoxin-sensitive IPSC amplitude, decay time, and charge transfer (Fig. [5](#F5){ref-type="fig"}). In addition to a pool of synaptic receptors containing an α1 subunit that mediates fast IPSCs \[[@B44]-[@B47]\], α4 subunit-containing receptors accounts for \~30% of the total GABA~A~receptor population in the thalamus \[[@B48],[@B49]\]. α4-containing receptors are recognized by \[^3^H\]Ro15-4513 \[[@B48]\], indicating the presence of a γ2 subunit, and are expressed synaptically \[[@B40],[@B50]\]. Therefore, the pool of synaptic GABA~A~receptors expressed by VB neurons is heterogeneous, consisting primarily of α1- and α4-subunit containing receptors. Propofol markedly increased IPSC amplitude, suggesting potentiation of synaptic receptors containing either an α1 or α4 subunit \[[@B30],[@B35]\]. These receptors are likely to contain a β2 subunit as this subunit contributes to propofol-induced potentiation of GABA-evoked currents \[[@B85]\]. Finally, propofol-increased IPSC decay time suggested that the γ subunit might be involved, since propofol prolonged the deactivation time in receptors expressed in HEK cells containing a γ2L subunit \[[@B31]\]. These data strongly support the conclusion that propofol caused shunting inhibition by enhancing GABA~A~receptor-mediated chloride conductance in VB neurons through both synaptic and extrasynaptic receptors.
Conclusions
===========
The GABAergic general anesthetic propofol, at clinically relevant concentrations, markedly suppressed excitability and synaptic responsiveness to corticothalamic activation in thalamocortical relay neurons in VB. Propofol enhancement of postsynaptic GABA~A~receptor-function on VB neurons resulted in shunting inhibition of excitatory input. Recent clinical findings \[[@B22],[@B24],[@B25],[@B79],[@B86],[@B87]\] and *in vivo*electrophysiological evidence \[[@B28]\] have all suggested that thalamocortical circuits may constitute a strategic target for some general anesthetics including propofol. Our results support that hypothesis, and clearly establish the link between propofol-mediated inhibition of corticothalamic activation of VB neurons and propofol-enhanced GABA~A~receptor function.
Methods
=======
Brain slice preparation
-----------------------
Experiments were performed in accordance with institutional and federal guidelines. Thalamocortical (TC) slices were prepared as described \[[@B88]\] with a slight modification. Briefly, mice (C57BL/6, P25--55) were anesthetized by halothane and decapitated. The head was immediately submerged in ice-cold carbogenated (95% O~2~/5% CO~2~) slicing solution, and the brain was rapidly dissected out. The rostral portion of the brain was cut at 45° or 55°; the rostral end of the brain block was glued to a homemade platform. Slices (240 or 300 μm) were cut on a microslicer (Leica VT 1000S, Wetzlar, Germany) using a sapphire blade (Leica) to yield smooth-surface slices, gently rinsed once in cold artificial cerebrospinal fluid (ACSF) bath solution, and incubated in carbogenated ACSF at 34°C for 1 hr for recovery and at 24°C for at least another 1 hr before use. For horizontal slices, the brain was sagittally cut into two halves along the midline; 240 μm-thick slices containing both VB and RTN were prepared. Experiments were generally performed on TC slices, except those with RTN stimulation that were carried out in horizontal slices.
Electrophysiology
-----------------
Current-clamp recordings were performed at 35°C. Slices were perfused with carbogenated ACSF; neurons were visualized and identified using a Zeiss Axioskop (Jena, Germany) equipped with a 2.5 × objective and 40 × water immersion objective with a 2.4 mm working distance and IR-DIC optics. The resistance of the pipette was 3.5--6 MΩ when filled with internal solution. Tight seal (\> 2 GΩ) was achieved by application of a small negative pressure, using a 1 ml-syringe. Access resistance (R~a~) ranged from 10--14 MΩ, and was compensated by up to 60%; data were discarded if Ra \> 15 MΩ. Input resistance was measured at a holding membrane potential level close to resting membrane potential (RMP) from the voltage response elicited by a small current pulse (-60 pA). Only neurons that showed a stable RMP negative to -60 mV, action potential (AP) overshoot of \> 10 mV and R~i~\> 150 MΩ (in current-clamp mode) were selected for study. Although cells so selected generally showed stable data records for up to 240 min, pharmacological tests were completed within 90 min to minimize the variation of responses; only one experiment per slice was performed. Liquid junction potentials (11--12.2 mV) for intracellular and bath solutions were calculated by Junction Potential Calculator (Clampex 8, Axon Instruments, Union City, CA), and corrected online or offline. Membrane voltage was filtered at 5 kHz, membrane current at 2 kHz and then digitized at 10 kHz using an Axopatch 200A amplifier connected to a DigiData 1200 interface (Axon).
Extracellular electrical stimulation
------------------------------------
To stimulate CT fibers, a concentric bipolar tungsten electrode (FHC Inc., Bowdoinham, ME) was placed in either layer VI of the barrel cortex or the white matter in TC slices \[[@B60]\]. Single pulses or train pulses were delivered using a Master-8 pulse generator (A.M.P.I., Jerusalem, Israel) controlled by a PC and a constant current stimulus isolator (World Precision Instruments, Sarasota, FL). Responses were considered monosynaptic if the latency jitter was less than 0.4 ms and their rise times were consistent from trial to trial (3 trials). Latency was calculated from start of stimulus to onset of response. To confirm that the effects of propofol were GABA~A~receptor mediated, responses were blocked by a GABA~A~receptor antagonist (bicuculline 10 μM or gabazine 10 μM) or Cl^-^channel blocker (picrotoxin 100 μM). To evoke IPSCs, the stimulation electrode was placed in RTN, and synaptic currents were recorded in the presence of the GABA~B~receptor antagonist 2-OH saclofen (100 μM), and in some cases the non-NMDA receptor antagonist CNQX (20 μM) and NMDA receptor antagonist D-AP5 (40 μM) were added. CNQX and D-AP5 were also used to block evoked excitatory postsynaptic potentials (EPSPs) and the Na^+^channel blocker tetrodotoxin (500 nM) was used to block evoked action potentials.
Drug application
----------------
Drugs were applied by bath superfusion (unless otherwise noted) for at least 10 min prior to data collection using polytetrafluoroethylene (Teflon^®^) tubing and connectors; solution flow rates were 3 ml/min. Propofol was freshly prepared in DMSO and diluted with ACSF to clinically relevant concentrations (0.3 -- 3 μM); the final concentration of DMSO was 0.01%, which had no effects on the cells examined. The concentration range was selected based on the fact that a free aqueous concentration of \~2 μM is required to inhibit a response to a painful stimulus in 50% of test mammalian subjects \[[@B89]\].
Solutions
---------
Slicing solution contained (in mM): 2.5 KCl, 24 NaHCO~3~, 1.25 NaH~2~PO~4~, 234 sucrose, 11 glucose, 10 MgSO~4~, and 0.5 CaCl~2~. ACSF bath solution contained (in mM): 124 NaCl, 26 NaHCO~3,~2.5 KCl, 1.25 NaH~2~PO~4~, 1.2 MgCl~2~, and 2 CaCl~2~and 11 glucose. Intracellular solution contained (in mM): 130 K-gluconate, 5 NaCl, 2 MgCl~2~, 10 HEPES, 0.5 EGTA, 2 ATP-K, 0.3 GTP-Na, pH adjusted to 7.25 with KOH. K-gluconate was used because the impermeant ion gluconate does not contribute to anesthetic-induced changes in RMP or I-V relationship \[[@B90]\]. Voltage-clamp recordings of inhibitory postsynaptic currents (IPSCs) were made at 25°C, using a Cs^+^-based internal solution \[[@B91]\]. The bath solution for voltage-clamp contained (in mM): 117 NaCl, 25 NaHCO~3,~3.6 KCl, 1.2 NaH~2~PO~4~, 1.2 MgCl~2~, and 2.5 CaCl~2~and 11 glucose; osmolarity was adjusted to 300 mOsm with sucrose. All bath solutions were freshly prepared on the same experimental day.
Intracellular biocytin filling
------------------------------
Neurons from 30 mice were intracellularly filled with biocytin (0.5% in the pipette solution). After recording, slices were fixed for 24--72 hrs in phosphate buffer (PB) solution containing 4% paraformaldehyde, transferred to 20% sucrose solution in 0.1 M PB and re-sectioned to 60 -100 μm. After endogenous peroxidases were blocked with phosphate-buffered 3% H~2~O~2~, the slices were incubated with biotinylated horseradish peroxidase conjugated to avidin (ABC-*Elite*, Vector Labs, Burlingame, CA), washed and incubated with DAB for 15 min. Filled neurons were visualized and reconstructed.
Chemicals
---------
Compounds from Tocris Cookson (Ellisville, MO) were: (+) bicuculline, picrotoxin, gabazine, 2-OH saclofen, (2S)-3-\[\[(1S)-1-(3,4-dichlorophenyl) ethyl\] amino-2-hydroxypropyl\] (phenylmethyl) phosphinic acid (CGP55845), 6-cyano-7-nitroquinoxaline-2, 3-dione (CNQX), D-2-amino-5-phosphopentanoic acid (D-AP5), (±)1-aminocyclopentane-trans-1, 3-dicarboxylic acid (*trans*-ACPD). Tetrodotoxin (TTX) was from Alomone Labs (Jerusalem, Israel), and propofol was from Aldrich (Milwaukee, WI).
Data and statistical analysis
-----------------------------
Membrane voltages and currents were analyzed using both Clampfit 9.0 and MiniAnalysis 6 (Synaptosoft, Decatur, GA). To analyze temporal summation containing five responses, the peak of the first and fifth responses were measured from baseline and expressed as ΔV~1~and ΔV~5~, respectively; responses were calculated as: % increase = \[(ΔV~5~/ ΔV~1~) - 1\] × 100. Temporal summation was defined as % increase in depolarization occurring at the soma during a train \[[@B92]\]. Statistical analyses were performed with Sigmastat V 3.0 (SPSS, Chicago, IL) using *t*-test or one-way ANOVA. Data were expressed as means ± SE.
Declaration of Competing Interests
==================================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
(SWY) -- study design, data collection and analysis, manuscript preparation. (PAG) -- study design, manuscript preparation.
Acknowledgements
================
Funding for this work was provided by the Dept. of Anesthesiology, WMC, and by NIH grant GM66840 (to PAG). We would like to thank Edward Hurlock for advice on biocytin labeling and Dr. Neil Harrison for suggestions.
|
PubMed Central
|
2024-06-05T03:55:55.088794
|
2005-1-14
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074352/",
"journal": "Mol Pain. 2005 Jan 14; 1:2",
"authors": [
{
"first": "Shui-Wang",
"last": "Ying"
},
{
"first": "Peter A",
"last": "Goldstein"
}
]
}
|
PMC1074353
|
Background
==========
Tissue damage and inflammation produce an array of chemical mediators such as ATP, bradykinin, prostanoids, protons, cytokines and peptides including substance P that can excite or sensitize nociceptors to elicit pain at the site of injury. Among them prostanoids were shown to influence inflammation, and their administration was found to reproduce the major signs of inflammation including augmented pain \[[@B1]\]. Prostaglandin E~2~(PGE~2~) and prostaglandin I~2~(PGI~2~) are the products of arachidonic acid metabolism through the cyclooxygenase pathway. In addition to numerous other physiological actions *in vivo*, previous studies have indicated important roles for PGE~2~in nociception and inflammation \[[@B2],[@B3]\]. PGE~2~is generated in most cells in response to mechanical, thermal or chemical injury and inflammatory insult, resulting in sensitization or direct activation of nearby sensory nerve endings. Analgesic effects of non-steroidal anti-inflammatory drugs (NSAIDs) are attributed predominantly to inhibition of prostaglandin synthesis. Prostaglandins act upon a family of pharmacologically distinct prostanoid receptors including EP~1~, EP~2~, EP~3~, EP~4~and IP that activate several different G protein-coupled signaling pathways \[[@B2],[@B4],[@B5]\]. Primary sensory neurons in dorsal root ganglion (DRG) are known to express mRNAs encoding several prostanoid receptor subtypes, IP, EP~1~, EP~3~and EP~4~\[[@B6],[@B7]\]. The role of IP in inflammation has been clearly shown by the analysis of IP-deficient mice, although the underlying cellular mechanisms still remain to be elucidated \[[@B8]\]. In contrast, the potential involvement of EP receptors other than IP in inflammation and pain generation has not been well studied, although some earlier studies have suggested that prostanoids contribute to the development of pain through EP receptors \[[@B9],[@B10]\].
The capsaicin receptor TRPV1 is a non-selective cation channel expressed predominantly in unmyelinated C-fibers \[[@B11]\]. TRPV1 is activated not only by capsaicin, but also by protons or heat (with a threshold \> \~43°C), both of which cause pain *in vivo*\[[@B11]-[@B13]\]. A prominent role of TRPV1 in nociception has been demonstrated in studies of TRPV1-deficient mice \[[@B14],[@B15]\].
Recently, we reported that inflammatory mediators such as ATP, bradykinin and trypsin or tryptase potentiate TRPV1 activity in a PKC-dependent manner \[[@B16]-[@B18]\], and identified two target serine residues in TRPV1 as substrates for PKC-dependent phosphorylation \[[@B19]\]. On the other hand, there are several reports showing that a PKA signaling pathway mediates PGE~2~-induced potentiation of capsaicin-evoked responses in rat sensory neurons \[[@B20]-[@B22]\]. Therefore, we examined the effects of PGE~2~and PGI~2~on TRPV1 activity. Surprisingly, we found the functional interaction of TRPV1 with PGE~2~or PGI~2~occurs mainly through a PKC-dependent pathway at both cellular and behavioral levels.
Results
=======
Functional interaction between TRPV1 and PGE~2~
-----------------------------------------------
In order to examine the possibility that TRPV1 is involved in PGE~2~-induced hyperalgesia *in vivo*, we performed a behavioral analysis using wild type and TRPV1-deficient (TRPV1^-/-^) mice. PGE~2~(500 pmol/20 μL) produced a significant reduction in paw withdrawal latency in response to radiant heat (thermal hyperalgesia) at 5 to 90 min following intraplantar injection in wild type mice (Figure [1A](#F1){ref-type="fig"}). On the other hand, the PGE~2~-induced thermal hyperalgesia was almost completely abolished in TRPV1^-/-^mice, suggesting a functional interaction between PGE~2~and TRPV1 (Figure [1A](#F1){ref-type="fig"}), consistent with a previous report that capsaicin-ablation of primary afferent neurons prevents PGE~2~-induced thermal hyperalgesia \[[@B23]\]. We next examined the interaction between PGE~2~and TRPV1 in mouse DRG neurons using the patch-clamp technique. Capsaicin (100 nM) evoked small inward currents in DRG neurons. The capsaicin-evoked currents were significantly potentiated by 1.5 min pretreatment with PGE~2~(1μM) in 19 of 23 cells as previously reported \[[@B21]\] (Figure [1C](#F1){ref-type="fig"}) (3.36 ± 0.55 fold increase, n = 23 for PGE~2~(+); 0.78 ± 0.08 fold for PGE~2~(-) (Cont.), n = 5, p \< 0.05). Because it has been suggested that a PKA-dependent pathway is predominantly involved in the PGE~2~-induced potentiation of capsaicin-activated currents in rat DRG neurons \[[@B21]\], we examined the potential involvement of such a mechanism both in mouse DRG neurons and human embryonic kidney-derived HEK293 cells expressing TRPV1. No potentiation of the capsaicin-activated current responses was observed in DRG neurons treated with a mixture of forskolin (FSK, 10 μM), 3-isobutyl-1-methylxanthine (IBMX, 1 mM) and dibutyryl-cAMP (dbcAMP, 3 mM) for the same time period (1.5 min) (1.15 ± 0.20 fold increase, n = 9) although a significant increase in cAMP level was confirmed during such the treatment (Figures [1B](#F1){ref-type="fig"} and [1C](#F1){ref-type="fig"}). When we treated cells longer (6.5 min), 7 out of 14 cells showed increase of capsaicin-activated currents (2.15 ± 0.77 fold increase, n = 14, p = 0.28) (Figure [1C](#F1){ref-type="fig"}). In HEK293 cells, two different short (1.5 min) treatments to activate PKA produced no potentiation (Figure [1D](#F1){ref-type="fig"}) (treatment with a mixture of FSK, IBMX and dbcAMP in cells expressing TRPV1, 1.20 ± 0.19 fold increase, n = 11, and treatment with isoproterenol (Isop.) in cells expressing both TRPV1 and mouse β~1~-adrenergic receptors (β~1~-ADR), 0.83 ± 0.12 fold increase, n = 4) although a significant increase in cAMP level was confirmed following both treatments in HEK293 cells (Figure [1B](#F1){ref-type="fig"}). We also examined the effects of long treatment (6.5 min) with a mixture of FSK, IBMX and dbcAMP. This treatment caused significant potentiation of capsaicin-activated currents (2.39 ± 0.60 fold increase, n = 7, p \< 0.05) (Figure [1D](#F1){ref-type="fig"}). These results suggest that both PKA-dependent and -independent pathways are involved in the potentiation of the capsaicin-activated currents by PGE~2~, that it takes longer to cause potentiation of capsaicin-activated currents through a PKA-dependent pathway, and that the PKA-independent pathway is predominantly involved under the short treatment conditions. Indeed, it has been reported that capsaicin-activated currents were not increased upon FSK/IBMX or 8-bromo-cAMP (8-Br-cAMP)/IBMX treatment in *Xenopus*oocytes expressing TRPV1, or treatment with isoproterenol in oocytes expressing both TRPV1 and β~1~-ADR \[[@B24]\].
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Physiological interaction of PGE~2~with TRPV1 in mice. (**A**) PGE~2~-induced thermal hyperalgesia in TRPV1^+/+^mice (○, n = 6) or TRPV1^-/-^mice (▲, n = 6). Reduction of paw withdrawal latency (thermal hyperalgesia) by intraplantar PGE~2~(500 pmol/ 20 μL) injection was significantly diminished in TRPV1^-/-^mice. \* p \< 0.05, \*\* p \< 0.01 vs. TRPV1^+/+^mice. (**B**) Intracellular cAMP levels in mouse DRG neurons or HEK293 cells treated with a mixture of forskolin (FSK, 10 μM), IBMX (1 mM) and dibutyryl cAMP (dbcAMP, 3 mM), or PGE~2~(1 μM) or isoproterenol (Isop., 10 μM). \*, \# p \< 0.05 vs. Cont., \*\*, \#\# p \< 0.01 vs. Cont. (**C**) Representative traces of potentiation of capsaicin (100 nM)-activated current by extracellular PGE~2~(1 μM, 1.5 min) or a mixture of FSK(10 μM), IBMX (1 mM) and dbcAMP (3 mM) (6.5 min) in mouse DRG neurons. Currents were normalized to values induced by first capsaicin application in the absence of additives (bar graph). Capsaicin was reapplied 1.5 or 6.5 min after exposure to bath solution with additives. Numbers in parenthesis indicate cells tested. \* p \< 0.05 vs. Cont. Holding potential (V~h~): -60 mV. (**D**) Long (6.5 min) but not short (1.5 min) activation of PKA pathway has effect on TRPV1 responses in HEK293 cells. FSK (10 μM), IBMX (1 mM) and dbcAMP (3 mM) were applied to cells expressing rat TRPV1. Isop. (10 μM) was applied to cells expressing both rat TRPV1 and β~1~-adrenergic receptors (β~1~-ADR). Numbers in parenthesis indicate cells tested. V~h~: -60 mV. \* p \< 0.05 vs Cont.
:::
:::
PGE~2~increases TRPV1 activity through EP~1~receptors
-----------------------------------------------------
To explore the mechanism underlying the PKA-independent PGE~2~(1.5 min)-induced potentiation of the capsaicin-evoked responses observed in DRG neurons, we first examine the effects of PGE~2~on capsaicin-activated currents in HEK293 cells expressing TRPV1 and each EP receptor. PGE~2~(1 μM, 1.5 min) caused a robust increase in the magnitude of low dose (20 nM) capsaicin-activated currents in HEK293 cells co-expressing TRPV1 with EP~1~(0.90 ± 0.04 fold increase, n = 9 for control (Cont.); 4.60 ± 1.03 fold, n = 17 for PGE~2~, p \< 0.05) (Figures [2A](#F2){ref-type="fig"} and [2B](#F2){ref-type="fig"}). This increase lasted more than three minutes, as we previously reported for PAR-2 (proteinase activated receptor 2)-mediated potentiation of TRPV1 activity \[[@B16]\]. In contrast, no such potentiation was detected in cells expressing TRPV1 with other EP receptor subtypes (0.91 ± 0.09 fold increase, n = 7; 0.77 ± 0.13, n = 9; 0.72 ± 0.24, n = 5; 0.98 ± 0.18, n = 7; 0.89 ± 0.15, n = 9 for EP~2~, EP~3α~, EP~3β~, EP~3γ~or EP~4~, respectively) (Figure [2B](#F2){ref-type="fig"}). Protracted (6.5 min) treatment with PGE~2~caused a significant increase in capsaicin-activated currents in cells expressing TRPV1 and EP~4~, a phenomenon like that observed following treatment with a mixture of FSK, IBMX and dbcAMP (3.03 ± 0.48 fold increase, n = 6, p \< 0.05 vs. Cont.) (Figure [2B](#F2){ref-type="fig"}), suggesting that the EP~4~receptor, known to be expressed in DRG and coupled to Gs protein, is the receptor that activates a PKA-dependent signaling pathway upon prostaglandin exposure. All cells exhibiting an increase of capsaicin-activated currents upon treatment with a mixture of FSK, IBMX and dbcAMP also showed an increase in current in the presence of PMA (data not shown), suggesting that both PKA- and PKC-dependent pathways work in the same cells. To examine how PGE~2~changes TRPV1 responsiveness, we measured TRPV1 current in single cells by applying a range of concentrations of capsaicin in the absence or presence of PGE~2~. The currents were normalized to the maximal current produced by 1 μM capsaicin in each cell. Maximal current in the presence of PGE~2~was almost the same as that in the absence of PGE~2~. The resultant dose-response curves clearly demonstrate that PGE~2~enhances capsaicin action on TRPV1 by lowering EC~50~values without altering maximal responses (EC~50~from 81.0 nM to 27.6 nM) (Figure [2C](#F2){ref-type="fig"}). We next examined the effects of PGE~2~on the thermal sensitivity of TRPV1. When temperature ramps were applied to HEK293 cells expressing both TRPV1 and EP~1~in the absence of PGE~2~, heat-evoked currents developed at 40.7 ± 0.3°C (n = 8) (Figure [2D](#F2){ref-type="fig"}). In contrast, the temperature threshold for TRPV1 activation was significantly reduced to 30.6 ± 1.1°C in the presence of PGE~2~(n = 8, p \< 0.05) (Figure [2D](#F2){ref-type="fig"}) implying that under these conditions, TRPV1 could be activated at normal body temperature. A similar potentiating effect of PGE~2~was observed for proton (pH 6.2)-evoked TRPV1 current responses (0.91 ± 0.06 fold increase, n = 3 for control; 4.47 ± 1.09 fold, n = 7 for PGE~2~, p \< 0.01) (Figure [2E](#F2){ref-type="fig"}). These data clearly show that TRPV1 currents evoked by any of three different stimuli (capsaicin, proton, or heat) are potentiated or sensitized by PGE~2~through EP~1~receptor activation. On the other hand, the temperature threshold for TRPV1 activation was not changed upon treatment with a mixture of FSK, IBMX and dbcAMP in HEK293 cells expressing TRPV1 (40.8 ± 0.8°C, n = 4), suggesting different actions on TRPV1 by PKA and PKC.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
PGE~2~increases TRPV1 activity through EP~1~receptors in a PKC-dependent manner in HEK293 cells. (**A**) and (**B**) Treatment with PGE~2~(1.5 min) potentiates capsaicin-evoked responses in cells expressing rat TRPV1 with mouse EP~1~receptors, but not with other mouse EP receptors. Cells were pretreated with PGE~2~(1 μM) for 1.5 or 6.5 min before second capsaicin (20 nM) application. V~h~: -60 mV. Currents were normalized as described in Figure 1. \* p \< 0.05 vs. control (Cont.). Numbers in parenthesis indicate cells tested. (**C**) Capsaicin dose-response curves for TRPV1 activation in the absence (•) and presence (○) of extracellular 1 μM PGE~2~. Currents were normalized to the current maximally activated by 1 μM capsaicin in the absence of PGE~2~. Figure shows averaged data fitted with the Hill equation. EC~50~= 81.0 nM and Hill coefficient = 1.33 in the absence of PGE~2~. EC~50~= 27.6 nM and Hill coefficient = 1.01 in the presence of PGE~2~. Data were obtained from 54 different cells. (**D**) Temperature threshold for TRPV1 activation was reduced in the presence of extracellular PGE~2~(1 μM). Representative temperature-response profiles in the absence (upper) and presence (lower) of PGE~2~(left). Temperature threshold for TRPV1 activation in the presence of PGE~2~(30.6 ± 1.1°C) was significantly lower than that in the absence of PGE~2~(40.7 ± 0.3°C) (right). \* p \< 0.05 vs. PGE~2~(-). Numbers in parenthesis indicate cells tested. (**E**) Proton-evoked TRPV1 responses were significantly potentiated by PGE~2~(1 μM). \* p \< 0.01 vs. PGE~2~(-). (**F**) PKC-dependent pathway is involved in the PGE~2~(1 μM, 1.5 min)-induced potentiation of capsaicin-activated currents. In some experiments, calphostin C (Calp. C) (1 μM) or PKCε translocation inhibitor (PKCε-I) (200 μM) was included in the pipette solution. Currents were normalized as described in Figure 1. Numbers in parenthesis indicate cells tested. \* p \< 0.05 vs. Cont. V~h~: -60 mV.
:::
:::
The signaling pathway downstream of EP~1~remains to be clarified. We have reported that G~q/11~-coupled metabotropic receptor activation such as ATP (P2Y), bradykinin (B2) and proteinase-activated receptor 2 (PAR2) receptors causes potentiation or sensitization of TRPV1 through the PKC-dependent phosphorylation of TRPV1 \[[@B16]-[@B18],[@B25]\]. Therefore, we examined whether a similar signal transduction pathway is involved in the regulation of TRPV1 responses through EP~1~. When calphostin C (Calp.C), a specific PKC inhibitor, was added to the pipette solution, the effect of PGE~2~was almost completely inhibited (0.92 ± 0.15 fold increase, n = 10) (Figure [2F](#F2){ref-type="fig"}). Similarly, a PKCε translocation inhibitor (PKCε-I) abolished the potentiation of TRPV1 response by PGE~2~(1.11 ± 0.25 fold increase, n = 11) (Figure [2F](#F2){ref-type="fig"}). These data suggest that PGE~2~-induced potentiation of TRPV1 responsiveness develops through activation of PKCε. To further confirm the involvement of PKC-dependent phosphorylation, PGE~2~effects were examined using cells expressing a TRPV1 mutant, S502A/S800A which is insensitive to PKC-dependent phosphorylation \[[@B19]\]. No potentiation of capsaicin-activated currents was observed upon PGE~2~treatment of cells expressing S502A/S800A (0.85 ± 0.15 fold increase, n = 5) (Figure [2F](#F2){ref-type="fig"}), further indicating the involvement of PKC-dependent phosphorylation. Since S502 is a PKA-phosphorylation site as well \[[@B26]\], we examined the effects of treatment with a mixture of FSK, IBMX and dbcAMP on the capsaicin-activated currents in cells expressing S502A/S800A. Such treatment failed to potentiate the capsaicin-activated currents (1.13 ± 0.07 fold increase, n = 10), suggesting that S502 is a substrate for PKA-dependent phosphorylation of TRPV1 as well.
Sensitization of TRPV1 by EP~1~receptors in mouse
-------------------------------------------------
To examine the involvement of EP~1~in PGE~2~(1.5 min)-induced potentiation of capsaicin-evoked response in native neurons, we used a specific EP~1~agonist, ONO-DI-004 \[[@B27]\], and a specific EP~1~antagonist, ONO-8713 \[[@B28]\], in mouse DRG neurons. ONO-DI-004 was found to significantly increase the capsaicin-activated currents to an extent similar to that observed with PGE~2~(3.36 ± 0.68 fold increase for PGE~2~, n = 23, p \< 0.05 vs. control (Cont.); 3.30 ± 0.68 fold for ONO-DI-004 (EP~1~Agon.), n = 9, p \< 0.05 vs. Cont.) (Figures [3A](#F3){ref-type="fig"} left and [3B](#F3){ref-type="fig"}). Furthermore, potentiation of the capsaicin-activated currents by PGE~2~was inhibited in the presence of ONO-8713 (EP~1~Antg., 1.00 ± 0.17 fold increase, n = 8) (Figures [3A](#F3){ref-type="fig"} right and [3B](#F3){ref-type="fig"}). These results indicate that PGE~2~(1.5 min)-induced potentiation of the capsaicin-activated current responses occurs through EP~1~receptors in DRG neurons. To confirm the involvement of PKC-dependent events downstream of PGE~2~effects in DRG neurons, we first examined the effects of a specific phospholipase C (PLC) inhibitor, U73122 (3 μM). PGE~2~-induced potentiation of capsaicin-activated current was significantly diminished in the presence of U73122 while control U73343 exhibited no such effects (0.73 ± 0.11 fold increase, n = 8 for U73343; 3.40 ± 1.11 fold, n = 8 for U73433, p \< 0.05) (Figure [3B](#F3){ref-type="fig"}). Furthermore, PGE~2~failed to potentiate the capsaicin-activated currents when PKCε-I was included in the pipette solution (0.86 ± 0.09 fold increase, n = 12) (Figure [3B](#F3){ref-type="fig"}). A robust potentiating effect of phorbol 12-myristate 13-acetate (PMA, 100 nM) also supported the involvement of PKC-dependent events (16.36 ± 3.68 fold increase, n = 11, p \< 0.05) (Figure [3B](#F3){ref-type="fig"}). To further confirm the involvement of EP~1~receptors, DRG neurons of EP~1~deficient mice (EP~1~^-/-^) were subjected to patch-clamp analysis. PGE~2~failed to potentiate capsaicin-activated currents in the DRG neurons from EP~1~^-/-^mice (1.45 ± 0.70 fold increase, n = 10) (Figure [3B](#F3){ref-type="fig"}). Functional interaction of PKCε with TRPV1 prompted us to examine the expression of the two proteins in mouse DRG. Three hundred seventy eight out of 541 TRPV1 positive neurons (69.9 %) were stained with anti-PKCε antibody, supporting the TRPV1 activation pathway through PKCε (Figure [3C](#F3){ref-type="fig"}).
::: {#F3 .fig}
Figure 3
::: {.caption}
######
EP~1~receptor involvement in PGE~2~(1.5 min)-induced potentiation of capsaicin-activated currents in mouse DRG neurons. (**A**) Representative traces of potentiation of capsaicin-activated currents by a specific EP~1~agonist, ONO-DI-004 (10 μM, 1.5 min), and reverse of the PGE~2~(1.5 min)-induced potentiation by a specific EP~1~antagonist, ONO-8713 (1 μM). V~h~: -60 mV. (**B**) Effects of PGE~2~(1 μM), ONO-DI-004 (EP~1~Agon., 10 μM), PGE~2~plus ONO-8713 (EP~1~Antg., 1 μM), PGE~2~plus U73122 (3 μM), PGE~2~plus U73343 (3 μM), phorbol 12-myristate 13-acetate (PMA, 100 nM) or PGE~2~plus PKCε-I (200 μM) on capsaicin-activated currents in DRG neurons from wild type (EP~1~^+/+^) mice, and effects of PGE~2~on capsaicin-activated currents in DRG neurons from EP~1~^-/-^mice. Currents are normalized as described in Fig. 1. \* p \< 0.05 vs. Cont., + p \< 0.05 vs. U73343. Numbers in parenthesis indicate cells tested. (**C**) Co-expression of TRPV1 (green) and PKCε (blue) in mouse DRG. Arrowheads indicate neurons positive for TRPV1 but not for PKCε. Arrows indicate neurons positive for both TRPV1 and PKCε (light blue). Bar, 100 μm.
:::
:::
We next investigated the involvement of EP~1~receptors in PGE~2~-induced thermal hyperalgesia at the behavioral level. PGE~2~-induced thermal hyperalgesia was significantly diminished at 15 to 45 min after injection in EP~1~^-/-^mice (Figure [4A](#F4){ref-type="fig"}), relative to that observed in wild type mice. The involvement of EP~1~receptors in the PGE~2~-induced hypersensitivity was supported by another behavioral analysis in which PGE~2~caused less reduction of paw withdrawal latency in wild type mice pretreated with a specific EP~1~antagonist (500 pmol/ 20 μL) than in vehicle control (Figure [4A](#F4){ref-type="fig"}). These results suggest that a PKC-dependent pathway downstream of EP~1~activation is mainly involved in PGE~2~-induced thermal hyperalgesia. We have hypothesized that the potentiation of TRPV1 activity by several inflammatory mediators could represent one important mechanism underlying acute inflammatory pain sensation. To prove the accuracy of this hypothesis, we investigated the involvement of EP~1~in inflammatory pain-related responses using mustard oil which is known to cause inflammation \[[@B29],[@B30]\]. Topical application of mustard oil induced clear thermal hyperalgesia (Figure [4B](#F4){ref-type="fig"}). The mustard oil-induced thermal hyperalgesia was significantly reduced both in TRPV1^-/-^mice and EP~1~^-/-^mice. Thus, these data show that EP~1~contributes to inflammatory nociception in mice and support the hypothesis.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
Interaction between TRPV1 and EP~1~receptors in a behavioral level. (**A**) PGE~2~-induced thermal hyperalgesia in wild type mice with (▴, n = 6) or without (○, n = 6) pretreatment (ONO-8713, 500 pmol/ 20 μL), or in EP~1~^-/-^mice (•, n = 6). \* p \< 0.05 vs. wild type mice. (**B)**10% Mustard oil-induced thermal hyperalgesia in wild type mice (○, n = 12), TRPV1^-/-^mice (▲, n = 6) or EP~1~^-/-^mice (•, n = 6). \* p \< 0.05, \*\* p \< 0.01 vs. wild type mice.
:::
:::
Sensitization of TRPV1 by IP receptors
--------------------------------------
In order to determine whether the observed responses are specific to PGE~2~, we extended our analysis to PGI~2~whose receptor has been reported to be involved in nociception \[[@B8]\]. We first examined the effects of PGI~2~on capsaicin-activated currents in mouse DRG neurons. PGI~2~pretreatment (1000 nM, 1.5 min) potentiated capsaicin (100 nM)-activated currents (3.23 ± 0.55 fold increase, n = 14 or 0.78 ± 0.08 fold, n = 5 with or without (Cont.) PGI~2~, respectively; p \< 0.05) whereas at 100 nM, PGI~2~(1.5 min) showed no such effects (1.24 ± 0.22 fold, n = 11) (Figures. [5A](#F5){ref-type="fig"} and [5B](#F5){ref-type="fig"}). On the other hand, long (6.5 min) treatment with PGI~2~(100 nM) caused significant potentiation of capsaicin-activated currents as in the treatment with a mixture of FSK, IBMX and dbcAMP (2.06 ± 0.54 fold increase, n = 11, p \< 0.05 vs. Cont.) (Figures [5A](#F5){ref-type="fig"} and [5B](#F5){ref-type="fig"}). The potentiation effects of PGI~2~appear to occur through IP receptors because a specific IP agonist, ONO-54918-07 (100 nM) \[[@B31]\] caused similar potentiation of capsaicin-activated currents (Agon., 3.71 ± 0.81 fold increase, n = 9, p \< 0.05 vs. Cont.) (Figures [5A](#F5){ref-type="fig"} and [5B](#F5){ref-type="fig"}) although PGI~2~is known to cross react with some EP receptors \[[@B2]\]. The fact that a specific EP~1~antagonist, ONO-8713 failed to prevent the PGI~2~-induced potentiation (+EP~1~Antg., 3.55 ± 1.17 fold increase, n = 6) (Figure [5B](#F5){ref-type="fig"}) further suggests the involvement of IP receptors in the potentiation process. The involvement of IP receptors in the PGI~2~-induced potentiation of capsaicin-activated currents was further supported by the ineffectiveness of PGI~2~on DRG neurons from IP-deficient mice (IP^-/-^) (1.25 ± 0.16 fold increase, n = 9, p \< 0.01 vs. 1000 nM of PGI~2~) (Figure [5B](#F5){ref-type="fig"}). It has been reported that low concentrations of PGI~2~stimulate Gs protein coupled to IP receptors whereas high concentrations of PGI~2~stimulate not only Gs but also Gq \[[@B32]\]. This property might explain the dose-dependent effects of PGI~2~on capsaicin-activated currents: PKC-dependent sensitization of TRPV1 occurs downstream of Gq-coupled IP receptor activation at high concentrations (1000 nM) of PGI~2~(1.5 min) while long (6.5 min) treatment with low concentrations (100 nM) of PGI~2~causes potentiation of TRPV1 activity through Gs activation. To test this hypothesis, PGI~2~(1.5 min)-induced potentiation of capsaicin-activated currents was examined in the presence of U73122. When U73122 was included in the pipette solution, PGI~2~(1.5 min) failed to potentiate the currents whereas U73343 exhibited no such effects, indicating the involvement of PLC activation in the potentiating process (0.97 ± 0.40 fold increase, n = 9 for U73122, 2.58 ± 0.66 fold, n = 5 for U73343, p \< 0.05) (Figure [5B](#F5){ref-type="fig"}). Furthermore, PKCε-I included in the pipette solution almost completely blocked the PGI~2~(1000 nM)-induced potentiation (1.49 ± 0.60 fold increase, n = 14), suggesting the involvement of PKCε-dependent regulation mechanism (Figure [5B](#F5){ref-type="fig"}).
::: {#F5 .fig}
Figure 5
::: {.caption}
######
PGI~2~causes potentiation or sensitization of TRPV1 through mainly through PKC activation. (**A**) Representative traces of potentiation of capsaicin-activated currents by PGI~2~(1000 nM, 1.5 min), a specific IP agonist, ONO-54918-07 (1.5 min) or PGI~2~(100 nM, 6.5 min), but not by PGI~2~(100 nM, 1.5 min) in mouse DRG neurons. V~h~: -60 mV. (**B**) Effects of treatments (1.5 or 6.5 min) with PGI~2~(100 or 1000 nM), ONO-54918-07 (IP Agon., 100 nM), PGI~2~(1000 nM) plus ONO-8713 (EP~1~Antg., 1 μM), PGI~2~(1000 nM) plus U73122 (3 μM), PGI~2~(1000 nM) plus U73343 (3 μM) or PGI~2~(1000 nM) plus PKCε-I (200 μM) on capsaicin-activated currents in DRG neurons from wild type (IP^+/+^) mice, and effects of PGI~2~on capsaicin-activated currents in DRG neurons from IP-deficient (IP^-/-^) mice. Currents are normalized as described in Figure 1. \* p \< 0.05 vs. Cont. ++ p \< 0.01 vs. U73343, \# p \< 0.05, \#\# p \< 0.01 vs. PGI~2~(1000 nM, 1.5 min) in DRG neurons from IP^+/+^mice. Numbers in parenthesis indicate cells tested. (**C**) A representative trace of potentiation of capsaicin-activated currents by PGI~2~(1000 nM, 1.5 min) in HEK293 cells expressing both TRPV1 and IP. V~h~: -60 mV. (**D**) Effects of treatments (1.5 or 6.5 min) with PGI~2~(100 or 1000 nM) or PGI~2~(1000 nM) plus calphostin C (Calp. C, 1 μM) on capsaicin-activated currents in HEK293 cells expressing rat wild type TRPV1 or S502A/S800A mutant with IP. Currents are normalized as described in Figure 1. \* p \< 0.05 vs. Cont. (**E**) Temperature threshold for TRPV1 activation in the presence of PGI~2~(32.2 ± 1.2°C) was significantly lower than that in the absence of PGI~2~(38.2 ± 0.5°C) in HEK293 cells expressing rat TRPV1 and IP. \* p \< 0.01 vs. PGI~2~(-).
:::
:::
Dose-dependent PGI~2~(1.5 min)-induced potentiation of capsaicin-activated currents was also observed in HEK293 cells expressing TRPV1 and IP receptors (0.90 ± 0.04 fold increase, n = 9 without PGI~2~(Cont.); 0.68 ± 0.08 fold, n = 12 with 100 nM of PGI~2~; 0.75 ± 0.07 fold, n = 6 with 300 nM PGI~2~, 4.96 ± 1.36 fold, n = 8 with 1000 nM of PGI~2~, p \< 0.01 vs. Cont.) (Figures [5C](#F5){ref-type="fig"} and [5D](#F5){ref-type="fig"}, and data not shown). Calp. C blocked PGI~2~-induced potentiation of TRPV1 currents (0.75 ± 0.15 fold increase, n = 6) (Figure [5D](#F5){ref-type="fig"}). Furthermore, PGI~2~(1000 nM) failed to potentiate capsaicin-activated currents in HEK293 cells expressing the S502/S800 mutant (0.80 ± 0.05 fold, n = 6) (Figure [5D](#F5){ref-type="fig"}). Long (6.5 min) treatment with PGI~2~(100 nM) caused an increase in capsaicin-activated currents in 4 out of 6 cells, as did long treatment with a mixture of FSK, IBMX and dbcAMP in HEK293 cells expressing TRPV1 (3.19 ± 1.45 fold increase, n = 6, p = 0.16). These results suggest that a mechanism involving PKC is predominantly involved in the regulation of TRPV1 activity during short treatment with PGI~2~although both PKA-dependent and PKC-dependent pathways may contribute. The temperature threshold for TRPV1 activation was significantly reduced (from 38.2 ± 0.5°C, n = 5 to 32.2 ± 1.2°C, n = 5) in the presence of PGI~2~, suggesting the possibility that IP receptor activation can cause nociception at body temperature (Figure [5E](#F5){ref-type="fig"}). Finally, PGI~2~-induced thermal hyperalgesia observed in wild type mice disappeared almost completely in both TRPV1-deficient (TRPV1^-/-^) mice and IP-deficient (IP^-/-^) mice, suggesting that the functional interaction of TRPV1 with IP causes thermal hyperalgesia at the behavioral level (Figure [6](#F6){ref-type="fig"}).
::: {#F6 .fig}
Figure 6
::: {.caption}
######
Interaction between TRPV1 and IP receptors at a behavioral level. PGI~2~-induced thermal hyperalgesia in wild type mice (○, n = 6), TRPV1^-/-^mice (▲, n = 6) or IP^-/-^mice (•, n = 6). Thermal hyperalgesia by intraplantar PGI~2~(500 pmol/ 20 μL) injection was significantly diminished in TRPV1^-/-^mice and IP^-/-^mice. \* p \< 0.05, \*\* p \< 0.01 vs. wild type mice.
:::
:::
Discussion
==========
The data presented herein demonstrate that TRPV1 is essential for the development of thermal hyperalgesia *in vivo*induced by two major inflammation-associated prostaglandins, PGE~2~and PGI~2~, and that TRPV1 and EP~1~or IP receptors can functionally interact, mainly through a PKC-dependent pathway. The temperature threshold for TRPV1 activation is reduced below 35°C in the presence of prostaglandins, so that TRPV1 can be activated at normal body temperature, possibly leading to spontaneous pain sensation. This interaction might be one important underlying mechanism for the well-recognized peripheral nociceptive actions of PGE~2~or PGI~2~in the context of inflammation. In the present study, 1 μM PGE~2~or PGI~2~was found to potentiate or sensitize TRPV1 activity. It is not well known how much PGE~2~or PGI~2~is released locally at the site of inflammation. However, more than micromolar-order concentrations of PGE~2~and PGI~2~have been reported to be synthesized by macrophages upon lipopolysacharide (LPS) stimulation \[[@B33],[@B34]\], suggesting that 1 μM is an attainable concentration in the context of inflammation. It has been previously reported that EP~1~is coupled to intracellular Ca^2+^mobilization in CHO cells \[[@B35]\]. However, the transduction events downstream of EP~1~signaling have been unclear. Together with a report suggesting the possible coupling of EP~1~with G~q/11~-protein \[[@B36]\], our data indicate that EP~1~receptors activate a PKC-dependent signal transduction pathway.
There has been extensive work demonstrating the activation of a PKA-dependent pathway by PGE~2~that influences capsaicin- or heat-mediated actions in rat sensory neurons \[[@B20]-[@B22],[@B37],[@B38]\] as well as interactions between cloned TRPV1 and PKA \[[@B26],[@B39]-[@B42]\]. These results suggest that PKA plays a pivotal role in the development of hyperalgesia and inflammation by prostaglandins. In our experiments using mouse DRG neurons and HEK293 cells expressing TRPV1, a PKC-dependent pathway was found to be predominantly involved in both PGE~2~(1.5 min)- and PGI~2~(1.5 min)-induced responses. The reason that there has been no study describing the involvement of a PKC-dependent pathway in the regulation of TRPV1 following prostaglandin receptor activation is not clear. In the present study, it was found that both PKA- and PKC-dependent pathways are involved downstream of prostaglandin actions on TRPV1 although the PKC-dependent one appears to predominate. A PKA-dependent pathway took a relatively long time to exert its potentiating effects on TRPV1 activity, suggesting some difference between PKA- and PKC-dependent phosphorylation of TRPV1. Indeed, Bhave et al. treated cells with 8-Br-cAMP for 30 min to inhibit TRPV1 desensitization through phosphorylation \[[@B39]\], and significant potentiation of capsaicin-activated currents in rat DRG neurons was observed upon prolonged (greater than 10 min) exposure to PGE~2~\[[@B21]\]. Furthermore, there is a report describing the ineffectiveness of PKA stimulation on TRPV1 currents in *Xenopus*oocytes treated with 8-Br-cAMP and IBMX for relatively short periods \[[@B24]\]. Both PKA-dependent and PKC-dependent pathways might work in concert in native cells. Patch-clamp recordings in the previous studies were performed in the Ca^2+^-containing solutions, whereas we did all of our experiments under Ca^2+^-free conditions, to avoid Ca^2+^-dependent TRPV1 desensitization \[[@B43]\]. Potentiation of capsaicin-activated currents by PGE~2~was observed in embryonic rat DRG neurons \[[@B21]\] while we used adult mouse DRG neurons. Furthermore, potentiation of heat-activated currents \[[@B26]\], inhibition of desensitization of capsaicin-activated currents \[[@B39],[@B41],[@B44]\] or anandamide-induced cytosolic Ca^2+^increase \[[@B40]\] but not potentiation of capsaicin-activated current response were examined in the previous studies investigating the involvement of PKA-dependent pathway in TRPV1 activity. Thus, difference in experimental conditions or readout might also account for the different outcomes. The physiological relevance of the two different pathways downstream of prostaglandin exposure remains to be elucidated. The fact that only PKC activation leads to the reduction of temperature threshold for TRPV1 activation might be pertinent to this issue. Disruption of interaction between phosphatidylinositol-4, 5-bisphosphate (PIP~2~) and TRPV1 has also been reported to be involved in the sensitization of TRPV1 downstream of PLC activation \[[@B45],[@B46]\]. In our study, however, both PGE~2~- and PGI~2~-induced potentiation of TRPV1 activity was completely inhibited by treatments with two kinds of PKC inhibitors. Thus, we believe that a PKC-dependent pathway is predominantly involved in the PGE~2~- and PGI~2~-induced potentiation or sensitization of TRPV1 activity in mice.
The inhibition of PGE~2~-induced thermal hyperalgesia observed in EP~1~^-/-^mice, while significant, was not very robust, compared with that in TRPV1^-/-^mice (Figure [4](#F4){ref-type="fig"}). Other pathways, most likely including one involving PKA, might account for the residual component. Further, inhibition of mustard oil-induced thermal hyperalgesia observed in TRPV1^-/-^or EP~1~^-/-^mice might seem not to be robust or dramatic (Figure [4](#F4){ref-type="fig"}). Since many inflammatory factors activating PLC-coupled receptors are involved in the inflammatory response \[[@B47],[@B48]\]. In such a complicated environment, thermal hyperalgesia was significantly diminished in TRPV1^-/-^mice or EP~1~^-/-^mice albeit at a few time points, suggesting the importance of the two molecules in the context of inflammatory pain sensation. Given the fact that one of the final targets of both PGE~2~and PGI~2~is TRPV1 as shown in our study, compounds acting on EP~1~, IP or TRPV1, or interfering with their interaction could prove useful in the treatment of pain and inflammation.
Conclusions
===========
Potentiation or sensitization of TRPV1 activity through EP~1~or IP activation, mainly through PKC- and PKA-dependent mechanisms, might be important mechanism underlying the peripheral nociceptive actions of PGE~2~or PGI~2~.
Methods
=======
Animals
-------
Male C57BL/6-strain mice (4 weeks, SLC, Shizuoka, Japan), EP~1~-deficient mice (4 weeks, from Dr. Narumiya), IP-deficient mice (4 weeks, from Dr. Narumiya) or TRPV1-deficient mice (4 weeks, from Dr. Julius, UCSF) were used. They were housed in a controlled environment (12 h light/dark cycle, room temperature 22--24°C, 50--60% relative humidity) with free access to food and water. All procedures involving the care and use of mice were carried out in accordance with institutional (Mie University) guidelines and the National Institute of Health guide for the care and use of laboratory animals.
Behavioral study
----------------
Thermal nociceptive threshold was assessed using the paw withdrawal test. Mice were placed in a transparent Perspex box on a thin glass platform (Plantar test, Ugo Basile, Italy). They were injected intraplantarly with PGE~2~(500 pmol/ 20 μL, Sigma) with or without ONO-8713 (500 pmol/ 20 μL), or with PGI~2~(500 pmol/ 20 μL, Sigma), or applied topically to the plantar surface of right hind paw with 10% mustard oil (Sigma) (diluted with mineral oil), and the paw withdrawal latency to radiant heat applied to the plantar surface of hind paw was measured as the time from onset of the radiant heat to the withdrawal of the mouse hind paw.
Cell culture
------------
Human embryonic kidney-derived (HEK293) cells were maintained in Dulbecco\'s modified Eagle\'s medium (Invitrogen; supplemented with 10% fetal bovine serum, penicillin, streptomycin and L-glutamine) and transfected with 0.5 μg of rat TRPV1 cDNA and 0.5 μg of mouse EP or IP receptor cDNAs (EP~1~, EP~2~, EP~3α~, EP~3β~, EP~3γ~, EP~4~or IP) using Lipofectamine Plus Reagent (Invitrogen). Primary cultures prepared from adult C57BL/6-strain mice, EP~1~-deficient mice or IP-deficient mice dorsal root ganglion (DRG) neurons were incubated in medium containing nerve growth factor (Sigma, 100 ng/ml).
Electrophysiology
-----------------
Whole-cell patch-clamp recordings were performed 1 day after transfection to HEK293 cells or dissociation of the DRG neurons. Standard bath solution contained 140 mM NaCl, 5 mM KCl, 2 mM MgCl~2~, 5 mM EGTA, 10 mM HEPES, 10 mM glucose, pH7.4 (adjusted with NaOH). Pipette solution contained 140 mM KCl, 5 mM EGTA, 10 mM HEPES, pH7.4 (adjusted with KOH). All patch-clamp experiments were performed at room temperature (22°C). Thermal stimulation was applied by increasing the bath temperature at a rate of 1.0°C/sec with a preheated solution. When the heat-activated currents started to inactivate, the preheated solution was changed to a 22°C one. Chamber temperature was monitored with a thermocouple placed within 100 μm of the patch-clamped cell. For this analysis, heat-evoked current responses were compared between different cells, rather than within the same cell, because repetitive heat-evoked currents show significant desensitization even in the absence of extracellular Ca^2+^\[[@B13]\] and because the thermal sensitivity of TRPV1 increases with repeated heat application \[[@B49]\]. Threshold temperature for activation was defined as the intersection where two lines approximating the stable baseline current and the clearly increasing temperature-dependent current cross in the temperature-response profile. The sensitivity of DRG neurons to capsaicin is slightly lower than that of TRPV1-transfected HEK293 cells as previously reported \[[@B18],[@B50]\]. Therefore, we applied capsaicin at 100 nM to DRG neurons and at 20 nM to HEK293 cells.
cAMP measurement
----------------
Intracellular cAMP level was examined using \'cAMP Biotrak Enzymeimmunoassay System\' according to the manufacture\'s direction (Amersham Biosciences). In brief, intracellular cAMP released upon membrane hydrolysis of treated cells (10,000 cells/ well) after stimulation (90 sec) was measured based on competition between unlabelled cAMP and a fix quantity of Peroxidase-labeled cAMP for a limited number of the binding sites on a cAMP specific antibody.
Immunostaining
--------------
DRG was removed from male C57BL/6-strain mice and frozen in liquid nitrogen, and the frozen tissue was cut on a cryostat at a 10 μm thickness. The sections were incubated with the rabbit anti-rat TRPV1 polyclonal antibody (1: 500; Oncogene) and anti-rat PKCε monoclonal antibody (1: 250; Transduction lab) at 4°C for 2 days. Slides with the section were washed with PBS, followed by incubation with Alexa 488-conjugated goat anti-rabbit IgG (1: 700, Molecular Probes), Alexa 350-conjugated anti-mouse IgG (1: 500, Molecular Probes) and Texas Red-phalloidin (1: 500, Molecular Probes). Images were obtained using an Olympus fluorescent microscope with a cooled-CCD camera (ORCA-ER, Hamamatsu Photonics) and IP-Lab Image software (Scanalytics Inc.).
Chemicals
---------
ONO-DI-004, ONO-8713 and ONO-54918-07 were obtained from Ono Pharmaceutical Co., Ltd (Osaka, Japan). Calphostin C, phorbol 12-myristate 13-acetate, forskolin, 3-isobutyl-1-methylxanthine, dibutyryl-cAMP, isoproterenol, U73122 and U73343 were from Sigma, and PKCε translocation inhibitor was from Calbiochem.
Statistics
----------
Values are shown as the mean ± S.E. and data are analyzed using an unpaired *t*test. P values of \< 0.05 were considered significant.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
TM and TH carried out most of the experiments in this study. KT carried out the immunostaining experiments. TI carried out some electrophysiological experiments. ES made and maintained EP~1~- and IP-deficient mice, and participated in the interpretation of data. YS and SN participated in experimental design and discussion. TT carried out some biochemical experiments. MT contributed to all aspects of the study and wrote the manuscript.
Acknowledgements
================
We thank D. Julius (University of California, San Francisco) for giving us TRPV1-deficient mice, and M.J. Caterina (Johns Hopkins University), N. Saito (Kobe University) and M. Numazaki (University of Tsukuba) for their critical reading of the manuscript, and N. Suzuki and H. Tsumura (Mie University) for their support for maintaining mice. This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology in Japan, Japan Brain Foundation, Yamanouchi Foundation for Research on Metabolic Disorders, Uehara Memorial Foundation, AstraZeneca Research Foundation and ONO Medical Research Foundation to M.T.
|
PubMed Central
|
2024-06-05T03:55:55.091774
|
2005-1-17
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074353/",
"journal": "Mol Pain. 2005 Jan 17; 1:3",
"authors": [
{
"first": "Tomoko",
"last": "Moriyama"
},
{
"first": "Tomohiro",
"last": "Higashi"
},
{
"first": "Kazuya",
"last": "Togashi"
},
{
"first": "Tohko",
"last": "Iida"
},
{
"first": "Eri",
"last": "Segi"
},
{
"first": "Yukihiko",
"last": "Sugimoto"
},
{
"first": "Tomoko",
"last": "Tominaga"
},
{
"first": "Shuh",
"last": "Narumiya"
},
{
"first": "Makoto",
"last": "Tominaga"
}
]
}
|
PMC1074354
|
Background
==========
Spinal cord dorsal horn, the first central site for sensory processing, is divided into structurally and functionally distinct lamina regions \[[@B1]\]. Lamina I, or marginal zone of the spinal cord dorsal horn, is a critical region in nociceptive transmission. Different from other lamina regions, many lamina I neurons receive nociceptive inputs and directly relay the nociceptive information through ascending pathways to the brain \[[@B1],[@B2]\]. Nociceptive transmission through lamina I and their modulation there have important implications in both physiological and pathological pain conditions \[[@B2],[@B3]\]. Thus, a chemical mediator may have significant influence on pain conditions if it has an effect on this nociceptive pathway.
Extracellular ATP is a chemical mediator that has multiple effects in different tissues including nervous systems \[[@B4]\]. ATP is involved in sensory signaling at peripheral sites \[[@B5]-[@B7]\] and synaptic modulation at central sites in the somatic sensory system \[[@B8],[@B9]\]. At the periphery, ATP may directly stimulate nociceptive afferent fibers through the activation of P2 receptors, resulting in nociceptive inputs to the dorsal horn of the spinal cord or the equivalent sensory structure in the brain \[[@B6],[@B10]\]. At the central sites in the dorsal horn of the spinal cord, the inner part of lamina II (lamina IIi) may be a region where ATP-induced sensory inputs are transmitted. This is based on the strong P2X3 receptor-immunoreactivity in this region \[[@B11]\] as well as previous electrophysiological evidence \[[@B12],[@B13]\]. Lamina V may be another CNS site where ATP-sensitive inputs are transmitted. Many neurons in lamina V region receive sensory inputs with a wide dynamic range including both nociceptive and non-nociception signals \[[@B1]\]. Previous studies have shown that ATP modulates synaptic transmission to lamina V of the dorsal horn. The synaptic modulation is mediated by the activation of P2X receptors and these P2X receptors are localized at the presynaptic terminals of afferent fibers innervating lamina V neurons. The ATP-sensitive afferent terminals to lamina V neurons were found to be capsaicin-insensitive \[[@B14]\].
ATP and its receptors (P2 receptors) have been shown to be involved in inflammatory and neuropathic pain conditions \[[@B15],[@B16]\]. P2 receptors may be therapeutic targets for the management of pathological pain conditions \[[@B17],[@B18]\]. Although the significance of lamina I neurons in inflammatory and neuropathic pain conditions have been well recognized \[[@B19],[@B20]\], it is currently not known whether lamina I is one of the CNS sites for transmitting ATP-sensitive nociceptive inputs. This issue is important for understanding how ATP and its receptors are involved in nociceptive transmission under physiological and pathological conditions. This critical information has not been provided in the previous studies, which is mainly due to the difficulty in making electrophysiological recordings from lamina I neurons \[[@B2],[@B3]\]. In the present study, we performed patch-clamp recordings from lamina I neurons and studied effects of ATP and its analog αβmeATP on excitatory synaptic transmission to lamina I neurons. We demonstrated, for the first time, a P2X receptor-mediated nociceptive pathway to the lamina I of the spinal cord dorsal horn.
Results
=======
Effects of ATP and αβmeATP on miniature EPSCs of lamina I neurons
-----------------------------------------------------------------
We performed patch-clamp recordings from lamina I neurons in spinal cord slice preparations (Figure [1A](#F1){ref-type="fig"}). Under 40X objective with IR-DIC microscopic system, white matter and gray matter of the dorsal horn could be distinguished. All recordings (Figure [1A](#F1){ref-type="fig"}) were made from dorsal horn neurons whose somas were either within or adjacent to the white matter (within a distance of 30 μm).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Effects of ATP and αβmeATP on mEPSCs recorded from lamina I neurons of rat spinal cord slice preparationsA.**The image on the left shows a spinal cord slice section viewed under a 4X objective. An electrode is shown to the left side of the tissue section. The tip of the electrode is in lamina I. Scale bar: 100 μm. The image in the middle shows the tissue section viewed under a 40X objective with the IR-DIC system. Near the center of the field is a lamina I neuron (arrow indicated) that is patched with an electrode. The cell is on the border between white matter and gray matter. White matter (W), lamina I (I), and lamina II (II) are indicated. Lamina I region is outlined with two dash lines. Scale bar: 10 μm. The drawing on the right indicates the locations of all neurons recorded in this study. **B.**Representative traces show mEPSCs at the basal level (control) and following the application of 100 μM ATP. The graph on the right shows cumulative probability histogram of inter-event intervals in the same neuron. The inter-event intervals, which reflect the changes of mEPSC frequency, were significantly shifted following the application of 100 μM ATP (P \< 0.01, Kolmogorov-Smirnov test). **C.**The experiment was the same as **B**except 100 αβmeATP was tested. The inter-event intervals were significantly shifted following the application of 100 μM αβmeATP (P \< 0.01, Kolmogorov-Smirnov test). **D.**The bar graph shows pooled results from experiments represented in **B**and **C.**Both ATP (n = 13) and αβmeATP (n = 14) increased mEPSC frequency without affecting mEPSC amplitude. **E.**Effects of capsaicin (2 μM) on mEPSCs in 8 cells that were tested for ATP and αβmeATP in D. The scale is in logarithm. Data represent Mean ± SEM, \*p \< 0.05, \*\*p \< 0.01, compared with controls, paired Student\'s *t*-test.
:::
:::
We first studied whether lamina I neurons directly received synaptic inputs from ATP-sensitive presynaptic terminals. This was done by testing the effects of ATP (100 μM) on mEPSCs recorded from lamina I neurons. Miniature EPSCs were recorded in the presence of 10 mM lidocaine. The use of the high concentration of lidocaine was to block both TTX-sensitive and TTX-resistant Na^+^channels so that the effects of ATP reflected its direct action at the synaptic sites of the recorded neurons \[[@B21]\]. In this study, basal levels of mEPSCs were first recorded for 10 min. Miniature EPSC frequency and mEPSC amplitude at basal levels were served as controls, which had a variation within ± 10% in the same cells. We defined an increase of mEPSC frequency and amplitude to be more than 120% of control following ATP application, provided they returned to basal levels after washout of ATP. Bath application of 100 μM ATP for 2 min increased mEPSC frequency in 13 out of 30 lamina I neurons recorded (Figure [1B,D](#F1){ref-type="fig"}). Miniature EPSC frequency of these 13 cells increased to 210 ± 22% of control, from 0.56 ± 0.11 Hz at the basal level to 1.08 ± 0.19 Hz following the application of 100 μM ATP (n = 13, P \< 0.01, Figure [1D](#F1){ref-type="fig"}). Miniature EPSC frequency in all these 13 cells returned to basal levels 10 min after washout of ATP. In the cells for which mEPSC frequency was increased, the amplitude of mEPSCs was not significantly changed following ATP application (Figure [1D](#F1){ref-type="fig"}; 118 ± 7% control, 11 ± 1 pA in control and 12.0 ± 0.9 pA following ATP application, n = 13, P \> 0.05). For the remaining 17 cells, neither mEPSC frequency nor mEPSC amplitude was increased following ATP application (not shown). In all the cells recorded, we did not observe any direct whole-cell inward current during ATP application, a result consistent with previous studies in tissue slice preparations. In the above experiments and the experiments described below, 100 μM ATP was applied in the presence of 10 μM ARL67156, an ecto-ATPase inhibitor that was used to prevent ATP metabolism \[[@B14]\].
We tested αβmeATP, a metabolically stable ATP analog, on mEPSCs recorded from lamina I neurons. Similar to ATP, application of 100 μM αβmeATP increased mEPSC frequency in 14 out of 35 lamina I neurons (Figure [1C,D](#F1){ref-type="fig"}). Of these 14 cells, mEPSC frequency increased to 195 ± 25% of control (n = 14, P \< 0.01, Figure [1D](#F1){ref-type="fig"}). Miniature EPSC amplitude in these cells was not significantly different from the control (Figure [1D](#F1){ref-type="fig"}; 102 ± 2% control, n = 14, P \> 0.05). In the above experiments, αβmeATP did not directly evoke whole-cell inward currents in any cell. These results provided electrophysiological evidence indicating that the presynaptic terminals contacting the recorded lamina I neurons are ATP-sensitive/αβmeATP-sensitive. The expression of P2 receptors at these presynaptic terminals is suggested.
To test whether ATP-sensitive presynaptic terminals that contacted lamina I neurons were potentially related to nociceptive transmission, we tested capsaicin sensitivity in 8 cells that had responses to αβmeATP and ATP. Capsaicin was used because it selectively actives a subset of nociceptive afferent fibers that express VR1 receptors. We found that all of the 8 αβmeATP-responsive neurons also responded to capsaicin (2 μM) with a large increase of mEPSC frequency (Figure [1E](#F1){ref-type="fig"}; 0.4 ± 0.1 Hz in control *vs*18 ± 7 Hz following capsaicin, n = 8, P \< 0.05). Miniature EPSC amplitude was also found to be increased by capsaicin (10.8 ± 1.5 pA in control *vs*15.8 ± 2.2 following capsaicin, n = 8, P \< 0.05), which was most likely due to the high mEPSC frequency that produced temporal peak summation. There was no direct whole-cell inward current that was observed during capsaicin application.
Effects of ATP and its analogs on spontaneous EPSCs recorded from lamina I neurons
----------------------------------------------------------------------------------
We tested effects of P2 receptor agonists on spontaneous EPSCs (sEPSCs). In these experiments, Na^+^channels were not blocked so that action potentials were permitted to be generated. This allowed us to observe a more global effect of P2 receptor agonists on synaptic activity in lamina I. Of 106 lamina I neurons that were recorded, 44 cells showed increases in sEPSC frequency after application of 100 μM ATP (Figure [2A,B](#F2){ref-type="fig"}). The changes of sEPSC frequency in these 44 cells were 474 ± 70% of control (Figure [2B](#F2){ref-type="fig"}; 0.50 ± 0.09 Hz in control *vs*1.98 ± 0.30 Hz following ATP, n = 44, P \< 0.01). sEPSC amplitude was also significantly increased (Figure [2B](#F2){ref-type="fig"}; 157 ± 14% of control, 13.4 ± 1.0 pA in control *vs*19.8 ± 2.3 pA following ATP, n = 44, P \< 0.01,).
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Effects of ATP and αβmeATP on sEPSCs of lamina I neuronsA.**The trace on the tope panel shows sEPSCs recorded from a lamina I neuron before and following the application of 100 μM ATP. Two small portions, one before ATP and the other following ATP application (arrows indicated), are shown at the expanded scale. Two histograms below the traces show time courses of sEPSC frequency (left) and amplitude (right) in the same cell. Time bin is 20 sec. **B.**The bar graph shows pooled results of the effects of ATP (100 μM, n = 44), 10 μM αβmeATP (n = 8), and 100 μM αβmeATP (n = 26) on sEPSC frequency (filled bars) and amplitude (open bars). Data were normalized by the basal levels. **C.**The bar graph shows effects of capsaicin on sEPSCs in 6 neurons that were tested in **B**with 100 μM ATP and 100 μM αβmeATP.
:::
:::
Similar response was observed when αβmeATP was tested. Of 47 cells tested with 100 μM αβmeATP, 26 of them showed increases in sEPSC frequency and amplitude. The change of sEPSC frequency was 358 ± 67% of control (Figure [2B](#F2){ref-type="fig"}, n = 26, P \< 0.01,). The change of sEPSC amplitude was 163 ± 23% of control (Figure [2B](#F2){ref-type="fig"}, n = 26, P \< 0.05). At a lower concentration of 10 μM αβmeATP, 8 out of 19 cells showed increases of sEPSC frequency and the change of sEPSC frequency in these cells was 224 ± 62% of control (Figure [2B](#F2){ref-type="fig"}, n = 8, P \< 0.05).
We further tested effects of capsaicin on sEPSCs in 6 cells that showed the increases of sEPSC by αβmeATP and ATP. All 6 cells showed significant increases of sEPSC frequency following the application of 2 μM capsaicin (Figure [2C](#F2){ref-type="fig"}; 0.55 ± 0.18 Hz in control *vs*22.1 ± 7.8 Hz following capsaicin, n = 6). sEPSC amplitude was also found to be significantly increased by 2 μM capsaicin (Figure [2C](#F2){ref-type="fig"}; 14.6 ± 4.9 pA in control *vs*22.5 ± 3.5 pA following capsaicin, n = 6).
Effects of P2 antagonists on ATP-induced increases of sEPSCs
------------------------------------------------------------
We tested the effects of PPADS and suramin on ATP-induced increases of sEPSCs (Figure [3A,B](#F3){ref-type="fig"}). Of 5 cells for which 100 μM ATP increased sEPSC frequency to 512 ± 127% of basal levels in the absence of PPADS, sEPSC frequency was 155 ± 52% of the basal levels following 100 μM ATP in the presence of 10 μM PPADS (n = 4), and was 117 ± 17% of basal levels following 100 μM ATP in the presence of 30 μM suramin (n = 3). Similarly, αβmeATP-induced increases of sEPSCs in 3 cells (294 ± 78% of basal sEPSC frequency) were abolished in the presence of 10 μM PPADS (117 ± 14% of basal sEPSC frequency). These results suggested that P2 receptors are involved in the increases of spontaneous excitatory synaptic activity to lamina I neurons.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Block of ATP-induced synaptic responses by P2 antagonistsA.**The histogram shows the effects of ATP on sEPSC frequency in a lamina I neuron in the absence and presence of suramin (30 μM) or PPADS (10 μM). Suramin and PPADS were pre-applied for 10 min and were present during the recording. After 30 min washout of antagonists, 100 μM ATP was applied again for a test of recovery. **B.**Summarized data show effects of ATP on sEPSC frequency (filled bars) and amplitude (open bars) in the absence (n = 5) and presence of PPADS (10 μM, n = 4) and suramin (30 μM, n = 3). **C.**Histogram on the left shows a time course of ATP-induced increase of sEPSC frequency in a lamina I neuron in the presence of 100 μM reactive blue 2(RB2). Bar graph on the right is a summary that shows the effects of 100 μM ATP on the sEPSCs in the absence and presence of 100 μM RB2 (n = 11).
:::
:::
While ATP-induced increases of sEPSCs were abolished by both PPADS and suramin, ATP still could increase sEPSCs in the presence of reactive blue 2, a P2Y receptor antagonist. As shown in Figure [3C](#F3){ref-type="fig"}, in 11 cells showing increases of sEPSCs in response to 100 μM ATP (354 ± 85% of basal levels, P \< 0.05), 100 μM ATP increased the sEPSC frequency to 221 ± 49% of basal levels when 100 μM RB-2 was present (P \< 0.05, compared with basal levels, Figure [3C](#F3){ref-type="fig"}). Taken together, these results suggest that P2X receptors are involved in ATP-induced increases of sEPSCs.
Effects of ATP and αβmeATP on synaptic inputs from primary afferent fibers to lamina I neurons
----------------------------------------------------------------------------------------------
To further demonstrate that ATP-sensitive terminals are derived from primary afferent fibers, we studied effects of ATP and αβmeATP on synaptic inputs elicited by stimulation of dorsal root (i.e. primary afferent fibers). Dorsal root stimuli resulted in the evoked EPSCs (eEPSCs) recorded from lamina I neurons. Monosynaptic eEPSCs showed large variations in their latency among different cells (Figure [4A](#F4){ref-type="fig"}). The conduction velocity of afferent inputs, calculated from the latency of eEPSCs and the length of the dorsal roots, was from 0.33 to 4 m/s (1.1 ± 0.11 m/s, n = 57, Figure [4A](#F4){ref-type="fig"}), in the range of C-fiber and Aδ-fiber conduction velocity at these ages of rats \[[@B22]\]. Synaptic failure rates were low before the application of ATP and αβmeATP. However, following the application of 100 μM ATP or 100 μM αβmeATP, there was a significant increase in synaptic failure rates and a significant decrease of the averaged eEPSC amplitude in many lamina I neurons (Figure [4B](#F4){ref-type="fig"} to [4E](#F4){ref-type="fig"}). Of 13 cells tested with 100 μM ATP in the presence of 10 μM ARL67156 and 2 mM caffeine, 9 of them showed increases in synaptic failure rates and decreases in the averaged eEPSC amplitude (Figure [4B,D](#F4){ref-type="fig"}). The failure rates were 8 ± 3% in control and increased to 78 ± 7% following 100 μM ATP applications (n = 9, P \< 0.01); the averaged eEPSC amplitude was 35 ± 9 pA in control and reduced to 12 ± 4 pA following 100 μM ATP application (24 ± 7% of control, n = 9, P \< 0.01). Of 29 cells tested with 100 μM αβmeATP, 18 of them showed depression of the averaged amplitude of eEPSCs, accompanied with the increases of failure rates (Figure [4C,E](#F4){ref-type="fig"}). The failure rates were 12 ± 5% in control and increased to 53 ± 10% following 100 μM αβmeATP applications (n = 18); the eEPSC amplitude was reduced to 35 ± 7% of control following 100 μM αβmeATP application (n = 18, p \< 0.01).
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Suppression of eEPSCs by ATP and αβmeATPA.**The trace on the left shows an averaged eEPSC recorded from a lamina I neuron that received monosynaptic inputs from Aδ-afferent fibers. The averaged eEPSC was obtained from eEPSCs elicited by 10 sweeps of stimuli. The trace on the right is an averaged eEPSC from a lamina I neuron that received monosynaptic inputs from C-fibers. The graph shows numbers of lamina I neurons recorded for all the eEPSC experiments in this study and their corresponding afferent conduction velocity (n = 57). **B.**Top panel shows superimposed eEPSCs following 10 sweeps of stimuli in normal bath solution (control), following the application of 100 μM ATP, and washout of ATP (recovery). The averaged eEPSCs from 10 sweeps of stimuli are shown in the bottom panel. **C.**Similar to **B**except αβmeATP was tested in a different lamina I neuron. **D.**The bar graph summarizes the increases of synaptic failure rates by 100 μM ATP (black bar, n = 9) and 100 μM αβmATP (gray bar, n = 18). **E.**A summary of the reduction of average eEPSC amplitude by 100 μM ATP (black bar, n = 9) and 100 μM αβmATP (gray bar, n = 18).
:::
:::
The suppression of eEPSCs could be due to a direct action on primary afferent axons whose terminals synapse to lamina I neurons or due to an effect indirectly through GABAergic presynaptic inhibition. To exclude the latter possibility, we tested the effects of 100 μM ATP and 100 μM αβmeATP on eEPSCs in the presence of the GABA-A receptor inhibitor bicuculline (20 μM). We found that both ATP and αβmeATP still could suppress eEPSCs (Figure [5](#F5){ref-type="fig"}). Of 7 cells tested with 100 μM ATP in the presence of bicuculline, 5 of them showed suppression of eEPSCs. Synaptic failure rates increased from 26 ± 6% before ATP to 68 ± 15% following 100 μM ATP (n = 5, p \< 0.01, Figure [5B](#F5){ref-type="fig"}). The averaged eEPSC amplitude was also significantly decreased (Figure [5C](#F5){ref-type="fig"}). Of 14 cells tested with 100 μM αβmeATP in the presence of bicuculline, 6 of them showed depression of eEPSCs. For these 6 cells, synaptic failure rates were increased from 27 ± 10% before αβmeATP to 50 ± 15% following the application of 100 μM αβmeATP (n = 6, p \< 0.05, Figure [5B](#F5){ref-type="fig"}), the averaged eEPSC amplitude decreased to 33% ± 9% of control following 100 μM αβmeATP (n = 6, p \< 0.01, Figure [5C](#F5){ref-type="fig"}). These results suggest that the suppression of primary afferent synaptic inputs to lamina I neurons was due to a direct action of ATP or αβmeATP on primary afferent fibers.
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Suppression of eEPSCs by ATP and αβmeATP in the presence of bicucullineA.**The traces show eEPSCs recorded from a lamina I neurons under the following conditions: in bath solution with 20 μM bicuculline (control), following the application of 100 μM ATP in the presence of 20 μM bicuculline, and washout of ATP (recovery). In each set of a test, eEPSCs were elicited by 10 sweeps of stimuli. **B.**The bar graph shows a pooled result of increases in synaptic failures by ATP (100 μM, n = 5) and αβmeATP (100 μM, n = 6). **C.**A summary shows the decreases of average eEPSCs by 100 μM ATP (n = 5) and 100 μM αβmeATP (n = 6). All recordings were made from lamina I neurons in the presence of 20 μM bicuculline.
:::
:::
If P2X receptors were involved in the suppression of eEPSCs, the suppression should be abolished in the presence of P2X antagonists. We tested the effects of PPADS on αβmeATP-induced suppression of eEPSCs (Figure [6](#F6){ref-type="fig"}). Of 4 cells for which 100 μM αβmeATP suppressed eEPSC amplitude to 31 ± 11% of control (n = 4, P \< 0.05) in the absence of PPADS, αβmeATP did not significantly suppress eEPSC amplitude when 10 μM PPADS was present (82 ± 7% of control, n = 4, Figure [6C](#F6){ref-type="fig"}). Thus, P2X receptors are expressed on primary afferent fibers that contact lamina I neurons.
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**Block of αβmeATP-induced suppression of eEPSCs by PPADSA.**The three traces on the top, middle and bottom panels show eEPSCs recorded from a lamina I neuron in normal bath solution (Control), during the application of 100 μM αβmeATP and after washout of αβmeATP (Recover), respectively. **B.**The same neuron shown **A**was tested again in the presence of 10 μM PPADS. In both **A**and **B**, a trace of eEPSC was obtained from an average of eEPSCs evoked by 10 sweeps of stimuli. **C.**Pooled results (n = 4) show effects of 100 μM αβmeATP on eEPSC amplitude in the absence and presence of 10 μM PPADS. Data represent percent of control values, and eEPSCs in control is scaled as 100%.
:::
:::
::: {#F7 .fig}
Figure 7
::: {.caption}
######
**Capsaicin-sensitivity of αβmeATP-sensitive primary afferents to lamina IA.**Example recordings from a lamina I neuron show eEPSCs following 10 sweeps of stimuli in normal bath solution (control), following the application of 100 μM αβmeATP, and washout of the drug (recovery). **B.**Capsaicin (2 μM) was tested in the same neuron as shown in **A**. **C.**The bar graph summarizes capsaicin-induced synaptic failures in 6 lamina I neurons that received ATP/αβmeATP-sensitive primary afferent inputs. **D.**A summary shows capsaicin-induced reduction of averaged eEPSC amplitude (n = 6). All neurons were recorded from lamina I that received ATP/αβmeATP-sensitive primary afferent inputs.
:::
:::
To study whether ATP-sensitive afferent fibers that synapsed to lamina I neurons were nociceptive afferent fibers, we tested capsaicin sensitivity of the evoked eEPSCs. Of the 6 lamina I neurons that showed eEPSC suppression by ATP and αβmeATP (Figure [7A](#F7){ref-type="fig"}), all of them also showed eEPSC suppression following the application of 2 μM capsaicin (Figure [7B,C, D](#F7){ref-type="fig"}). The synaptic failure rates of these 6 cells were 12 ± 6% before capsaicin and increased to 65 ± 16% (n = 6, p \< 0.05) following the application of 2 μM capsaicin. The averaged amplitude of eEPSCs was 46 ± 17 pA before capsaicin and reduced to 18 ± 10 pA following 2 μM capsaicin (n = 6, p \< 0.01). Of these six recordings, the afferent conduction velocities were less than 1 m/s (from 0.5 to 0.8 m/s) in 4 cells and more than 1 m/s (1.1 and 1.5 m/s) in 2 cells. These results suggest that ATP-sensitive afferent fibers that synapse to lamina I neurons are nociceptive afferent fibers.
Discussion
==========
In the present study we have shown that ATP and its analog αβmeATP produce profound effects on excitatory synaptic transmission to lamina I neurons of the spinal cord dorsal horn. Pharmacological tests indicate that P2X receptors are involved in the synaptic responses and that P2X receptors are expressed on many capsaicin-sensitive afferent fibers innervating lamina I neurons. In view of the significance of lamina I in transmitting nociceptive signals and in the development of pathological pain conditions \[[@B2],[@B3]\], our findings provide a new insight into P2X receptor functions in physiological and pathological pain conditions \[[@B16],[@B17],[@B23]\].
In this study, the presynaptic actions of ATP and αβmeATP were evidenced electrophysiologically by their effects on mEPSCs and eEPSCs \[[@B21]\]. Both ATP and αβmeATP increased mEPSC frequency without affecting mEPSC amplitude; both ATP and αβmeATP also increased synaptic failure rates of eEPSCs. The involvement of P2X receptors is supported by several lines of pharmacological evidence. First, αβmeATP is a selective agonist to P2X receptors \[[@B24]\] and its effects on mEPSCs, sEPSCs and eEPSCs were similar to ATP in this study. Second, at low micromolar concentrations, PPADS is a selective antagonist for P2X receptors \[[@B24]\]; we found that PPADS at 10 μM abolished synaptic effects of ATP and αβmeATP. Third, ATP and αβmeATP still could induce synaptic responses in the presence of the P2Y receptor antagonist RB2. Fourth, the potential complication by the ATP metabolite adenosine was excluded with the use of caffeine to block adenosine receptors and the used of ARL67156 to prevent ATP metabolism \[[@B14]\].
We have found that in the same lamina I neurons for which ATP and αβmeATP had synaptic effects, capsaicin produced similar responses. For example, in the lamina I cells for which αβmeATP increased synaptic failure rates of eEPSCs, capsaicin also enhanced synaptic failure rates in these cells. These findings suggest that many ATP- and αβmeATP-sensitive presynaptic terminals to lamina I neurons were derived from capsaicin-sensitive nociceptive afferent fibers. Our previous study showed that P2X receptor-expressing afferent fibers innervating lamina V neurons were capsaicin-insensitive Aδ-fibers \[[@B14]\]. In the present study, we have found that P2X receptor-expressing afferent fibers to lamina I are capsaicin-sensitive and could be either Aδ or C-fibers. Thus, the P2X-expressing afferent fibers are distinct in their lamina distribution and in their capsaicin sensitivity. The lamina I innervation and the capsaicin sensitivity of these ATP-sensitive afferent fibers strongly suggest the nociceptive function of these P2X receptors.
The synaptic failures of eEPSCs by ATP and αβmeATP could be due to conduction block following the activation of P2X receptors on primary afferent axons. Alternatively, the increase of synaptic failure rates was caused by presynaptic inhibition through GABAergic inhibitory interneurons \[[@B25]\]. This latter possibility, however, is discounted because ATP and αβmeATP still increased the failure rates of the evoked EPSCs in the presence of bicuculline to block GABA-A receptors. These results further support the idea that P2X receptors are expressed on primary afferent fibers innervating lamina I neurons. The synaptic failure of eEPSCs following P2X receptor activation does not discount the involvement of P2X receptors in transmitting nociceptive signals to lamina I regions. In opposite, it suggests that P2X receptor action may directly depolarize primary afferent fibers to initiate nociceptive signals, which are then transmitted to lamina neurons. Consistent with this idea, out results showed that both frequency and amplitude of sEPSCs were significantly increased following the application of ATP or αβmeATP.
Of more than 10 subtypes of functional P2X receptors identified so far, low concentrations of αβmeATP selectively activate those that contain P2X1 or P2X3 subunits \[[@B24]\]. Thus, the sensitivity to 10 μM αβmeATP in our study narrows down the possible P2X receptor subtypes \[[@B26]\]. The sensitivity to the block by 10 μM PPADS and 30 μM suramin in our study may allow us to further exclude a number of P2 receptors including P2Y receptors, homomeric P2X4 receptors, homomeric P2X6 receptors, and homomeric P2X7 receptors \[[@B26]\]. Previous studies have shown the involvement of P2X3 receptors in nociception \[[@B7]\]. However, because P2X3-expressing afferent terminals are restricted in the inner layer of lamina II \[[@B11]\], it is less likely that the functional P2X receptors in our study are P2X3-containing subtypes. The P2X receptor subtype(s) involved in the nociceptive pathway shown in this study may not be easily identified due to the lack of selective agonists and antagonists for most of P2X receptor subtypes and due to the possible presence of more P2X receptor subtypes. The answer to this critical issue in future studies may provide new therapeutic targets for pain management.
Methods
=======
Spinal cord slice preparation
-----------------------------
Principles of laboratory animal care (NIH publication No. 86-23, revised 1985) were followed in all the experiments described in the present study. Spinal cord slice preparations and patch-clamp recordings were described in details in our previous studies \[[@B14]\]. In brief, Sprague Dawley rats at the postnatal age of 10--21 days were used. Transverse spinal cord slices were prepared from lumbar enlargement of the spinal cords. Two types of spinal cord slice sections were made. One type was in the thickness of 400 μm without dorsal root, and was used for studying mEPSCs and sEPSCs. The other type was in the thickness of 600 μm with attached dorsal roots, and was use for experiments testing eEPSCs. The length of the dorsal roots was in a range of 8--12 mm. The spinal cord slices were maintained in a basket submerged in \~200 ml Krebs solution (24°C). The Krebs solution contained (in mM): NaCl 117, KCl 3.6, CaCl~2~2.5, MgCl~2~1.2, NaH~2~PO~4~1.2, NaHCO~3~25 and glucose 11; the solution was saturated with 95 % O~2~and 5% CO~2~and had pH of 7.3.
Patch-clamp recordings
----------------------
In each experiment a spinal cord slice was transferred to a 0.5 ml recording chamber and placed on the stage of an upright microscope. The microscope was equipped with an infrared differential interference contrast (IR-DIC) system. The spinal cord slice was superfused with the Krebs\' solution flowing at 10 ml/min at room temperature (24°C); the Krebs\' solution was equilibrated with 95% O~2~and 5% CO~2~. Lamina I, the marginal region of the dorsal horn, was identified under the microscope. Individual neurons were identified with a 40× water immersion objective. Whole-cell patch-clamp recordings were made from lamina I neurons with electrodes filled with an internal solution containing (in mM): 135K-gluconate, 5KCl, 0.5CaCl~2~, 2MgCl~2~, 5EGTA, and 5HEPES; the pH of the solution was adjusted to 7.3 with NaOH. The resistances of the electrodes were \~5 MΩ when filled with the internal solution. The access resistance was below 30 MΩ and was not compensated. Signals were amplified and filtered at 2 kHz (Axopatch 200B) and sampled at 5 kHz using pCLAMP 7.0 (Axon instruments). EPSCs were recorded with cells held at -60 mV. The holding potential was close to the reversal potential for GABA~A~and glycine receptors under the experimental conditions so that the outward IPSCs were minimized and usually undetectable.
Spontaneous EPSCs were recorded in the normal Krebs\' bath solution, and mEPSCs were recorded in the presence of 10 mM lidocaine. To record eEPSCs, stimuli were applied to a dorsal root with a suction electrode. Stimulation intensities were 60--150 μA for Aδ fiber and 200--800 μA for C-fibers; stimulation duration was always 0.1 msec. Evoked EPSCs were usually elicited by 10 sweeps of stimuli at a time interval of 5 s, which were used to obtain the average of eEPSCs. Monosynaptic eEPSCs were judged by a constant latency to the repeated stimuli. Conduction velocity was calculated from the latency of eEPSCs and the length of dorsal roots.
Effects of P2 agonists and capsaicin on mEPSCs, sEPSCs and eEPSCs were tested using ATP (100 μM), α,β-methylene-ATP (αβmeATP, 10 and 100 μM), and 2 μM capsaicin. When ATP was tested, bath solution always contained the ecto-ATP inhibitor ARL67156 (10 μM), and the adenosine receptor antagonist caffeine (2 mM) was also present. P2 antagonists including pyridoxalphosphate-6-azophenyl-2\',4\'-disulfonic acid (PPADS, 10 μM), suramin (30 μM) and reactive blue 2 (RB2, 100 μM) were used in this study. When the antagonists were tested, they were first pre-applied for 10 min, and then were co-applied with P2 agonists. The time intervals for multiple applications of testing compounds were 20 min. All compounds were applied via the bath solution. All compounds were purchased from Sigma (St. Louis, MO).
Data analysis
-------------
Synaptic events including sEPSCs and mEPSCs were analyzed using Mini Analysis Program (Jaejin software, Anderson Place, GA) with criteria being the same as previously described \[[@B21],[@B14]\]. In analyzing the changes of mEPSC frequency or sEPSC frequency following the bath application of P2 agonists, frequency time courses before and after P2 agonists were first constructed with time bin of 10 s. Then the average response in continuous 6 bins (60 s) around the peak was used to calculate the changes in reference to the basal level (control). Cells were assigned to be responsive to a test compound when there were more than 20% increases in mEPSC frequency or sEPSC frequency as previously described \[[@B21]\]. For eEPSCs, eEPSCs were elicited by 10 sweeps of stimuli, which were used for calculating the averaged amplitude of eEPSC. Cells were considered to be responsive to a testing compound when there were more than 20% increases in the averaged amplitude of eEPSCs. Unless otherwise indicated, data were presented as mean ± SEM. Paired Student\'s *t*-tests were used for statistical comparison, and significance was considered at the level of the p \< 0.05.
Competing interests
===================
The author(s) declare that they have no competing interests.
Acknowledgments
===============
This work was supported by a grant NS38254 to J.G.G.
|
PubMed Central
|
2024-06-05T03:55:55.094932
|
2005-1-17
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074354/",
"journal": "Mol Pain. 2005 Jan 17; 1:4",
"authors": [
{
"first": "Meng",
"last": "Chen"
},
{
"first": "Jianguo G",
"last": "Gu"
}
]
}
|
PMC1074355
|
Glutamate is the major excitatory amino acid neurotransmitter that plays an important role in many physiological functions. Maintaining a physiological range of extracellular glutamate concentration is key to preventing glutamate over-excitation and neurotoxicity that could occur under a variety of pathological conditions. Regulating extracellular glutamate is primarily carried out by an efficient, high-capacity glutamate transporter (GT) system, because clearance of extracellular glutamate via glutamate metabolism or diffusion is negligible. To date, at least five cell membrane GT proteins have been cloned \[[@B2],[@B3]\]. GT is labeled by a common name \'excitatory amino acid transporter\' (e.g., EAAT1). Among cell membrane GT, EAAT1 (GLAST), EAAT2 (GLT1), and EAAT3 (EAAC1) are particularly relevant to the regulation of glutamate uptake in broad CNS regions. EAAC1 is generally considered as a neuronal GT, whereas GLAST and GLT1 are primarily astroglial GT, although both GLAST and GLT1 also have been located in neuronal cells during the developmental stage \[[@B2],[@B3]\].
Glutamate has a dual role both as an excitatory neurotransmitter essential for physiological functions and a neurotoxic mediator contributory to pathological processes. Since the homeostasis of extracellular glutamate concentration is critically regulated by neuronal and glial GT, reduced GT expression and/or function would be expected to increase extracellular glutamate concentration with subsequent excessive activation of glutamate receptors and excitotoxicity. Indeed, a large number of studies have shown the detrimental effects from reduced GT expression and function on the pathogenesis of neurological disorders including brain ischemia, epilepsy, spinal cord injury, amyotrophic lateral sclerosis, AIDS neuropathy, and Alzheimer\'s disease.
Given the well-documented role of the glutamatergic system in the mechanisms of pathological pain, it is not surprising that regulation of GT also has been implicated in the central mechanisms of nociceptive processing, such as that following the hindpaw formalin injection or the application of exogenous NMDA or protaglandins in rats \[[@B4],[@B5]\]. A series of recent experiments have demonstrated that the expression of spinal GT was altered following peripheral nerve injury and contributed to neuropathic pain behaviors in rats \[[@B6]\]. The altered GT expression after nerve injury was mediated, at least in part, through a tyrosine kinase receptor (TrkB) and intracellular mitogen-activated protein kinases. Moreover, peripheral nerve injury significantly reduced spinal glutamate uptake activity, supporting a functional role of spinal GT, via regulating regional glutamate homeostasis, in the mechanisms of nerve injury-induced neuropathic pain behaviors \[[@B6]\]. Of interest is that recent studies also have demonstrated that chronic morphine administration regulated the spinal GT expression, which contributed to the mechanisms of morphine tolerance and associated neuronal apoptosis and hyperalgesia in rats \[[@B7]-[@B9]\]. Since neuropathic pain and opioid tolerance have been shown to share a common glutamatergic mechanism \[[@B10]\], these findings indicate that regulation of the GT expression and function would be an important approach to preventing and reversing glutamate over-excitation and neurotoxocity associated with the mechanisms of neuropathic pain and opioid tolerance.
Despite a positive regulatory role of certain compounds such as riluzole and MS-153 in GT function in pre-clinical studies of neuropathic pain and morphine tolerance \[[@B6]-[@B9]\], the exact mechanisms of GT regulation remain unclear. A recent article by Rothstein et al \[[@B1]\] reports that the most commonly used class of antibiotics (β-lactam antibiotics) such as ceftriaxone promoted the expression of GLT1 and demonstrated a functional role in both *in vitro*and *in vivo*models of glutamate neurotocixity. These findings indicate that positive promoters of GT expression may have a unique role in neuroprotection through regulating GT expression. The regulatory role of β-lactam antibiotics in promoting GT expression appears to be selective, because other classes of antibiotics such as vancomycin were ineffective \[[@B1]\]. The results from this study are encouraging in search for new pharmacological tools for neuroprotection and pain management. A potential issue concerning such an approach, however, is the development of antibiotic-resistant bacteria associated with its application, making it less practical to directly use an antibiotic therapy for managing neuropathic pain, opioid tolerance, and neurological disorders. Therefore, it may be expected that future studies would explore the genomic regulation of GT expression promoted by β-lactam antibiotics in order to elucidate the regulatory pathway(s) of GT expression at the molecular level.
In contrast to antagonism of glutamate receptors, a distinct advantage of regulating GT expression and activity is the possibility of minimizing the pathological impact from a glutamate overload while retaining the physiological role of glutamate. A similar approach is used in developing anti-psychiatric drugs, including broad and selective monoamine reuptake agents, and has yielded rather positive outcomes in the treatment of psychiatric disorders. Extensive research is under way to explore the cellular and molecular mechanisms of GT expression and function in relation to the pathogenesis of neuropathic pain. In addition, studies on the role of GT regulation in opioid tolerance and dependence may provide new insights into the cellular mechanisms of substance abuse, an emerging issue associated with clinical opioid therapy.
Acknowledgements
================
This work was supported in part by PHS grants DA08835, NS42661, and NS 45681.
|
PubMed Central
|
2024-06-05T03:55:55.097380
|
2005-2-9
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074355/",
"journal": "Mol Pain. 2005 Feb 9; 1:5",
"authors": [
{
"first": "Jianren",
"last": "Mao"
}
]
}
|
PMC1074356
|
Background
==========
Pain in humans is an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage \[[@B1]\]. The pain experience contains at least two major components: the first is the encoding and perception of sensory noxious stimulus (e.g., pain intensity); and the second is the encoding of the unpleasantness of the noxious stimuli \[[@B2],[@B3]\]. Exploration of the centers for pain-related unpleasantness has recently been carried out in human studies using modern imaging techniques \[[@B4]-[@B8]\]. Among many central regions investigated, the anterior cingulate cortex (ACC) is believed to be a key structure that contributes to pain affect or unpleasantness. Early human observations showed that surgical ablation of the ACC significantly reduced pain unpleasantness without influencing the ability to detect the intensity or location of the pain \[[@B9],[@B10]\]. Rainelle et al \[[@B5]\] reported that specific manipulation of pain unpleasantness produced significant changes in the imaged activity of the ACC, while the manipulation of pain intensity produced changes mainly in the primary somatosensory cortex (S1) \[[@B2],[@B5]\]. More recently, electrophysiological recordings from the ACC in humans found that some ACC neurons respond to noxious stimuli \[[@B6]\]. More interestingly, a recent study reported that the ACC was also activated during social exclusion \[[@B11]\]. In addition to pain, the ACC has been proposed as the neurobiological substrate for executive control of cognitive and motor processes \[[@B12]\]. Human imaging studies demonstrate that the ACC region is activated by different factors including motivational drive, reward, gain or loss, conflict-monitoring or error prediction, and attention or anticipation \[[@B13]-[@B23]\]. The neuronal mechanisms for these different functions within the ACC remain mostly unknown due to the limitation of human studies.
Studies from our group and other investigators, using animal models, provide evidence for the importance of the ACC in behavioural responses related to noxious stimuli \[[@B24]-[@B32]\] and the \"top-down\" descending modulatory effects \[[@B33]\]. Lesion in the medial frontal cortex, including the ACC, significantly reduced the behavioral response to noxious stimuli and aversive memory behaviors \[[@B24]-[@B26]\]. Also, electrophysiological recordings demonstrate that neurons within the ACC respond to noxious stimuli \[[@B6],[@B28]\]. Tissue injury or digit amputation activates immediate early gene expression and triggers long-term potentiation of evoked sensory responses in the ACC \[[@B27]-[@B29]\]. In mice genetically modified to over express NMDA NR2B receptors in forebrain areas, including the ACC, behavioral responses to tissue inflammation were significantly enhanced \[[@B29]\]. Behavioural allodynia related to inflammation was reduced by injection of antagonists of NMDA receptors or inhibitors of cAMP-dependent protein kinases \[[@B30],[@B32]\]. These findings indicate that ACC neurons are clearly involved in the processing of noxious stimuli, and demonstrate activity-dependent long-term plasticity in the ACC after tissue injury. In addition, ACC can also serve as a \"top-down\" descending modulatory system that regulates spinal nociceptive reflexes. Electrical stimulation or chemical injection of glutamate receptor agonists facilitated a spinal nociceptive tail-flick reflex through a descending facilitatory system relayed to the brainstem rostral ventromedial medulla (RVM) \[[@B33]-[@B37]\].
It is difficult to distinguish the role of the ACC in pain-related unpleasantness from its descending pain modulatory effects on sensory transmission in the spinal cord by using behavioral withdrawal responses to noxious stimuli. A recent human imaging study reported that the ACC is activated during placebo analgesia \[[@B7]\]. These results suggest that the ACC may also play roles in placebo analgesia. While the physiological nature of the imaged \'hot\' spots (i.e., excitation of excitatory versus inhibitory neurons) remains to be determined, it has been proposed that the ACC may activate endogenous analgesia systems due to its projections to the periaqueductal gray (PAG) in the midbrain \[[@B7],[@B38]\].
Here we propose that the ACC serves as a region for pain unpleasantness in the brain, and excitation of neurons in the ACC can trigger pain unpleasantness but not analgesia. Because animals will never report human-like \'unpleasantness\', we used a classic Pavlovian fear memory to measure the effects of stimulation in the ACC. Our operational definition of \'unpleasantness\' in mice is based on the formation of fear memory. If ACC stimulation triggers \'unpleasantness\' in mice, we expect to observe behavioural freezing responses in mice receiving paired but not unpaired fear-conditioning training. Another obvious advantage of using the Pavlovian fear memory model is to avoid using behavioural withdrawal responses (e.g., hind paws or tail) that are constantly under descending inhibitory and facilitatory modulatory influences. Experiments were performed in both mice and rats to test the hypothesis.
Results
=======
We employed direct focal electrical stimulation of the ACC in freely moving adult mice and used fear memory to test if the ACC encodes for pain-related unpleasantness. Animals were implanted with stimulating electrodes in the ACC, and behavioural measurements were performed at least 2 weeks after the surgery. Mice were placed in a specially designed sound insulated chamber. After at least 30 min of baseline observation, a brief electrical current was delivered to the ACC through the implanted electrodes. We monitored ultrasonic activity, another index of emotional responses \[[@B39]\], in some of the mice throughout the experiments. Ultrasonic responses were increased during ACC stimulation at 0.3 mA (n = 6 mice; *P*\<*0.05*as compared with stimulation at 0 mA; see Fig. [1](#F1){ref-type="fig"}). Significant changes were detected for both the frequency and duration of ultravocalization events (Fig. [1b](#F1){ref-type="fig"} and [1c](#F1){ref-type="fig"}). The ultrasonic activity induced by electrical stimulation was intensity dependent. Stimulation at 0.01 mA produced almost no significant changes, while stimulation at 1.0 mA induced greater responses (n = 6 mice, *P*\<*0.01*as compared with stimulation at 0.1 or 0.3 mA; Fig. [1d](#F1){ref-type="fig"}). Ultrasonic activity was found mostly between 30--50 kHz. This frequency range of ultrasonic activity was also increased by a noxious foot shock and by a painful chemical injection of capsaicin. No obvious changes in motor activity or seizure-like activity were found in mice after receiving the focal ACC stimulation.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**ACC stimulation induces ultrasonic vocalization in freely moving mice.(a)**An example of ultrasonic responses from a single mouse at four different frequencies before, during and after ACC stimulation (at 0.3 mA). 1 min duration; see filled circle for the stimulation site within the ACC in (e); **(b)**ACC stimulation (0.3 mA; n = 6) increased the frequency of individual ultrasonic responses; \* *P*\<*0.05*, comparing the frequency during ACC stimulation with baseline response before the stimulation; **(c)**ACC stimulation (0.3 mA; n = 6) also increased the duration of single ultrasonic response; \* *P \< 0.05*, comparing the duration during ACC stimulation with baseline duration before the stimulation; **(d)**Summarized data of ACC stimulation (n = 6) produced ultrasonic responses at different intensities. Total vocalization responses (sec) within 2 min ACC stimulation were plotted against the intensity of stimulation; **(e)**Stimulating sites in ACC on the schematic representation of coronal section 0.62 mm anterior to the Bregma. Filled circle, for data shown in (a); open circles, other sites for data shown in (b-d).
:::
:::
Next, we wanted to test if the fear-like freezing responses induced by ACC stimulation were due to unpleasantness (similar to that caused by a noxious foot shock). If so, we would expect to see long-term fear memory induced by ACC stimulation. We paired ACC electrical stimulation with a tone presented in a conditioning chamber (see Fig. [2a](#F2){ref-type="fig"}). After pairing ACC stimulation with the tone, long-term fear memory was detected in most of the mice (n = 16 of 21 mice; 76.2%), both to the tone presentation in a novel context (auditory memory) (Fig. [2b](#F2){ref-type="fig"}) and to the conditioning environment (contextual memory) (Fig. [2c](#F2){ref-type="fig"}). In the other animals, there is no specific freezing either to the conditioning context or to the tone (Fig. [2d](#F2){ref-type="fig"}; n = 5 mice). We did not find any clear anatomic separation between effective versus ineffective stimulation sites in the ACC for producing fear memory (Fig. [2d](#F2){ref-type="fig"}). There are two possible explanations for this result. One is that some of sites may have a higher stimulation threshold for inducing fear memory. Alternatively, ACC neurons are functionally heterogeneous and some sites within the ACC do not contribute to the unpleasantness of pain.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**ACC stimulation induces long-term fear memory.(a)**Three pairings of 30 s tone and 10 s electrical train stimulation were delivered in the paired group on the conditioning day; **(b**,**c)**Percentage freezing to the tone (**b**) and the conditioning context (**c**) measured at 1 hr, 1 day and 3 days after paired training. After paired training, long-lasting fear memory was detected in most of the mice (n = 16 mice, filled circles), while some other mice showed no freezing across the test periods (n = 5, data not shown). After unpaired training of tone and ACC stimulation, mice showed no freezing to the tone (n = 6, open squares) but clear memory to the training environment; \* *P \< 0.05*compared with the unpaired group;**(d)**Stimulating sites in ACC on the schematic representation of coronal section 0.62 mm anterior to the Bregma. Filled circles, effect sites of ACC paired group; open circles, no effect sites of ACC paired group; open squares, ACC unpaired group.
:::
:::
Pavlovian fear memory occurs when a subject learns to associate a certain conditioned stimulus (CS) or cue with a noxious unconditioned stimulus (US) \[[@B40]-[@B43]\]. To determine whether ACC stimulation-induced fear memory results from association between the tone and ACC stimulation, we performed experiments where the tone and ACC stimulation were applied but unpaired in another group of animals (Fig. [2b--c](#F2){ref-type="fig"}). We predicted that while similar contextual memory may form (since animals received the same amount of ACC stimulation in the same environment), auditory memory, which requires the precise pairing of US-CS, would be blocked. Indeed, mice receiving unpaired training demonstrated clear freezing response in the conditioning context (n = 6 mice; *P \< 0.05*versus baseline responses), whereas no freezing response was observed during tone presentation in the novel context (*P \> 0.05*, comparing behavioral responses before and during the tone; Fig. [2b---c](#F2){ref-type="fig"}).
Glutamate receptors including metabotropic glutamate receptors (mGluRs) are found in the ACC \[[@B33],[@B38]\]. To further determine if local activation of glutamate receptors in the ACC may also trigger similar responses as focal electrical stimulation, we performed microinjection of a mGluR agonist tACPD (0.25 μg in 0.5 μl) \[[@B33]\]. Microinjection of tACPD into the ACC produced freezing responses within 10 min, indicating that activation of ACC neurons induced freezing responses (n = 5 mice). Furthermore, we also paired the tACPD microinjection with a tone (see Fig. [3a](#F3){ref-type="fig"}) to test if tACPD application with tone may also cause fear memory. We measured fear responses at one and three days the conditioning. We found that tACPD microinjection pairing produced significant behavioral freezing responses at one and three days later (n = 5 mice; Fig. [3b](#F3){ref-type="fig"} and [3c](#F3){ref-type="fig"}). Mice receiving the tACPD microinjection in the ACC did not cause any visible abnormal motor hyperactivity or seizure-like behaviors, although we cannot completely rule out other sub-threshold changes caused by the microinjection.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Activation of metabotropic glutamate receptors in the ACC caused long-term fear memory.(a)**ACC tACPD microinjection (0.25 μg in 0.5 μl; indicated by an arrow) was paired with a tone (indicated by filled bar) on the conditioning day; **(b, c)**Percentage freezing to the tone (**b**) and the conditioning context (**c**) measured at 1 day and 3 days after paired training (n = 5 mice). \**P \< 0.05*compared with the control. **(d)**The sites in the ACC for tACPD microinjection on the schematic representation of coronal section 0.62 mm anterior to the Bregma.
:::
:::
Next, we asked whether stimulation-induced unpleasantness is regional selective. If ACC activation is specifically involved in fear conditioning, electrical stimulation outside of the ACC would not produce similar results. To test this possibility, stimulating electrodes were implanted into the primary somatosensory cortex (S1). It has been well documented that somatosensory cortex contributes to the central processing of sensory inputs, including pain intensity (e.g., noxious stimulus parameters) \[[@B44]\]. As shown in Fig. [4](#F4){ref-type="fig"}, this group of mice did not form fear memory, even though they underwent exactly the same conditioning procedures as the paired groups (*P \> 0.05*versus baseline; n = 5 mice).
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Primary somatosensory cortex (S1) stimulation produces no long-term fear memory.(a, b)**Three pairings of 30 s tone and 10 s electrical train stimulation were delivered to somatosensory cortex (S1) caused neither long-term auditory (a) nor contextual memory (b) (n = 5). **(c)**Sites for stimulation in the S1.
:::
:::
Amygdala and its related structures are well known for the induction of fear memory \[[@B40]-[@B43]\]. While the prefrontal cortex has been recently implicated for its involvement in the extinction of fear memory \[[@B45]-[@B47]\], it has not been shown that ACC inputs to the amygdala may contribute to the formation of fear memory. To test if fear-like long-term memory induced by ACC conditioning requires the involvement of amygdala, we selectively blocked the NMDA receptors during training (ACC stimulation paired with the environment or a tone). The NMDA receptors are known to be important for the induction of fear memory \[[@B48],[@B49]\]. We implanted additional guide cannulas into the basolateral amygdaloid complex (BLAC) for the microinjection of drugs (Fig. [5c](#F5){ref-type="fig"}). Before conditioning, the NMDA receptor antagonist AP-5, was injected bilaterally into the BLAC. At 15 min later, ACC stimulation was paired with the tone presentation. For the control group, the same volume of saline was injected. As showed in Fig. [5a](#F5){ref-type="fig"}, AP-5 significantly reduced ACC stimulation-induced fear memory to the tone 1 day after conditioning (*P \< 0.05*versus saline injected group; n = 5 mice), indicating that the NMDA receptor function in the amygdala is required for the formation of auditory fear memory. Importantly, the contextual memory was not affected (n = 5 mice; Fig. [5a](#F5){ref-type="fig"}), supporting the notion that other structures such as the hippocampus may contribute to contextual fear memory.
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**NMDA receptors in the amygdala is required for auditory fear memory induced by ACC stimulation.(a)**Bilateral microinfusions of AP-5 (2 μg/μl, 0.5 μl/side) in the BLAC 15 min before conditioning impaired auditory fear memory but no effect on contexual fear memory when tested 1 day later. Filled bars, AP-5 group (n = 5); open bars, saline group (n = 4). \* P \< 0.05 compared with saline-injected group. **(b)**Stimulation sites in ACC. Filled circles, AP-5 group; open circles, saline group. **(c)**Microinjection sites in the BLAC on the schematic representation of coronal section 1.94 mm posterior to the Bregma. Filled circles, AP-5 group; open circles, saline group.
:::
:::
These results suggest that ACC stimulation produces unpleasantness in mice and form fear auditory memory through its interaction with the amygdala. Both animal and human studies indicate that neurons in the ACC respond to peripheral painful stimuli \[[@B6],[@B28],[@B50]\], and injury triggers the activation of immediate early genes in the ACC \[[@B27],[@B29]\]. Unlike the somatosensory cortex, neurons in the ACC have wide diffuse receptive fields, and often contain the whole body of an animal, supporting its role in coding unpleasantness of pain \[[@B6],[@B51],[@B52]\]. Supporting this hypothesis, lesions of the ACC or inhibition of excitatory synaptic transmission in the ACC produces antinociceptive effects or analgesia \[[@B24],[@B25]\] and block formalin-induced conditioned place avoidance \[[@B26]\]. These results argue that activation of the ACC or increased the excitatory activity within the ACC is unpleasant or painful, acting directly or indirectly through other related brain regions. However, a recent imaging study reported that the activity in the ACC was increased during placebo analgesia \[[@B7]\]. We think that the activation of endogenous analgesia system is unlikely to account for placebo analgesia. Previous studies showed that stimulation of ACC did not trigger the endogenous analgesia system \[[@B33]\]. Rather, we propose that enhanced inhibitory activity may contribute to the placebo effects. To test if increasing inhibitory transmission in the ACC may relieve unpleasantness due to foot shock, we microinjected a GABA~A~receptor agonist, muscimol into the ACC prior to the fear conditioning. Muscimol significantly reduced the fear memory score 1 day and 3 days after classical fear conditioning (*P \< 0.05*versus saline injected group, n = 6 mice for each group) (Fig. [6a--c](#F6){ref-type="fig"}).
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**ACC inactivation by a GABA~A~receptor agonist impairs fear memory by foot shock.(a**,**b)**Mice receiving muscimol microinjection (1 μg/μl, 0.5 μl/each side, n = 6 mice, black bars) into bilateral ACC 15 min before conditioning showed reduced auditory (a) and contextual (b) fear memory induced by classic foot shock conditioning). \* *P \< 0.05*compared with the saline treated group (n = 6, open bars). **(c)**Microinjection sites in the ACC. Filled circles: muscimol group; open circles, saline group.
:::
:::
One possible explanation is that activation of ACC neurons is aversive in the absence of peripheral noxious stimuli, and is antinociceptive in case of peripheral noxious stimuli. To test this possibility, we decided to test if activation of the ACC alters animals\' avoidance responses using the modified noxious hot-plate escape test. In this test, mice learn to escape the noxious area by moving into an unheated area. In the first trail, mice found the escape route to the unheated area within 79.6 ± 17.7 sec after being placed on the 50°C hot-plate, (n = 7 mice, Fig. [7a](#F7){ref-type="fig"}). After repetitive training (n = 3 in total, with a 10 min interval), mice learned to move into the \'safe\' area within a significantly shorter period of time (12.1 ± 1.8 sec; *P \< 0.01*compared with the first measurement). Furthermore, at 24 hours after training, mice quickly moved into the safe area within a similar time (mean 8.1 ± 1.8 sec, n = 7 mice; Fig. [7b](#F7){ref-type="fig"}). Next, we wanted to test if activation of the ACC may affect behavioral escape responses. tACPD (0.25 μg in 0.5 μl) was microinjected into the ACC before the training. While the escape time at the first trial was not significantly affected by tACPD microinjection (n = 7 mice, mean 70.1 ± 8.0 sec), mice moved into the safe area within a significantly shorter period of time during the second and third trials (*P \< 0.01*as compared with control mice). Furthermore, at 24 hours after the training, mice moved into the safe area significantly faster than control mice (*P \< 0.05*, Fig. [7b](#F7){ref-type="fig"}). To determine if the effects of tACPD may affect subsequent extinction, we repetitively exposed both groups of mice to the same plate at room temperature (22°C) at 30 min intervals. Mice spent significantly longer time in the plate after learning that the plate was not hot or noxious (n = 7 mice; Fig. [7c](#F7){ref-type="fig"}). As shown in Fig. [7c](#F7){ref-type="fig"}, both groups of animals showed similar \'learning\' responses. These results indicate that local activation of mGluRs in the ACC selectively enhanced the learning ability to avoid the noxious plate.
::: {#F7 .fig}
Figure 7
::: {.caption}
######
**Activation of mGluRs in the ACC facilitated escaping behavioral responses in a hot-plate.(a)**A diagram explaining a new behavioural escape test using the modified hot-plate instrument. **(b)**Mice receiving tACPD microinjection into the unilateral ACC 10 min before training showed faster escape response. \* *P \< 0.05*compared with the control group (n = 6 mice, open squares). **(c)**Extinction responses in control and tACPD-treated groups were similar. MPE were calculated as: (response latency -- baseline latency)/(180 -- baseline latency). 100% MPE indicates that mouse stayed in the same plate for 3 min without moving into the safe area.
:::
:::
To examine the effects of microinjection of tACPD in the ACC on behavioral nociceptive responses, we examined the effects of chemical activation of mGluRs in the ACC on the spinal nociceptive tail-flick reflex and the hot-plate test in awake mice. Microinjection of tACPD into the ACC produced facilitation of spinal tail-flick reflex (n = 7 mice; Fig. [8b](#F8){ref-type="fig"}). Furthermore, the response latency in the hot-plate test was also significantly reduced (n = 5 mice, Fig. [8a](#F8){ref-type="fig"}). These results consistently suggest that activation of mGluRs in the ACC is able to facilitate or enhance behavioral responses to noxious stimuli. Previous studies in anesthetized rats demonstrate that descending facilitation induced by electrical/chemical activation in the ACC depend on central relays in the brainstem RVM \[[@B33]\]. Descending serotonergic systems originated from the brainstem raphe nucleus as well as adjacent areas are thought to be important for descending facilitation of spinal nociceptive transmission and reflexes \[[@B34]-[@B37]\]. To examine if spinal serotonergic receptors are required for tACPD-induced facilitation, we directly injected a serotonergic receptor antagonist into the spinal cord through intrathecal catheters in freely moving adult rats. Similar to adult mice, microinjection of tACPD into the ACC produced a significant reduction in the tail-flick responses latencies in rats (n = 3 rats, mean 5.8 ± 0.8 sec vs 3.2 ± 0.1 sec, *P \< 0.01*, paired t-test). Intrathecal administration of methysergide at a dose of 32.0 nmoles (10 μl, i.th.) that blocked descending facilitatory effects from the RVM \[[@B34]\] completely abolished the facilitation induced by tACPD in the ACC (n = 3 rats; Fig. [8c](#F8){ref-type="fig"}). To avoid possible side effects of stress, we also performed experiments in anesthetized rats. As reported previously \[[@B33]\], microinjection of tACPD produced significant facilitation of the tail-flick response latency in rats (n = 4 rats, mean 6.3 ± 0.3 sec vs 5.3 ± 0.5 sec, *P \< 0.05*; Fig. [8d](#F8){ref-type="fig"}). Methysergide (32.0 nmoles/10 μl) injected intrathecally blocked the facilitatory effects produced by tACPD in the ACC (n = 4 mice, Fig. [8d](#F8){ref-type="fig"}). These results consistently indicate that activation of the ACC causes facilitation or anti-analgesic effects in case of noxious stimulation.
::: {#F8 .fig}
Figure 8
::: {.caption}
######
**Spinal serotonin receptors contribute to descending facilitatory modulation from the ACC.(a, b)**Microinjection of tACPD (0.25 μg in 0.5 μl) in the ACC facilitated the hot-plate responses (a, i.e., reduced response latency) and spinal nociceptive tail-flick reflex (b) in freely moving mice; **(c, d)**In freely moving (c) or anesthetized rats (d), tACPD microinjection in the ACC also facilitated the tail-flick reflex. Intrathecal injection of a serotonergic receptor antagonist methysergide (32.0 nmoles/10 μl) at a dose that blocked descending facilitation also blocked the facilitation of the tail-flick reflex.**(e)**A model explaining the neuronal pathways, which contribute to ACC activation, produced fear memory and descending facilitatory modulation of spinal nociception.
:::
:::
What about the possibility that synaptic transmission within the ACC may contribute to memory formation as those reported in the amygdala? We suspect that synaptic changes between two sides of the ACC connected by callosal projection fibers may serve as an ideal candidate for the storage of such information, since the thalamic-cortical connections are important for ascending nociceptive sensory transmission. In order to detect possible changes in the ACC, we performed recordings of synaptic responses in the ACC before and after fear conditioning. As shown in Fig. [9](#F9){ref-type="fig"}, we found that synaptic responses to stimulation of the callosal projection fibers from the other side of the ACC were not affected by fear conditioning (n = 5 mice). Furthermore, measurements of the same responses at one day and 3 days after fear conditioning also did not reveal any significant changes (Fig. [9](#F9){ref-type="fig"}), indicating that synaptic transmission between the ACCs did not undergo synaptic potentiation after fear conditioning. To test if synaptic transmission may undergo potentiation by theta burst stimulation, we also tested responses in freely moving mice after applying theta burst stimulation (TBS) locally into the ACC (n = 3 mice). As shown in Fig. [9](#F9){ref-type="fig"}, we found that synaptic transmission was enhanced for the initial 3 hours.
::: {#F9 .fig}
Figure 9
::: {.caption}
######
**Fear conditioning did not cause long-term plasticity in ACC-ACC synapses.(a)**Evoked fast responses in the ACC by stimulation applied to the other side of ACC were not affected by single foot shock that induced classic fear memory. However, TBS induced synaptic potentiation in the ACC. Insets: Traces of evoked responses before, 3 hour, 1 day and 3 days after fear conditioning.**(b)**Summarized data of experiments shown in (a) and mice receiving the treatment without the foot shock (n = 2 mice).
:::
:::
Discussion
==========
We present strong evidence that the ACC serves as a critical region for pain unpleasantness in the brains of adult mice. Our results are consistent with previous reports from human imaging that show that ACC is important for pain affect or unpleasantness \[[@B2],[@B3],[@B5]\]. Clinical reports show that lesions in the human ACC selectively influence the unpleasant component of pain \[[@B9],[@B10]\]. In animals, lesions in the ACC blocked formalin-induced conditioned place avoidance \[[@B26]\]. Both animal and human studies indicate that neurons in the ACC respond to peripheral painful stimuli or electric shocks \[[@B6],[@B28]\], and injury triggered activation of immediate early genes in the ACC \[[@B27],[@B29]\]. Unlike neurons in the somatosensory cortex, neurons in the ACC tend to have widely diffuse receptive fields, and often contain the whole body of animals, supporting its role in coding unpleasantness of pain. Our results using fear memory to measure pain triggered by ACC stimulation avoid the possible contribution of descending modulation to commonly used behavioural nociceptive responses \[[@B33]-[@B37]\]. The present study provides a new approach to study the role of pain in higher brain function.
The ACC is a complex and heterogeneous cortex. Neurophysiological recordings, neuropsychological tests and human imaging studies suggest that the ACC plays a key role in cognitive control and is involved in response conflict monitoring \[[@B12],[@B14],[@B19]\]. The ACC may be involved in the neural representation of motivational drives, including sexual desire, hunger and the motivational aspect of pain \[[@B13],[@B22]\]. However, synaptic and molecular mechanisms of the ACC in these higher order functions are largely unknown, due to the lack of animal models. The different roles of the ACC in various functions require caution when explaining the current results. One possible explanation is that ACC neurons express pain affect, general unpleasantness, aspects of cognition and motor response, and pain analgesia. These different functions might be topographically organized along the extent of the ACC, and through its potential interactions with other cortical areas. Furthermore, possible differences between mouse and human brains certainly contribute to some of the different reports on the functions of human ACC vs mice as presented here. For example, Shidara and Richmond \[[@B15]\] reported that neural activity in the ACC codes the degree of reward expectancy \[[@B15]\]. It is unclear if any of those neurons are also involved in encoding pain unpleasantness. Due to the important role of the ACC in attention and anticipation, another possibility is that ACC stimulation alters learning by affecting such processes. Specifically, the ACC stimulation produces some anticipatory \"state\" for unpleasantness rather than one of aversion or unpleasantness. To test this possibility, we performed experiments using a newly developed pain-related hot-plate escaping test. Chemical activation of excitatory mGluRs in the ACC enhanced the escape responses. These results suggest that enhanced escape responses and behavioural freezing responses are most likely due to pain-related unpleasantness, and are unlikely to be explained by simply the cessation of activity or goal directed activity. It is also possible that ACC stimulation only causes pain, and the unpleasantness is coded in other regions of the brain. For fear memory, the projection from the ACC to the amygdala plays an important role. Future experiments are clearly needed to address these possibilities. Interestingly, a recent study from Johansen and Fields \[[@B53]\] showed that excitatory amino acid microinjection into the ACC in rats during conditioning produced avoidance learning in the absence of a peripheral noxious stimulus. These findings indicate that that neurons in the ACC of adult rats and mice mediate both pain-induced negative affect and a nociceptive aversive teaching signal (see \[[@B53]\]).
Insular cortex has recently been shown to contribute to pain perception \[[@B38],[@B54]\]. It was shown that the insular cortex may tonically control spinal nociceptive transmission through descending inhibitory systems \[[@B54]\]. By contrast, it has been reported that stimulation in the ACC produced no antinociceptive effect, and facilitated spinal nociceptive tail-flick reflex \[[@B33]\]. Inhibition of excitatory transmission in the ACC by microinjection of opioids produced powerful analgesic effects in freely moving animals \[[@B25]\], suggesting that the ACC is unique as a centre for unpleasantness. Recent studies in knockout mice for adenylyl cyclases1 and 8 as well as transgenic mice over expressing NR2B receptors consistently indicate that activity-dependent plastic signaling pathways in the ACC may contribute to persistent pain, a classic condition with long-term pain-related unpleasantness \[[@B29],[@B30]\]. One possibility is that ACC stimulation may activate endogenous analgesia from the ACC, in the absence of nociception, might itself be aversive. To test this possibility, we performed experiments in awake mice and rats using the spinal nociceptive tail-flick reflex, a classic behavioural test for endogenous analgesic/antinociceptive systems \[[@B33]\]. Activation of mGluRs in the ACC by tACPD actually facilitated behavioural responses in both the tail-flick reflex and hot-plate tests, providing direct evidence that endogenous facilitatory but not inhibitory systems are activated. Moreover, intrathecal injection of a serotonergic receptor antagonist methysergide that blocked descending facilitatory modulation from the RVM \[[@B34],[@B38]\] completely blocked the facilitatory effects in awakened or anaesthetized rats. These findings thus provide the first evidence, to our knowledge, in freely moving animals, excitatory activity in the ACC exert descending facilitatory modulation on spinal nociception
Experiments using lesions in the central nervous system provide important evidence for the involvement of certain structures in learning and memory \[[@B55]\]. However, when using the lesion technique it is difficult to distinguish between roles of the ACC in the expression of freezing responses, or the formation of fear memory. In the present study, we performed a reversible blockade of ACC activity during fear conditioning and found that fear memory was blocked. Our results indicate that the blockade of fear memory is not simply due to the ACC-dependent expression of freezing. One possible mechanism for the roles of the ACC in fear memory is that inputs from the ACC feedback to the amygdala for the formation of fearful memory. Indeed, many ACC neurons directly project to the amygdala \[[@B56],[@B57]\]. We found that auditory fear memory induced by pairing the ACC stimulation with a tone was blocked by bilateral injection of the NMDA receptor antagonist AP-5 into the amygdala. Contextual memory was not significantly affected, suggesting that other structures may be involved or required for the expression of contextual fear memory \[[@B55]\]. Our results provide new evidence that cortical input from the ACC (coding unpleasantness) is critical for the formation of fear memory (see Fig. [7e](#F7){ref-type="fig"}). Due to limitations of the microinjection technique, we cannot completely rule out the possible diffusion of AP-5 into other nuclei within the amygdala. Finally, we believe that mouse genetic models will provide ample opportunities for us to explore the molecular mechanisms for high-order brain functions in experimental conditions. Together with imaging and electrophysiological studies in humans and primates, we hope to reveal new central and molecular targets for the treatment of central pain, phantom pain and the unpleasantness related to various mental disorders.
Methods
=======
Animals
-------
Animals were adult male C57BL6/J mice or Sprague-Dawley albino rats that were housed individually and maintained on a 12/12 h light/dark cycle. Food and water were provided *ad libitum*.
Brain electrode and microinjection cannula implantation
-------------------------------------------------------
Animals were anesthetized with sodium pentobarbital (80 mg/kg, i.p.) and implanted unilaterally with a tungsten electrode in the ACC (0.62 mm anterior, 0.4 mm lateral and 1.7 mm ventral to the Bregma) or S1 (0.62 mm anterior, 2.8 mm lateral and 2.3 mm ventral to the Bregma) \[[@B58]\] under aseptic conditions. The tungsten wire extended 0.1--0.2 mm of the guide cannula that was used as a reference. Mice were allowed to recover for 2 weeks before the experiments. In preliminary experiments, we performed several experiments to locate electrodes into the mouse ACC. All procedures were in accordance with the Animal Studies Committee at the University of Toronto. At the end of the experiment, using standard histological methods, 30 μm brain sections were stained with cresyl violet and examined by light microscopy for electrode placements or cannula penetrations. We only used animals with stable implantation of electrodes that did not exhibit abnormal behaviour during surgical recovery. Results from stimulation sites outside of the ACC were not used in the current studies.
Behavioral conditioning
-----------------------
On the day of conditioning, the electrode assembly was connected to the stimulating hardware under brief isoflurane sedation. Mice were allowed 5 min to recover and habituate in the mouse conditioning chamber located in a sound-attenuating box (Med Associates). A commutator (CRIST INSTRUMENT CO.) was used to handle the connecting wires while mice were moving. Each mouse received 3 pairings of training with 1 min in between. The conditioned stimulus (CS) was a tone (2.8 kHz, 85 dB sound pressure level, 30 s), and the unconditioned stimulus (US) was the 10 s ACC stimulation that co-terminated with the tone. The electrical stimulation parameters are as following: 0.3 mA, 0.2 ms pulse duration, 5 pulses at 100 Hz per train, 200 ms train interval. 1 hour, 1 day and 3 days later, animals were first exposed to the conditioning context without tone for 3 min and then to a novel context without tone for 3 min followed by 3 min with tone presentation. Freezing responses was scored visually and presented as percentage of the total period of time observed \[[@B31]\]. Baseline responses were subtracted in order to evaluate responses to the context or tone. For the unpaired training, tone and ACC stimulation were delivered randomly with 40 s -- 80 s interval. In some experiments, tACPD (0.25 μg in 0.5 μl) was injected into the ACC unilaterally. After the injection, mice were placed in the conditioning chamber. Each mouse received a tone (see above) for 10 min (at 0--5 and 10--15 min after the tACPD injection).
Pharmacological treatment
-------------------------
Under anesthesia of sodium pentobarbital (80 mg/kg), 25-gauge guide cannulas were implanted bilaterally into the ACC (0.62 mm anterior to Bregma, 0.5 mm lateral from the midline, 0.9 mm beneath to the surface of the skull) or BLAC (1.94 mm anterior to Bregma, 3.5 mm lateral from the midline, 4.2 mm beneath to the surface of the skull). Mice were given at least 2 weeks for recovery after cannula implantation. 30-gauge injection cannula was 0.8 mm lower than the guide. For intra-ACC infusion, 0.5 μl muscimol (1 μg/μl) or saline was delivered bilaterally within 90 s using a pump. 15 min later mice were conditioned by 1 pairing of a tone (2.8 kHz, 85 dB, 30 s) and a foot shock (0.75 mA, 2 s) that terminated simultaneously with the tone. For intra-BLAC infusion, 0.5 μl AP5 (2 μg/μl) or saline was delivered bilaterally within 90 s. 15 min later mice received 3 pairings of tone and ACC stimulation, exactly as above.
Hot-plate escape test-a new behavioral test
-------------------------------------------
Adult mice were trained using a modified thermal hot-plate (10\" × 10\" heating surface) (Columbus Instruments, Columbus) maintained at 50°C, with an escapable non-thermal platform (8\" × 5.5\" surface). During the first training trial, the escape platform was blocked until the mice showed signs of nociception (e.g., licking of the hind paws). The escape route was then unblocked, and the mice were then free to explore both platforms. The first time entry to the escape platform was recorded. The total duration of each training trial was 3 min. Mice were returned to the same modified thermal hotplate one day after training, with the escape route remaining open, and the temperature set at a room temperature of 22°C. The first time entry to the escape platform was recorded, and the mice were returned back to their home cage upon escape. The maximum test time was 3 min. Mice were tested a total of eight times with an inter-trial interval of 30 min. Mice that remained on the hotplate for the total test duration were recorded with having an escape time of 3 min.
Spinal nociceptive tail-flick reflex
------------------------------------
Mice or rats were used either in the awake or halothane-anesthetized states. The spinal nociceptive tail-flick reflex was evoked by noxious radiant heat provided by a 50 W projector lamp focused on a 1.5 × 10 mm area on the underside of the tail. The latency to reflexive removal of the tail from the heat was measured by a digital photocell timer to the nearest 0.1 s. Baseline TF latency was the mean of 3 trials taken at 3 min intervals. For experiments using anesthesia, rats were anesthetized with 2--3% halothane (Ohio Medical Products) delivered via a specialized gas adapter (Stoleting Instrument; IL) with 30% O~2~balanced with nitrogen. Body temperature was maintained at 37 ± 0.5°C by a circulating water, thermostatically-controlled heating pad.
Chemical microinjection into the ACC
------------------------------------
A mGluR agonist, trans-(±)-1-amino-(1S, 3R)-cyclopentanedicarboxylic acid (tACPD), was microinjected into the ACC in a volume of 0.5 μl via an injection cannula (33-gauge, 0.20 mm O.D.) inserted through the 26-gauge guide cannula and also extending 2 mm beyond its tip. Injection of tACPD was monitored by following the movement of an air bubble in a length of calibrated tubing between the syringe and the cannula.
Intrathecal drug injection
--------------------------
Intrathecal catheters (PE-10 tubing, 8.5 cm in length) were inserted through a small opening in the cisterna magna and extended to the lumbar subarachnoid space. In case of experiments in freely moving rats, animals were recovered for at least two weeks before the testing. Intrathecal drug administration was done in the awake or halothane-anesthetized rats. In rats with ACC drug microinjection, baseline TF latencies were determined at 3, 6 and 9 min before and after tACPD injection. After observing the facilitatory effects at 10 min after the injection, methysergide or saline were injected intrathecally to examine if the facilitatory effects may be blocked. The selection of methysergide and dose used are based on previous studies of descending facilitatory modulation from the RVM to the spinal cord \[[@B33],[@B38]\].
Electrophysiological recordings in freely moving mice
-----------------------------------------------------
Mice were anesthetized with sodium pentobarbital (80 mg/kg, i.p.). A concentric stimulating electrode was positioned in the right ACC and another concentric recording electrode was placed to the left ACC (both were in Cg2). Dental cement was used to keep the electrodes in place for the recordings of field potential in freely moving mice. Following surgery, mice were allowed to recovery for at least two weeks. Test responses were elicited by monophasic stimuli (200 μs, 75--190 μA, 1/60 s) at an intensity that evoked 40--50% of the maximal responses. fEPSP potentiation is expressed as percentage change relative to the mean baseline response during the 30 min prior to the single footshock stimulation.
Data analysis
-------------
Data are presented as mean ± 1 standard error of mean (S.E.M). Facilitation of the TF reflex or hot-plate test is presented as a percentage of the control TF latency. Results were expressed as mean ± s.e.m. One-way ANOVA or two-way ANOVA with repeated measurements was used to compare the differences between treatments. If not stated otherwise, post-hoc comparisons were made with Tukey test. In all cases, p \< 0.05 was considered statistically significant.
Acknowledgements
================
Supported by grants from the EJLB-CIHR Michael Smith Chair in Neurosciences and Mental Health, Canada Research Chair, NIH NINDS NS42722 and Canadian Institutes of Health Research.
|
PubMed Central
|
2024-06-05T03:55:55.098344
|
2005-2-9
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074356/",
"journal": "Mol Pain. 2005 Feb 9; 1:6",
"authors": [
{
"first": "Jianrong",
"last": "Tang"
},
{
"first": "Shanelle",
"last": "Ko"
},
{
"first": "Hoi-Ki",
"last": "Ding"
},
{
"first": "Chang-Shen",
"last": "Qiu"
},
{
"first": "Amelita A",
"last": "Calejesan"
},
{
"first": "Min",
"last": "Zhuo"
}
]
}
|
PMC1074357
|
Background
==========
Opioid analgesics, such as morphine, currently are the most effective and frequently used pain reliever for moderate to severe pain. However, long-term administration of opioids can alter the central pain-related systems and results in opioid tolerance (decreased analgesic effect of opioids) and opioid dependence (a behavioral state requiring continued opioids to avoid a series of aversive withdrawal syndromes). Opioid tolerance and dependence significantly hamper the effective treatment of chronic pain with opioid analgesics \[[@B1]\]. Numerous agonists and antagonists of various receptors and inhibitors of second messenger pathways have been reported to block or reduce morphine tolerance and/or dependence \[[@B2]\]. It has been well established that glutamate receptors are critical in the development and maintenance of opioid tolerance \[[@B3]-[@B6]\]. However, the underlying mechanisms by which glutamate receptors mediate opioid tolerance and dependence remain unclear. An upregulation of the cAMP/PKA signaling pathway has been characterized as a typical molecular adaptation in several brain regions following chronic morphine treatment \[[@B1]\], but the detailed role of the cAMP pathway in analgesic tolerance to chronic opioids has yet to be demonstrated.
Nucleus raphe magnus (NRM), a key medullary relay for descending pain modulation, is critically involved in opioid-induced analgesia \[[@B7]\]. According to their electrophysiological characters and opioid responses, NRM neurons in an *in vitro*preparation have been divided into two general types, *primary cells*that lack the μ-opioid receptor and *secondary cells*that contain the μ receptor \[[@B8]\]. Based on the observation that acute opioids inhibit GABA synaptic transmission in primary cells, we have proposed that opioids produce analgesia in the NRM by disinhibiting or activating those primary cells that send descending projections to the spinal dorsal horn and inhibit spinal pain transmission \[[@B8],[@B9]\]. Several lines of evidence suggests that some NRM cells that are directly inhibited by opioids or contain μ receptors have a facilitating action on spinal pain transmission through their descending projections \[[@B7],[@B10],[@B11]\]. Thus, both activation of pain-inhibiting primary cells and inhibition of pain-facilitating secondary cells in the NRM may be involved in acute opioid-induced analgesia. The synaptic connections between primary cells and secondary cells and the neurotransmitter they release are currently unknown. Accumulating evidence has clearly demonstrated that the μ receptor-containing cells in the NRM are activated in many chronic pain conditions with pain sensitization \[[@B11]-[@B13]\], but the activation mechanisms remain unclear. The present study was aimed to investigate chronic morphine-induced adaptation of glutamate synaptic transmission in NRM neurons from morphine-tolerant rats and the intracellular signaling pathways involved in the synaptic adaptation.
Results
=======
Chronic morphine selectively increases presynaptic glutamate release
--------------------------------------------------------------------
Glutamate-mediated excitatory postsynaptic currents (EPSCs) were recorded under whole-cell voltage-clamp in NRM slices *in vitro*. The EPSCs were compared between NRM slices from saline-treated control rats and those from morphine-treated tolerant rats. Both groups of slices (control and tolerant) were maintained in 5 μM morphine throughout recording experiment *in vitro*to prevent morphine withdrawal (Ingram et al., 1998). A separate group of control slices kept in a morphine-free solution (normal group) was used as controls for the acute morphine added. We used the paired-pulse ratio (PPR) to assess chronic morphine-induced changes in glutamate synaptic transmission in the two types of NRM neurons, primary cells and secondary cells. In the μ receptor-containing secondary cells in control slices, the average PPR was 1.76 ± 0.05 (n = 16), indicating a common synaptic facilitation (PPR\>1) by the two stimuli. However, in secondary cells from morphine-tolerant rats, the PPR was significantly smaller than that in controls (1.43 ± 0.07, n = 41, P \< 0.01, Fig. [1A,C](#F1){ref-type="fig"}), indicating an increased probability of presynaptic glutamate release in these secondary cells in a morphine-tolerant state. The PPR in secondary cells in the normal group without 5 μM morphine was 1.89 ± 0.13 (n = 33), which was not statistically different from that in control cells kept in 5 μM morphine (p \> 0.05), excluding possible effect of the acute morphine on the PPR. In contrast to secondary cells, there was no significant difference between the EPSC PPRs in μ receptor-lacking primary cells from control and from morphine-tolerant rats (control, 1.43 ± 0.11, n = 10; tolerant, 1.44 ± 0.08, n = 17; P \> 0.05, Fig [1B,C](#F1){ref-type="fig"}). These data indicate that chronic morphine increases glutamate synaptic transmission selectively in the μ receptor-containing secondary cells in a morphine-tolerant state.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Chronic morphine decreases the paired-pulse ratio (PPR) of glutamate EPSCs in μ-opioid receptor-containing cells (termed secondary cells) in the NRM from morphine-tolerant rats. **A**, Representative EPSC pairs in secondary cells from a saline-treated rat, a morphine-treated tolerant rat and the same two EPSC pairs normalized to the first EPSC. **B**, Representative EPSC pairs in the other type of NRM cells (termed primary cells) lacking the μ receptor from two similar groups of rats. **C**, Group data of the PPR in the two cell types from saline control and morphine-tolerant rats. Stimulus artifacts are blanketed. Numbers in columns indicate cell numbers. \*\* p \< 0.01.
:::
:::
EPSC-enhancing effect of activators of cAMP pathway is increased
----------------------------------------------------------------
In control secondary cells, bath application of the adenylyl cyclase (AC) activator forskolin (10 μM) significantly increased the amplitude of evoked EPSCs (eEPSCs) by 53.9 ± 4.1% (control, 95.1 ± 7.3 pA, forskolin, 144.1 ± 14.2 pA, n = 12, P \< 0.01). Forskolin produced a comparable EPSC increase in secondary cells in normal slices without 5 μM morphine (57.4 ± 11.4%, n = 6, p \< 0.01, p \> 0.05 when compared to its effect in control slices with 5 μM morphine). However, in tolerant secondary cells, the EPSC-enhancing effect of forskolin was significantly increased to 95.3 ± 5.3% (p \< 0.01, compared to its effect in control group) (control, 88.9 ± 7.8 pA, forskolin, 174.3 ± 13.9 pA, n = 7, P \< 0.01, Fig. [2A,B](#F2){ref-type="fig"}). Similar to the effect of chronic morphine, forskolin (10 μM) also significantly decreased the PPR of glutamate EPSCs in secondary cells from both saline-treated and morphine-tolerant rats (saline: control, 1.73 ± 0.21, forskolin, 1.03 ± 0.14, n = 5, P \< 0.01; tolerant: control, 1.43 ± 0.14, forskolin, 0.90 ± 0.03, n = 5, P \< 0.01, Fig [2C](#F2){ref-type="fig"}). Due to a significant decrease in the basal PPR, direct comparison in percentage term of the forskolin effects between the two groups was not feasible. In primary cells, by contrast, although forskolin increased the eEPSC amplitude in both control and tolerant groups, the magnitude of its effects in the two groups was not changed (control, 72.8 ± 17.6%, n = 7, tolerant, 67.0 ± 14.9%, n = 5, P \> 0.05).
::: {#F2 .fig}
Figure 2
::: {.caption}
######
EPSC-enhancing effect of Forskolin is increased in secondary cells from morphine-tolerant rats. **A**, Glutamate EPSCs before (control) and during bath application of forskolin (10 μM) in cells from saline control and tolerant rats. **B**, Time course for the forskolin enhancement of EPSC amplitudes in saline control (n = 12) and tolerant cells (n = 7). The horizontal bar indicates application time for forskolin. **C**, Normalized EPSC pairs in the absence (control) and presence of forskolin in saline control and tolerant cells. Note forskolin-induced decrease in the PPR of both groups. Forsk, forskolin. \* p \< 0.05, \*\* p \< 0.01.
:::
:::
To further confirm the forskolin effect on presynaptic glutamate release, we examined forskolin action on miniature EPSCs (mEPSCs) in NRM neurons. In control secondary cells, forskolin (10 μM) increased the frequency, but not the amplitude, of glutamate mEPSCs (frequency: control, 5.70 ± 1.03 Hz, forskolin, 8.46 ± 1.07 Hz, n = 5, P \< 0.05; amplitude: control, 20.8 ± 2.0 pA, forskolin, 20.6 ± 2.4 pA, n = 5, P \> 0.05). This forskolin effect was also observed in secondary cells from tolerant rats (frequency: control, 7.64 ± 0.96 Hz, forskolin, 15.18 ± 0.98 Hz, n = 5, P \< 0.01; amplitude: control, 21.3 ± 2.4 pA, forskolin, 20.5 ± 2.23 pA, n = 5, P \> 0.05). By comparison, the forskolin-induced increase in the mEPSC frequency was significantly greater in tolerant secondary cells (control, 54.2 ± 9.8%, tolerant, 104.3 ± 11.5%, n = 5, P \< 0.05, Fig [3](#F3){ref-type="fig"} & Fig. [4D](#F4){ref-type="fig"}). The forskolin-induced increase in mEPSC frequency in tolerant primary cells was not altered (control, 62.6 ± 17.4%, n = 4, tolerant, 72.2 ± 36.4%, n = 5, P \> 0.05).
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Forskolin enhancement of miniature EPSC (mEPSC) frequency is greater in secondary cells from morphine-tolerant rats. **A**,**B**,**C**, Representative current traces with miniature synaptic events (**A**) and plots of cumulative distribution of mEPSC frequency (**B**) and amplitude (**C**) before and during application of forskolin (10 μM) in the same saline control cell. **D,E,F**, Current traces (**D**) and plots of mEPSC distribution data (**E,F**) from a tolerant cell.
:::
:::
::: {#F4 .fig}
Figure 4
::: {.caption}
######
EPSC-enhancing effect of 8-bromo-cAMP is increased in secondary cells from morphine-tolerant rats. **A**, Normalized EPSC pairs in control and in 8-broma-cAMP (8-br-cAMP, 1 mM) in a saline control cell and a tolerant cell. **B**, Group data of 8-br-cAMP-induced decrease in the PPR in saline control (n = 4) and tolerant cells (n = 5). **C**, Distribution data of mEPSC frequency before and during application of 8-br-cAMP in a saline control (left) and a tolerant cell (right). **D**, Percent increase by 8-br-cAMP and forskolin in mEPSC frequency in saline control and tolerant cells.
:::
:::
The cAMP analog 8-bromo-cAMP (8-br-cAMP) produced similar effects to those of forskolin. Bath application of 8-br-cAMP (1 mM) increased the eEPSC amplitude in secondary cells from both saline control and morphine-tolerant rats (saline: control, 110.9 ± 14.8 pA, 8-br-cAMP, 166.6 ± 9.2 pA, n = 4, P \< 0.01; tolerant: control, 130.1 ± 1.4 pA, 8-br-cAMP, 260.2 ± 28.5 pA, n = 7, P \< 0.001). This cAMP effect was significantly enhanced by chronic morphine treatment (saline, 52.1 ± 9.4%, tolerant, 103.6 ± 10.1%, p \< 0.05). 8-br-cAMP also decreased the PPR in secondary cells from both saline control and tolerant rats (saline: control, 1.70 ± 0.08, 8-br-cAMP, 1.10 ± 0.07, n = 4, P \< 0.01; tolerant: control, 1.49 ± 0.17, 8-br-cAMP, 1.14 ± 0.10, n = 5, P \< 0.05, Fig [4A,B](#F4){ref-type="fig"}).
Bath application of 8-br-cAMP (1 mM) also significantly increased the frequency of mEPSC in secondary cells from both saline control and tolerant rats (saline: control, 4.72 ± 0.47 Hz, 8-br-cAMP, 6.85 ± 0.70 Hz, n = 8, p \< 0.01; tolerant: control, 6.62 ± 0.88 Hz, 8-br-cAMP, 12.46 ± 1.25 Hz, n = 7, p \< 0.01, Fig. [4C](#F4){ref-type="fig"}). However, similar to the forskolin effect, this effect of 8-br-cAMP was also significantly enhanced in the slices from morphine-tolerant rats (saline, 46.9 ± 7.9%, n = 8, tolerant, 97.5 ± 21.2%, n = 7, p \< 0.05, Fig. [4D](#F4){ref-type="fig"}). 8-br-cAMP did not change the amplitudes of mEPSCs in either group (saline: control, 15.9 ± 0.6 pA, 8-br-cAMP, 15.9 ± 1.3 pA, n = 8, P \> 0.05; tolerant: control, 18.8 ± 1.2 pA, 8-br-cAMP, 18.9 ± 1.6 pA, n = 7, P \> 0.05).
The enhanced forskolin and 8-br-cAMP effects on eEPSC amplitude and mEPSC frequency in morphine-tolerant rats indicate a sensitized or upregulated AC system in a morphine-tolerant state.
Chronic morphine-induced EPSC increase involves cAMP pathway
------------------------------------------------------------
To determine the involvement of the AC pathway in the chronic morphine-induced increase of glutamate release in NRM secondary cells, NRM slices were treated with MDL12330a, a selective AC inhibitor. While pre-incubation with MDL12330a (100 μM) did not change the EPSC PPR in secondary cells from control rats (control, 1.71 ± 0.06, MDL, 1.76 ± 0.03, n = 5, p \> 0.05), it largely reversed the chronic morphine-induced PPR decrease in secondary cells from morphine-tolerant rats (tolerant, 1.43 ± 0.07, +MDL, 1.74 ± 0.07, n = 16, P \< 0.01, Fig. [5A,B,C](#F5){ref-type="fig"}). Bath application of MDL12330a (100 μM) also significantly decreased the frequency of mEPSCs in tolerant secondary cells (tolerant, 10.17 ± 0.89 Hz, +MDL, 5.86 ± 0.66 Hz, n = 5, P \< 0.01, Fig. [5D,E,F](#F5){ref-type="fig"}) whereas it did not alter the mEPSC amplitude (tolerant, 16.9 ± 1.1 pA, +MDL, 18.1 ± 1.2 pA, n = 5, P \> 0.05). MDL had no effect on either the frequency or the amplitude of mEPSCs in secondary cells from saline control rats (frequency: control, 4.39 ± 0.45 Hz, +MDL, 4.71 ± 0.50 Hz, n = 5, P \> 0.05; amplitude: control, 16.9 ± 2.0 pA, +MDL, 17.5 ± 1.5 pA, n = 5, P \> 0.05, Fig. [5F](#F5){ref-type="fig"}). Next, we used a PKA inhibitor to determine whether PKA was involved in the enhanced glutamate transmission in secondary cells induced by chronic morphine. Pre-incubation with H89 (10 μM) largely inhibited the chronic morphine-induced decrease of the EPSC PPR in secondary cells from morphine-tolerant rats (tolerant, 1.43 ± 0.07, +H89, 1.71 ± 0.06, n = 13, P \< 0.01, Fig. [6A,B](#F6){ref-type="fig"}). H89 was without effect on the EPSC PPR in secondary cells from control rats (tolerant, 1.76 ± 0.06, +H89, 1.90 ± 0.08, n = 6, P \> 0.05, Fig. [6B](#F6){ref-type="fig"}). Similarly, bath application of H89 (10 μM) significantly decreased the frequencies, but not the amplitude, of mEPSCs in tolerant secondary cells (frequency: tolerant, 8.55 ± 2.42 Hz, +H89, 4.66 ± 2.11 Hz, n = 5, P \< 0.05; amplitude: tolerant, 15.1 ± 0.9 pA, +H89, 14.3 ± 0.7 pA, n = 5, P \> 0.05, Fig [6C,D](#F6){ref-type="fig"}). In control secondary cells, H89 did not change either the frequency or the amplitude of mEPSCs (frequency: control, 3.96 ± 0.44 Hz, +H89, 3.43 ± 0.62 Hz, n = 5, P \> 0.05; amplitude: control, 20.5 ± 3.8 pA, +H89, 20.2 ± 3.0 pA, n = 5, P \> 0.05, Fig. [6D](#F6){ref-type="fig"}).
::: {#F5 .fig}
Figure 5
::: {.caption}
######
MDL12330a reverses the chronic morphine effect on glutamate EPSCs in tolerant secondary cells. **A**, Representative EPSC pairs in tolerant cells without (tolerant) and after treatment with the adenylyl cyclase (AC) inhibitor MDL12330a (MDL, 100 μM). **B**, Normalized EPSC pairs from **A**. **C**, Group data of the MDL effect on the PPR in cells from morphine-tolerant (n = 16) and saline control rats (n = 5). **D**,**E**, Representative current traces (**D**) and a plot of mEPSC frequency distribution (**E**) in a tolerant cell before and during application of MDL (100 μM). **F**, Group data of the MDL effect on mEPSC frequency in tolerant and saline control cells (n = 5 in each group).
:::
:::
::: {#F6 .fig}
Figure 6
::: {.caption}
######
H89 reverses the chronic morphine effect on glutamate EPSCs in tolerant secondary cells. **A**, Normalized EPSC pairs in tolerant cells without and after treatment with H89 (10 μM), a protein kinase A inhibitor. **B**, Group data of the H89 effect on PPR in tolerant (n = 13) and saline control cells (n = 6). **C**, A plot of mEPSC frequency distribution in tolerant cells with or without H89 treatment. **D**, Group data of the H89 effect on mEPSC frequency in tolerant and saline control cells (n = 5 in each group).
:::
:::
These data obtained with both MDL12330a and H89 suggest that the cAMP/PKA pathway is critically involved in the chronic morphine-induced enhancement of glutamate synaptic transmission in NRM secondary cells.
Chronic morphine-induced EPSC increase involves PKC pathway
-----------------------------------------------------------
Finally, we examined whether the PKC pathway was also involved in the chronic morphine effect on glutamate EPSCs. Phorbol 12-myristate 13-acetate (PMA, 1 μM), a phorbol ester activator of PKC, increased the amplitude of eEPSCs in NRM secondary cells from both saline control and morphine-tolerant rats (saline: control, 109.9 ± 15.8 pA, PMA, 180.4 ± 22.2 pA, n = 7, P \< 0.001; tolerant: control, 134.8 ± 12.9 pA, PMA, 284.7 ± 25.8 pA, n = 7, P \< 0.01, Fig. [7A](#F7){ref-type="fig"}). However, when the PMA effects were compared between the two groups, the EPSC-enhancing effect of PMA was significantly larger in cells from tolerant rats than that in control cells (control, 67.7 ± 6.9%, n = 7, tolerant, 126.6 ± 16.4%, n = 7, p \< 0.01, Fig. [7B](#F7){ref-type="fig"}), indicating a likely upregulated PKC pathway. GF109203X (2 μM), a selective PKC inhibitor, had no significant effect on the EPSC PPR in control secondary cells (control, 1.76 ± 0.14, GF109203X, 1.76 ± 0.05, n = 6, P \> 0.05), but it reversed the chronic morphine-induced decrease of the EPSC PPR in secondary cells from morphine-tolerant rats (tolerant, 1.43 ± 0.07, +GF109203X, 1.79 ± 0.10, n = 8, P \< 0.05, Fig. [7C,D,E](#F7){ref-type="fig"}).
::: {#F7 .fig}
Figure 7
::: {.caption}
######
Chronic morphine-induced EPSC augmentation involves the protein kinase C pathway. **A**, Representative EPSCs in the absence and presence of the protein kinase C (PKC) activator phorbol 12-myristate 13-acetate (PMA, 1 μM) in a saline control cell and a tolerant cell. **B**, PMA effect on the peak amplitude of eEPSCs in saline control and tolerance cells. **C**,**D**, Representative EPSC pairs (**C**) and the same pairs after normalization (**D**) in tolerant cells with or without treatment with the PKC inhibitor GF109203X. **E**, Group data of the GF109203X effect on PPR in tolerant (n = 8) and saline control cells (n = 6).
:::
:::
Discussion
==========
The present study has illustrated that chronic morphine increases presynaptic release of glutamate in the μ receptor-containing secondary cells in a morphine-tolerant state. It also provides evidence that both cAMP/PKA and PKC signaling pathways are critically involved in this chronic morphine-induced synaptic adaptation during morphine tolerance.
Upregulation of the cAMP pathway and glutamate release
------------------------------------------------------
In the present study, the effect of chronic morphine on glutamate synaptic transmission was primarily assessed by using the paradigm of EPSC PPR in NRM neurons kept in a tolerant state from morphine-tolerant rats \[[@B14]\]. The PPR has been widely used to determine the involvement of a presynaptic site in the mechanism of neurotransmitter release \[[@B15]-[@B21]\]. A change in the PPR is inversely related to the probability of transmitter release and thus, a manipulation that increases the probability of transmitter release results in a reduction in the PPR and vise versa. Since the PPR in a cell under a certain condition is relatively stable, the advantage of using the PPR is to avoid large variance usually present in eEPSC amplitudes and mEPSC frequencies among individual cells, making it possible to compare EPSCs in two separate groups of slices, such as those from saline-treated control rats and those from morphine-tolerant rats. Our current results obtained with the PPR analysis suggest that glutamate synaptic transmission is enhanced through a presynaptic mechanism in NRM secondary cells from morphine-tolerant rats. This conclusion is further supported by the effects of activators and inhibitors of the cAMP/PKA or PKC pathway, which mimicked and reversed the effect of chronic morphine, respectively. It is interesting to note that the chronic morphine-induced EPSC enhancement occurred exclusively in μ receptor-containing secondary cells. This may implicate a functional activation of these μ receptor-expressing NRM cells by glutamate inputs during morphine tolerance (see below). The source of these glutamate inputs is currently unclear.
Presynaptic glutamate release is modulated by many complex and interacting proteins and processes, including components of the cAMP/PKA pathway, in presynaptic terminals \[[@B22],[@B23]\]. It has been demonstrated that *in vitro*application of activators of the cAMP/PKA pathway enhances presynaptic release of glutamate in many types of central neurons \[[@B23]-[@B27]\]. *In vivo*administration of chronic morphine induces adaptive hyperactivation of the cAMP/PKA signaling pathway in several brain regions \[[@B1],[@B27],[@B28]\]. The hyperactivated cAMP/PKA pathway induced by chronic morphine may induce a series of cellular adaptations, including enhanced transmitter release. However, cAMP-dependent synaptic adaptation induced by *in vivo*administration of chronic morphine has been reported only in GABAergic synapses in central neurons during morphine withdrawal \[[@B21],[@B28],[@B29]\]. The current study provides direct evidence that chronic morphine enhances glutamate synaptic transmission also in a cAMP-dependent way in NRM neurons from morphine-tolerant rats.
Several observations in the current study support a critical role of an upregulated cAMP/PKA pathway in the enhanced glutamate neurotransmission in secondary cells from morphine-tolerant rats. First, the AC activator and cAMP analog mimic the effect of chronic morphine on EPSC PPR. Second, the potency of cAMP activators in enhancing eEPSC amplitude and mEPSC frequency is significantly augmented in neurons from morphine-tolerant rats. Such an augmented effect has been interpreted as the result of a sensitized or upregulated cAMP pathway \[[@B21],[@B27],[@B29]\]. Finally, inhibitors of both AC and PKA reverse the chronic morphine-induced EPSC enhancement. This effect of AC/PKA inhibitors observed in tolerant cells, but not in control cells, indicates an elevated basal activity of the cAMP/PKA pathway induced by chronic morphine. Such an upregulation of basal and stimulated PKA activity has been shown in other brain areas \[[@B1],[@B30]\].
Upregulation of the PKC pathway and glutamate release
-----------------------------------------------------
It has been shown that activation of PKC either by the intracellular messenger diacylglycerol or by phorbol esters produces an enhancement of neurotransmitter release including glutamate release in central neurons \[[@B23],[@B31],[@B32]\]. Activated PKC may phosporylate a number of proteins in nerve terminal and increase neurotransmitter release by calcium influx through voltage-gated calcium channels or by direct effects on exocytosis or on vesicle-recycling pathways \[[@B33]\]. Chronic morphine increases PKC activity in the rat brain and spinal cord \[[@B34],[@B35]\]. The increased PKC activity, which may be mediated through opioid activation of the phospholipase C pathway coupled to opioid receptors, can phosphorylate AC and increase its adaptive responses to chronic opioids \[[@B27],[@B30]\]. In the present study, the PKC activator, similar to activators of the cAMP/PKA pathway, exhibited an augmented potency in enhancing glutamate EPSCs in NRM neurons from morphine-tolerant rats, and the PKC inhibitor reversed the effect of chronic morphine. Although non-specific effects of these PKA and PKC inhibitors cannot be completely ruled out, these data indicate a likely upregulated PKC and PKA activity, which is at least partially responsible for the enhanced glutamate synaptic transmission in NRM secondary cells from morphine-tolerant rats.
Functional implications in morphine tolerance
---------------------------------------------
Extensive evidence has shown that some NRM cells inhibited by μ opioids have a facilitating action on spinal pain transmission through their descending projections \[[@B7],[@B10]\]. Recent pain research shows that those brainstem cells expressing μ-opioid receptors are commonly activated in various chronic pain conditions, including chronic opioid-induced abnormal pain, and contribute to the sensitized pain or hyperalgesia observed during these conditions \[[@B11]-[@B13],[@B36]\]. It remains unclear what mediates the activation of these μ receptor-containing cells in those pain conditions. Data from the current study indicate that excessive activity of glutamate synaptic inputs after chronic exposure to opioids may contribute to the activation of these cells during morphine tolerance. An overall effect of this enhanced glutamate synaptic activity in driving these cells needs to be evaluated by taking into account other synaptic inputs such as GABA synaptic transmission.
It has been proposed that the chronic opioid-induced activation of those presumably pain-facilitating cells and consequently, abnormal pain, constitute part of the mechanisms underlying opioid tolerance characterized by a reduced analgesic effect of opioids \[[@B36]\]. For example, microinjection of lidocaine into the NRM area to inactivate cell activity reverses the tactile allodynia and thermal hyperalgesia induced by prolonged exposure to morphine, and attenuates morphine tolerance \[[@B37],[@B38]\]. A critical role of glutamate receptors in the development of opioid tolerance and dependence has been established \[[@B3]-[@B6]\]. The present study illustrates an example of the mechanisms by which glutamate receptors and synapses participate in the development of opioid tolerance. An important role of PKC in opioid tolerance has also been demonstrated in previous studies. For instance, PKC inhibition reduces opioid tolerance \[[@B34],[@B35],[@B39],[@B40]\], and PKCγ mutant mice display alleviated morphine tolerance \[[@B41]\].
In summary, the current study shows that chronic morphine enhances glutamate synaptic transmission in μ receptor-containing NRM neurons. This synaptic adaptation appears to be mediated by an upregulated cAMP/PKA and PKC pathway in morphine-tolerant rats. These results may provide key information for pain therapies aiming at inhibiting those brainstem neurons and their descending pain facilitation, which is responsible for the pain sensitization in several chronic pain conditions.
Methods
=======
All procedures involving the use of animals conformed to the guidelines set by the University of Texas-MD Anderson Cancer Center Animal Care and Use Committee. *Chronic morphine treatment*Male, Wistar neonatal rats were injected (i.p.) twice daily with morphine solutions for 6 days. The dose of the morphine was 10 mg/kg on the first day, and increased by 5 mg/kg each day to reach a maximum dose of 30 mg/kg on day 5. On day 7, the rats were euthanized for brain slice preparation. Saline was injected at the same volume and schedule in a separate group of rats for controls. Morphine tolerance in these neonatal rats has been described previously \[[@B42]\].
*Whole-cell voltage-clamp recording*The methods for NRM brain slice preparations, visualized whole-cell recordings, cell classification and analysis of glutamate EPSCs have been published previously \[[@B15]\]. The brain of a rat (10--14 days old) was cut in a vibratome in cold (4°C) physiological saline to obtain brainstem slices (220--250 μm thick) containing the NRM. A single slice was submerged in a shallow recording chamber and perfused with preheated (35°C) physiological saline (in mM: NaCl, 126; KCl, 2.5; NaH~2~PO~4~, 1.2; MgCl~2~, 1.2; CaCl~2~, 2.4; glucose, 11; NaHCO~3~, 25, saturated with 95% O~2~and 5% CO~2~, pH 7.2--7.4). Slices were maintained at around 35°C throughout the recording experiment. Neonatal rats were used for better visualization of neurons in brain slices with an infrared Nomarski microscope. It has been demonstrated that the physiological and pharmacological properties of neurons from these young rats are indistinguishable from those of adult rats\[[@B9],[@B15]\]. Visualized whole-cell voltage-clamp recordings were made from identified neurons held at -60 mV with a glass pipette (resistance 3--5 MΩ) filled with a solution containing (mM): potassium gluconate, 126; NaCl, 10; MgCl~2~, 1; EGTA, 11; Hepes, 10; ATP, 2; GTP, 0.25; pH adjusted to 7.3 with KOH; osmolarity 280--290 mosmol/L. An AxoPatch 1-D amplifier and AxoGraph software 4.7 (Axon Instruments, Inc.) were used for data acquisition and on-line/off-line data analyses. A seal resistance of 2 GΩ or above and an access resistance of 15 MΩ or less were considered acceptable. Series resistance was optimally compensated. The access resistance was monitored throughout the experiment. Electrical stimuli of constant current (0.25 ms, 0.2--0.4 mA) were used to evoke EPSCs with bipolar stimulating electrodes placed lateral (200--400 μm) to the recording pipette within the NRM. The difference in glutamate synaptic transmission between control and morphine-tolerant slices was assessed by the paradigm of paired-pulse ratio (PPR). A pair of EPSCs was evoked by two stimuli with an interval of 40 ms. The PPR was determined by dividing the second EPSC amplitude by the first one. Miniature EPSCs were obtained in 60-sec epochs in the presence of tetrodotoxin (1 μM). The AxoGraph software was used to detect and measure the amplitude and intervals of the synaptic events, and to analyze their distribution data. All NRM cells recorded were classified into either a μ-containing secondary cell or a μ-lacking primary cell according to the criteria described in our previous study \[[@B8]\]. NRM slices from both morphine- and saline-treated rats were kept in 5 μM morphine throughout the recording experiment to maintain the slices in a morphine-tolerant state and prevent withdrawal \[[@B14]\]. Another group of slices from saline-treated rats kept in a morphine-free solution was used as controls for the acute morphine. Statistic analysis of mEPSCs were performed with the Mann-Whitney *U*test or the Kolmogorov-Smirnov test using the Statview software. Other numeral data were statistically analyzed with Students\' t tests and presented as mean ± S.E.M. Some slices were incubated in MDL12330a (100 μM), or H89 (10 μM) or GF109203X (2 μM) for at least 1 hour. Drugs were generally applied through the bath solution unless otherwise specified. Morphine sulfate was kindly supplied by the National Institute on Drug Abuse. All other drugs were purchased either from Sigma-Aldrich Co. or from Tocris Cookson Inc. (Ellisville, MO).
Competing interests
===================
The author(s) declare that they have no competing interests.
Acknowledgments
===============
This work was supported by the National Institute on Drug Abuse grant DA14524 and by an institutional fund of MD Anderson Cancer Center.
|
PubMed Central
|
2024-06-05T03:55:55.101709
|
2005-2-9
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074357/",
"journal": "Mol Pain. 2005 Feb 9; 1:7",
"authors": [
{
"first": "Bihua",
"last": "Bie"
},
{
"first": "Zhizhong Z",
"last": "Pan"
}
]
}
|
PMC1074358
|
Mechanosensation plays a pivotal role in many aspects of pain pathology, yet the mammalian molecular transduction apparatus responsible for this sensory modality remains unknown. In January\'s edition of *Nature Neuroscience*, O\'Hagan, Chalfie and Goodman \[[@B1]\] have provided direct electrophysiological evidence that somatic mechanotransduction in *C. elegans*is mediated by a complex of proteins previously identified in genetic screens for impaired touch sensation. Are the homologues of these proteins important for pain sensation in mammals? Perhaps surprisingly, the balance of evidence suggests that other proteins are better candidate noxious mechanosensors in mammals.
Many forms of pain, be it in acute, inflammatory or disease-related conditions, are triggered by mechanical stimuli. However, in mammals there is very little understanding of the molecular transduction process that converts mechanical stimuli into a change in membrane excitability. Studying mechanosensation in mammals is hampered by the diffuse and inaccessible distribution of nerve terminals in the periphery. The few studies of receptor potentials, made using extracellular recordings (mainly from Pacinian corpuscles of the cat\'s mesentery), do however suggest that mechanical stimuli depolarise termini by directly gating cationic channels \[[@B2]\].
It is genetic studies in *C. elegans*and *Drosophila*that have driven forward our molecular understanding of mechanosensation in a number of different cell types. The best-characterised system is the body touch receptor neuron of *C. elegans*; over 2 decades, Martin Chalfie and co-workers have, on the basis of genetic mutant interactions, behavioural analysis and gene cloning, devised an elegant molecular model of transduction in these cells (see Refs. 3 and 4). In this model at least 9 proteins form a mechanotransduction complex with an ion channel at its core formed by MEC-4 and MEC-10 (members of the DEG/ENaC ion channel superfamily) and apparently MEC-6 (a paraoxonase-like protein, \[[@B5]\]). The complex also contains extra- and intracellular structures that the ion channel is tethered to, via specific linker proteins (probably stomatin-like MEC-2 internally, \[[@B6]\]), such that sheering between them gates the channel (Fig. [1](#F1){ref-type="fig"}). Up until the present study however, no one had recorded ionic currents attributable to activation of this complex. Now though, Chalfie, Rob O\'Hagan and Miriam Goodman (a pioneer of *in situ*patch-clamping in nematodes) have measured mechanoreceptor currents (MRCs) in body touch receptors and provided direct evidence supporting the model of transduction \[[@B1]\].
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Schematic diagram of the proposed mechanotransduction complex in *C. elegans*body touch receptors. At its centre is an ion channel composed of MEC-4, 6 and 10, which interacts with the intracellular protein MEC-2. MEC-7 and 12 are microtubule proteins required for normal mechanosensation (they may be important for localisation or gating of the complex). MEC-1, 5 and 9 are extracellular proteins whose functions await further characterisation. (Figure adapted from Ref. 4.)
:::
:::
To record from body touch receptors, O\'Hagan *et al*used transgenic animals in which these cells were labelled with GFP. Using immobilised worms, the authors released the internal hydrostatic pressure away from the recording site and then exposed the cell bodies of posterior, lateral receptor neurons. Then patch-clamp recordings were made from the cell body while the mechanosensitive neurite was stimulated with a glass probe applied to the body wall. The authors observed that both the application and withdrawal of mechanical stimuli evoked rapidly adapting inward currents, whose amplitude was proportional to the magnitude of the stimulus. Consistent with the currents being mediated by members of the DEG/ENaC family, they were carried by sodium ions and blocked by amiloride. Next, given the extensive genetic analysis of mechanosensation in this species the investigators were able to extend their work by studying receptor currents in a range of mutant animals. Firstly, it was shown that null mutations in MEC-4, MEC-2 and MEC-6 abolished MRCs, suggesting that these 3 proteins (which physically interact) are essential for channel gating. An important control experiment was to show that voltage-gated currents in these mutants were normal. Subsequently, it was found that other (behaviourally less severe) mutations in MEC-4 and MEC-10 greatly reduced MRC amplitude and significantly altered the current-voltage relationship of MRCs. Hence, this is the first direct demonstration that MEC-4 and MEC-10 form the mechanotransducing ion channel in *C. elegans*. Finally, the group analysed MRCs in nematodes with a mutation in MEC-7, a β-tubulin required for formation of touch cell specific 15-protofilament microtubules, which had been hypothesised to be intracellular \"anchors\" required for channel gating. Interestingly, despite a large decrease in their amplitude and threshold, MRCs were not abolished in these mutants suggesting that MEC-7 is not an absolute requirement for channel gating.
This study represents a confirmation of the key aspects of a long-standing model of mechanotransduction. However, the relationship between this system and those in operation in mammalian somatic mechanosensation remains unclear. In mammals there are 9 identified DEG/ENac channels, which form two subfamilies; the epithelial sodium channels (ENaCα, β, γ and δ) and the acid sensing ion channels (ASIC1-4, and the closely related intestinal sodium channel, INaC). ENaCα, β and γ together form a constitutively active channel principally associated with non-neuronal tissues. β and γ ENaC do appear to be expressed in DRG neurons \[[@B7]\] but, as yet, their function there has not been studied. However, much interest was aroused in ASICs as potential mechanosensors because they are highly expressed in sensory neurons and 2 isoforms (ASIC3 and 1b) are almost exclusively expressed in these cells. Currently, the only known activator of these channels is external acidification, which gates 4 of the 6 known splice variants when they are expressed alone (interestingly MEC-4 and MEC-10 are not gated by protons) \[[@B8]\]. However, it has been suggested that if localised in a mechanotransduction complex analogous to that found in *C. elegans*, ASICs might mediate mammalian mechanosensation \[[@B9]\]. To test this hypothesis Michael Welsh and Gary Lewin collaborated in generating null mutants of ASIC1, 2 and 3 and assessing their somatosensory phenotypes using the skin-nerve preparation. In stark contrast to the dramatic effects of null mutations in MEC-4 and MEC-10, ablation of these genes had minor effects on mechanosensory responses. The first study found an approximate halving of the suprathreshold firing rates of rapidly adapting low threshold mechanoreceptors (LTMs) in ASIC2 nulls and a minor decrease in slowly adapting LTMs whilst the responses of all other fibre types were unchanged from wild type values \[[@B10]\]. In ASIC3 knockouts, rapidly adapting LTMs had an *increased*sensitivity to mechanical stimuli whereas Aδ-mechanonociceptors showed a decrease in responsiveness \[[@B11]\] and in ASIC1 null mutants cutaneous mechanosensation was unchanged from wild-type levels \[[@B12]\]. Whilst the analysis of double and triple knockouts would be worthwhile given the possibility that the remaining subunits functionally compensate for the missing ones in null mutants (although their expression was unchanged at the transcriptional level), the phenotypes of these animals is not consistent with ASICs being major transducers of mechanical stimuli in mammalian sensory nerves. Moreover, in an analysis of a separate line of ASIC2 nulls, no alteration in the sensitivity of rapidly adapting LTMs was found \[[@B13]\] and no group has reported mechanical gating of ASICs. Although mechanical gating of ion channels that are mechanosensitive *in situ*may be difficult using *in vitro*systems, different subpopulations of cultured DRG neurons are known to display distinct mechanically activated cationic currents \[[@B14]\] and these currents are unchanged in ASIC2 and/or 3 null mutants \[[@B15]\]. These data therefore suggest that other ion channels act as the primary mechanotransducers in mammals.
Whilst research on body touch receptors in *C. elegans*focussed attention on DEG/ENaC channels, genetic screens of other mechanosensory systems, particularly in *Drosophila*, have also revealed major roles for TRP channels in mechanosensation. In fruit flies, TRP-like channels NOMPC \[[@B16]\] and Nanchung \[[@B17]\] have been strongly implicated as mechanotransduction channels in Type I mechanosensors required for touch and hearing, respectively. In *Drosophila*larvae, Painless, a TRPV-like protein, is expressed in nociceptor-like cells and mutants have defective responses to noxious thermal and mechanical stimuli \[[@B18]\]. Also, in *C. elegans*OSM-9 is required for nose touch avoidance \[[@B19]\]. Research in mammalian systems has now produced evidence suggesting TRPA1 may be the transduction channel in hair cells \[[@B20]\] whilst TRPC1 has recently been shown to be directly mechanosensitive \[[@B21]\]. With regard to noxious mechanosensation, TRPV4 knockouts were found to have behavioural deficits in response to tail pressure \[[@B22]\], although this channel seems to be expressed at much higher levels in keratinocytes than in sensory neurons. Given that a number of TRP channels are already known to be central to thermosensation and inflammatory function in nociceptors, members of this family represent interesting candidates for mammalian noxious (and innocuous) mechanosensors.
In conclusion, the primary candidates for the role of mammalian mechanotransducers are members of the TRP and DEG/ENaC ion channel families, both of which are remarkably functionally diverse. However, the evidence supporting a function for any particular channel in mammalian mechanotransduction is much weaker than in invertebrate systems. Interestingly, the diversity of DEG/ENaC channels in *C. elegans*(28 homologues) in comparison to mammals (mice have 8) is striking, and the observation that mechanosensitive channels in nematodes form a distinct subgroup that all contain a specific extracellular regulatory domain \[[@B23]\] makes extrapolation of the *C. elegans*results to mammals less certain. Related to the diversity of putative mechanosensory ion channels is the issue of diversity in cellular systems that mediate mechanosensation. Despite similarities, the phylogenetic relationship between mammalian hair cells and primary somatosensory neurons and the analogous cells types in invertebrates is poorly established. Also, the extent to which chemically mediated mechanosensation functions in certain systems, potentially including some forms of mechanically induced pain, is currently unclear (for example see Ref. 24). Thus, much remains to be learnt regarding the molecular basis of mechanotransduction and when this is achieved, it should be possible to determine the evolutionary relationships of multiple mechanosensory systems. In addition, identification of the molecular basis of noxious mechanosensation should provide exciting new analgesic drug targets.
|
PubMed Central
|
2024-06-05T03:55:55.104220
|
2005-2-15
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074358/",
"journal": "Mol Pain. 2005 Feb 15; 1:8",
"authors": [
{
"first": "Liam J",
"last": "Drew"
},
{
"first": "John N",
"last": "Wood"
}
]
}
|
PMC1074751
|
Introduction {#s1}
============
Ongoing spontaneous activity is present throughout the nervous system \[[@pbio-0030141-b01]\], but its function remains enigmatic. In the embryo, spontaneous movements \[[@pbio-0030141-b02]\] and waves of endogenous retinal activity \[[@pbio-0030141-b03],[@pbio-0030141-b04]\] are thought to play an important role in the epigenesis of neural networks through selective synapse stabilization \[[@pbio-0030141-b05],[@pbio-0030141-b06]\]. Ongoing spontaneous activity is also present in the adult brain, where it is responsible for the highly variable patterns of the electroencephalogram (EEG). Thalamocortical networks generate a variety of oscillations whose rhythms change across the sleep-wake cycle \[[@pbio-0030141-b07],[@pbio-0030141-b08],[@pbio-0030141-b09]\]. Optical imaging methods in anesthetized animals also reveal fast spontaneous states of neuronal activity that, far from being random, exhibit patterns that resemble those evoked by external stimuli \[[@pbio-0030141-b10],[@pbio-0030141-b11]\]. In parallel, functional neuroimaging studies in humans have shown a globally elevated brain metabolism at rest, with localized patterns suggesting that particular cortical regions are maintained in a high, although variable, state of activity \[[@pbio-0030141-b12],[@pbio-0030141-b13],[@pbio-0030141-b14],[@pbio-0030141-b15],[@pbio-0030141-b16]\]. At present, the functional roles of this spontaneous activity in the adult brain at rest remains to be elucidated.
In previous neuronal modeling studies and computer simulations, we illustrated the possible contribution of spontaneous activity to tasks that involve a random search, such as the learning of a temporal sequence \[[@pbio-0030141-b17]\], the search for and selection of the correct rule in the delayed response and Wisconsin card-sorting tests \[[@pbio-0030141-b18]\], or the discovery of a multistep solution in the Tower of London test \[[@pbio-0030141-b19]\]. More recently, generalizing from this early work, we proposed a broader framework of a formal architecture of thalamocortical areas, in which top-down activity generated in hierarchically higher cortical areas plays a key role in what we referred to as "access to consciousness" in an effortful task \[[@pbio-0030141-b20],[@pbio-0030141-b21],[@pbio-0030141-b22],[@pbio-0030141-b23]\]. Like several previous proposals, our model of a conscious neuronal workspace distinguishes lower automatized systems from increasingly higher and more autonomous supervisory systems \[[@pbio-0030141-b24]\]. It also builds upon Baars\' cognitive theory of consciousness, which distinguishes a vast array of unconscious specialized processors running in parallel, and a single limited-capacity serial "workspace" that allows them to exchange information \[[@pbio-0030141-b25]\].
The proposed neuronal architecture separates, in a first minimal description, two computational spaces, each characterized by a distinct pattern of connectivity. Subcortical networks and most of the cortex can be viewed as a collection of specialized and automatized processors, each attuned to the processing of a particular type of information via a limited number of local or medium-range connections that bring to each processor the "encapsulated" inputs necessary to its function. On top of this automatic level, we postulate a distinct set of cortical "workspace" neurons characterized by their ability to send and receive projections to many distant areas through long-range excitatory axons, thus allowing many different processors to exchange information.
Our previous simulations demonstrated how this architecture could account for a psychological phenomenon, the "attentional blink." Because of its long-distance, brain-scale connectivity, the global workspace establishes a central processing bottleneck such that, in the presence of two competing stimuli, processing of the first temporarily blocks high-level processing of the second \[[@pbio-0030141-b22]\]. While this work simulated only sensory processing, a key hypothesis of the workspace model is that the neurons of the higher level, the workspace neurons, are the seat of a permanent spontaneous activity that creates a succession of active internal states \[[@pbio-0030141-b20],[@pbio-0030141-b21],[@pbio-0030141-b23]\]. The aim of the present paper is to explore in a more extensive and systematic manner the role of this ongoing spontaneous activity in a similar neural network comprising several nested levels of neuronal architecture. We propose a specific network architecture and perform explicit computer simulations that offer plausible explanations for the origins and function of structured spontaneous activity in adult thalamocortical circuits, and in particular its critical role in allowing or blocking access by sensory stimuli.
The observed dynamic properties of the network lead us to distinguish two main transitions in activation. First, a neuromodulatory substance is assumed to control the level of network activation; as its input increases continuously, the network exhibits a sudden surge in spontaneous activation and switches to a state of thalamocortical resonance characterized by temporary bouts of synchronized gamma-band oscillations of increasing amplitude. This state of activity leads to a facilitation of sensory processing, and is proposed to correspond to the state of vigilance or being awake.
When the simulated areas are reciprocally connected by long-distance excitatory connections, a second state transition can occur. A subset of areas may suddenly show a strong temporary increase in synchronized firing and form a coherent state of activity ("ignition"). The transition to this state of high correlated activity is fast and characterized by an amplification of local neural activation and the subsequent ignition of multiple distant areas. This state of activity competes with, rather than facilitates, sensory processing, and thus leads to an extinction of sensory processing. We propose that this blocking may account for the "inattentional blindness" phenomenon, in which normal subjects intensely engaged in mental activity fail to notice salient but task-irrelevant sensory stimuli \[[@pbio-0030141-b26],[@pbio-0030141-b27],[@pbio-0030141-b28]\].
Results {#s2}
=======
We used computer simulations to characterize spontaneous and evoked activity in a complex nested architecture comprising multiple neurons, columns, and areas ([Figure 1](#pbio-0030141-g001){ref-type="fig"}). To facilitate comprehension, we organize the results section as a progression from local to more global states of activity. We start by describing the spontaneous and evoked activity in the building blocks of the model, namely the single neuron and an isolated thalamocortical column. We then consider the extent to which those properties are affected when multiple thalamocortical columns are interconnected by long-distance, bottom-up and top-down connections.
::: {#pbio-0030141-g001 .fig}
Figure 1
::: {.caption}
###### Simulation Components and Resulting Spontaneous Activity
Shown are the constituents of the simulation (upper diagrams) and typical patterns of spontaneous activity that they can produce (lower tracings). We simulated a nested architecture in which spiking neurons (A) are incorporated within thalamocortical columns (B), which are themselves interconnected hierarchically by local and long-distance cortical connections (C) (see [Materials and Methods](#s4){ref-type="sec"} for details). While single neurons may generate sustained oscillations of membrane potentials (A), only the column and network levels generate complex waxing-and-waning EEG-like oscillations (B) and metastable global states of sustained firing (C).
:::
:::
Spontaneous Oscillatory Behavior in a Single Neuron and a Thalamocortical Column {#s2a}
--------------------------------------------------------------------------------
We first simulated a single neuron using the "cellular oscillator" model derived from Wang \[[@pbio-0030141-b29]\]. The results appear in [Figure 2](#pbio-0030141-g002){ref-type="fig"}A. In the absence of any depolarizing current, the resting membrane potential is stable at *V* ≈ −63 mV. Injection of an increasing depolarizing current *I* ~neuromodul~ leads to the sudden emergence, at a rather precise value (*I* ~neuromodul~≈ −1.1 μA/cm^2^), of oscillations in membrane potential in the gamma range. Two features characterize this transition as a supercritical Hopf bifurcation according to dynamical systems theory \[[@pbio-0030141-b30]\]. First, a discontinuous transition is observed in the frequency domain, with the oscillation emerging suddenly at a characteristic frequency of 30--35 Hz, and changing only slowly with increasing current (up to 40--45 Hz). Second, a continuous transition is observed in the amplitude domain, with oscillation amplitude increasing continuously from zero as the square root of the amount of deviation from the threshold current (and therefore power increases linearly, as shown in [Figure 2](#pbio-0030141-g002){ref-type="fig"}A).
::: {#pbio-0030141-g002 .fig}
Figure 2
::: {.caption}
###### Characterizing the Transition Toward Spontaneous Activity
In each panel, we show the evolution of the power of the local field potential (top left of each), frequency (top center), and mean firing rate of pyramidal neurons (top right) as a function of the injected neuromodulatory current *I* ~neuromodul~. The elongated graphs give examples of the temporal evolution of the local field potential (bottom trace) and spikes (top trace) as a function of time in milliseconds for three different levels of *I* ~neuromodul~ *.* The simulations were performed with a single neuron (A and C) or with a whole thalamocortical column (B, D, and E), and with different sources of spontaneous activity: intrinsic cellular oscillations (A and B), random spikes (C and D), or both random spikes and intrinsic thalamic oscillators (E). Regardless of the exact mechanism, the architecture of thalamocortical columns created a bifurcation characterized by the sudden emergence, at a relatively constant threshold current, of high-frequency oscillations in the gamma range (\> 20 Hz), with continuously increasing power as a function of *I* ~neuromodul~.
:::
:::
Around *I* ~neuromodul~ ≈ −1.7 μA/cm^2^, a second threshold is observed: When oscillation amplitude reaches the voltage threshold for spiking, a spike is generated. Firing rate increases essentially linearly above this threshold.
Overall, those properties of the model are similar to the gamma-band ("40 Hz") subthreshold membrane oscillations observed in intracellular recordings of thalamic and cortical neurons \[[@pbio-0030141-b07],[@pbio-0030141-b31],[@pbio-0030141-b32],[@pbio-0030141-b33]\]. In both our simulations and these experiments, oscillations emerge at a precise depolarization threshold, with a sudden well-defined frequency and a continuously increasing amplitude.
Similar properties continued to be observed when 120 such oscillatory neurons, with randomized membrane parameters, were interconnected in a model thalamocortical column (as described in [Materials and Methods](#s4){ref-type="sec"}). [Figure 2](#pbio-0030141-g002){ref-type="fig"}B shows the temporal evolution of the average local field potential (LFP) emitted by the cortical excitatory neurons in response to variable levels of injected current *I* ~neuromodul~. There is still a threshold, now lowered to *I* ~neuromodul~ ≈ −0.8 μA/cm^2^, at which gamma-band oscillations emerge with a fixed frequency (initially 30--35 Hz) and with continuously increasing amplitude. The lowering of the threshold is due to random variability between neurons in the conductance of the Na^+^ and K^+^ channels responsible for generating membrane oscillations. Some neurons begin to oscillate at a lower value of the injected current, thus smoothing out the sharp transition observed within each single neuron. For the same reason, the spiking threshold is also lowered in this single-column simulation, with the number of emitted spikes increasing smoothly starting around *I* ~neuromodul~ ≈ −0.9 μA/cm^2^.
In the absence of spikes, the membrane oscillations of different neurons are independent of each other. Spikes, however, introduce coupling and result in transient periods of synchrony, which appear as occasional increases changes in firing rate accompanied by high-amplitude LFPs and a strong synchronization of the thalamocortical column. Continuous plotting of the LFP reveals a spontaneous, semirandom waxing and waning of bouts of coherent gamma-band oscillations with typical durations of 100--150 ms ([Figures 1](#pbio-0030141-g001){ref-type="fig"}B and [2](#pbio-0030141-g002){ref-type="fig"}B). The temporal evolution of the LFP appears largely chaotic and unpredictable, although the underlying simulation is strictly deterministic.
Altogether, these properties are comparable to the synchronized, depolarization-dependent, high-frequency thalamocortical oscillations that have been observed, for instance, in the cat thalamus and cortex \[[@pbio-0030141-b34],[@pbio-0030141-b35]\]. For simplicity, we did not include mechanisms for sleep-related, low-frequency oscillations, which are not the focus of the present model but have been simulated by others \[[@pbio-0030141-b09],[@pbio-0030141-b36]\].
Thalamocortical Resonance without Intrinsic Oscillators {#s2b}
-------------------------------------------------------
To evaluate the role of intrinsic cellular oscillators in generating the above thalamocortical oscillations, we reiterated the simulations using the "random spikes" model, with simplified neurons devoid of the Na^+^ and K^+^ channels, but with a random (Gaussian) moment-to-moment variability in spike initiation threshold. We verified that a single such passive, single-compartment, integrate-and-fire neuron, faced with a constant input current, is incapable of generating membrane oscillations below the spiking threshold (p. 163, \[[@pbio-0030141-b37]\]). In a single neuron, oscillations appear only once the injected neuromodulatory current is sufficient to depolarize the neuron beyond the spiking threshold. Even then, they do not exhibit a fixed central frequency, but cover a broad spectrum that progressively increases and broadens starting at 0 Hz, and stays mostly below 30 Hz with the present parameters ([Figure 2](#pbio-0030141-g002){ref-type="fig"}C).
Despite those major differences at the single-unit level, when 120 such neurons were connected into a thalamocortical column architecture, we observed structured spontaneous activity and phase transitions analogous to those of the cellular oscillator model ([Figure 2](#pbio-0030141-g002){ref-type="fig"}D; compare with [Figure 2](#pbio-0030141-g002){ref-type="fig"}B). Once a sufficient number of neurons were depolarized above the spiking threshold, the local field potential began to wax and wane within a range of gamma-band frequencies. Although this band was initially rather broad, it quickly narrowed to a predominant band at 40--45 Hz for higher values of the injected current. As shown in [Figure 2](#pbio-0030141-g002){ref-type="fig"}D, the critical properties of continuously increasing oscillation amplitude with a well-characterized frequency range remained, due no longer to intrinsic membrane properties, but to the temporal filtering properties of the several connection loops present in the thalamocortical column. Those loops have the effect of filtering random spiking activity, thus biasing neurons toward generating spikes at recurrent, rather randomly organized times. For instance, a major excitatory loop circles from the thalamus to the layer IV, supragranular, and infragranular cortical neurons, and finally back to the thalamus. In the simulation, the total length of synaptic delays along this loop was 15 ms, which, combined with membrane integration times, resulted in a total of approximately 25 ms/cycle, thus biasing the system toward 40-Hz oscillations.
Because intrinsic membrane oscillations have been reported, particularly in thalamic neurons \[[@pbio-0030141-b07]\], we also simulated a third type of model in which only a small subset of neurons were intrinsic cellular oscillators (the excitatory thalamic neurons, or 16.6% of the simulated cells), and all other neurons were of the "random spikes" type. The results, which appear on [Figure 2](#pbio-0030141-g002){ref-type="fig"}E, indicate the presence of waxing-and-waning LFP oscillations within a much narrower band of the gamma range than in the model with nonoscillating integrate-and-fire neurons. Thus, a small proportion of intrinsic oscillators, in resonance with the delays associated with recurrent thalamocortical connectivity, suffices to generate spontaneous activity with precise characteristics.
Facilitation of External Inputs by Spontaneous Activity {#s2c}
-------------------------------------------------------
We then examined how spontaneous activity affects activation caused by external stimuli. To this end, we measured the number of spikes evoked during stimulation by a 500-ms depolarizing current pulse of variable intensity, while orthogonally varying the amount of ascending neuromodulatory current *I* ~neuromodul~. This paradigm was applied both to a single neuron and to an entire thalamocortical column (in which case, only the thalamic excitatory neurons were stimulated), in both the intrinsic oscillator and the random spikes models ([Figure 3](#pbio-0030141-g003){ref-type="fig"}).
::: {#pbio-0030141-g003 .fig}
Figure 3
::: {.caption}
###### Combination of Intrinsic and External Activity in the Firing Rate of a Single Neuron or a Thalamocortical Column
Simulations examined the joint effects of a variable external input (left axis \[stimulus intensity\], parameter *g* ~input~, expressed in mS/cm^2^) and a variable neuromodulatory current that modified spontaneous activity and neuronal responsiveness (right axis \[*I* ~neuromodul~\], expressed in μA/cm^2^). We simulated a single neuron (A and C) or a thalamocortical column (B and D), either with the cellular oscillator model (A and B) or with the random spikes model (C and D). In each case, we monitored the mean number of spikes emitted by pyramidal neurons during a 500-ms duration input.
:::
:::
In all cases, the results indicated a facilitating effect of ascending neuromodulatory afferences on external stimuli. The internal neuromodulatory current and the external input current combined in a smooth and largely additive fashion ([Figure 3](#pbio-0030141-g003){ref-type="fig"}). Only for small values of the external current did the firing threshold cause a small nonlinearity: Small external inputs that failed to reach the threshold for firing when *I* ~neuromodul~ was close to zero could be rescued and had an impact on firing rate when *I* ~neuromodul~ was increased to 1.0--1.5 μA/cm^2^ (e.g., [Figure 3](#pbio-0030141-g003){ref-type="fig"}D). Outside this special regime of very small external inputs, the results indicated that external inputs generally continue to be processed by a single neuron or thalamocortical column at all levels of the neuromodulatory current, albeit with a smooth, continuous increase in input efficacy as this neuromodulatory current is gradually increased.
We also performed simulations of the thalamocortical column while varying the duration of external stimulation. Those simulations showed that elevated firing rates are observed only for a temporal interval equal to the stimulus duration. As soon as the external stimulus is removed, a rapid decrease toward the basal spontaneous firing rate is seen. Thus, with the present values of the parameters, the recurrent thalamocortical connectivity alone is not sufficient to establish long-lasting sustained activity.
Global Ignition of the Workspace by External Stimuli {#s2d}
----------------------------------------------------
We now describe how the response to external stimuli is radically changed once multiple columns are interconnected by long-distance connections into a global workspace. We simulated a multicolumn cortical model with four hierarchical levels, interconnected by long-distance corticocortical connections, and with two representations at each level. As in the previous section, spontaneous activity could be generated by intrinsic oscillations, random spikes, or a mixture of both; yet, those simulations again showed few differences, and we therefore only report the results of the intrinsic oscillation model.
In the previous section we saw how, in a single thalamocortical column, the neuronal activity caused by thalamic stimulation does not last much beyond the duration of stimulation. With the present parameters, the local circuitry of a column, although it includes excitatory loops, does not have sufficient strength to maintain activation over a durable interval. In the entire global workspace model, the effect of an identical stimulation is quite different. As show in [Figure 4](#pbio-0030141-g004){ref-type="fig"}, stimuli of very brief duration elicit only a correspondingly brief pulse of activity, traveling in a feedforward manner through each of the thalamocortical columns. Crucially, however, there exists a critical stimulus duration beyond which activation begins to reverberate for a considerably longer duration ([Figure 5](#pbio-0030141-g005){ref-type="fig"}A--[5](#pbio-0030141-g005){ref-type="fig"}C). The existence of such a threshold is a characteristic of the dynamical system created by the recurrent bottom-up and top-down connectivity of the workspace. If the input is sufficiently long and strong, it is able to generate sufficient activation in the higher areas (i.e., areas labeled B, C, and D in [Figure 1](#pbio-0030141-g001){ref-type="fig"}C), and those areas, in turn, send descending activation to lower areas, thus supporting the very activity that activated them in the first place. The result is the "ignition" of a coherent reverberating neuronal ensemble spanning across all areas (areas A--D in [Figure 1](#pbio-0030141-g001){ref-type="fig"}C) and lasting 200--300 ms, as previously described \[[@pbio-0030141-b22]\]. The two successive processes of bottom-up propagation (proportional to stimulus duration) followed by top-down amplification and recurrent firing (incommensurate to stimulus duration) can be clearly seen as two successive firing peaks in the spike train of pyramidal neurons ([Figure 5](#pbio-0030141-g005){ref-type="fig"}C, right-hand inserts).
::: {#pbio-0030141-g004 .fig}
Figure 4
::: {.caption}
###### A Comparison of Spontaneous and Evoked Global Ignition in the Workspace Model
Each tracing shows the temporal evolution of the LFP (bottom trace) and spikes (top trace) in one area of the model over a 1-s period, with *I* ~neuromodul~ = −1.0 μA/cm^2^. In the two sets of tracings on the left, within an interval characterized by spontaneous membrane oscillations and low spike rates, the neurons coding for assembly 2 become spontaneously activated at an elevated rate for a period of 200--300 ms. The arrows mark the approximate onset of activation, which starts earlier from the higher areas D2 and C2, with a concomitant inhibition of spontaneous activity in areas D1 and C2. In the two sets of tracings on the right, the same network was stimulated by stimulus 2 at the time marked by the arrow. Note that activation now starts by a bottom-up propagation from areas A2 to D2, followed by top-down amplification in the reverse direction. This two-step process creates a characteristic firing profile with two successive peaks, particularly visible in area A2.
:::
:::
::: {#pbio-0030141-g005 .fig}
Figure 5
::: {.caption}
###### Reverberant Activity in the Global Workspace Model Imposes a Nonlinear Threshold on Incoming Stimuli
Each panel shows the firing rate of pyramidal neurons in the lowest area A1 (coded by gray level) as a function of time (x-axis), in response to stimuli of variable duration (y-axis). Ignition is present when the duration of firing extends much beyond the duration of the stimulus, i.e., when a long tail of firing is present. In (A--C), the vigilance level---set by the ascending neuromodulation parameter *I* ~neuromodul~---is progressively increased. This increase leads to a systematic change in the minimum duration necessary for ignition. Note that the figure is an average over 20 trials at each duration. Thus, the small trailing activation that can be seen in (A) and (B) even at durations below the threshold is due to a very small proportion of trials in which ignition did occur, due to stochastic variability, as further explained in [Figure 6](#pbio-0030141-g006){ref-type="fig"}. The insets in (C) show the peristimulus-time histograms for stimulus of 10- or 100-ms duration, showing clearly the two firing peaks successively evoked by bottom-up activation and by top-down amplification. In a simulation in which top-down connections are disabled (D), the first peak is preserved, but the second peak is abolished.
:::
:::
During ignition of one representation, inhibition spreads to competing assemblies, yielding hyperpolarization and a suppression of ongoing oscillations (see [Figure 4](#pbio-0030141-g004){ref-type="fig"}, third column). Thus, ignition occurs in a distributed but topologically delimited assembly---only a subset of neurons, forming a synchronous coding assembly, are active---but this assembly is surrounded by a broad context (or "penumbra" \[[@pbio-0030141-b38]\]) of temporarily inhibited neurons. In our previous simulations \[[@pbio-0030141-b22]\], we showed that this inhibition creates a competition that prevents more than one global representation from being active at the same time. Thus, if two stimuli occur in brief succession, the second one may fail to trigger global ignition, a feature akin to the classical psychological phenomenon of the "attentional blink" \[[@pbio-0030141-b39],[@pbio-0030141-b40]\].
Ignition is a high-level collective phenomenon and is critically dependent on the integrity of long-distance recurrent connections. When top-down connections are disabled, as shown in [Figure 5](#pbio-0030141-g005){ref-type="fig"}D, then ignition fails to occur and activation duration becomes exclusively proportional to stimulus duration, as in the single-column model. If top-down connection strengths are weakened, or if the connection probability is lowered, then a corresponding increase of the threshold duration for ignition is seen (up to a critical value beyond which ignition cannot occur, regardless of stimulus strength).
In a more complete model of cortical connectivity, global ignition would propagate to cortical areas beyond areas C and D, thus broadcasting the identity of the input stimulus to many cerebral processors. We therefore propose that the simulated process of ignition provides a minimal but plausible neural mechanism for the sudden access of sensory stimuli into conscious perception. The nonlinear dynamic threshold exhibited by the model may explain the sigmoid perceptual curve typically observed in psychophysical experiments using brief stimuli \[[@pbio-0030141-b41]\], and thus the very existence of a "limen" (threshold) of consciousness.
Changes in Ascending Neuromodulation Affect the Ignition Threshold {#s2e}
------------------------------------------------------------------
In our model, the ignition threshold is not fixed, but can be changed by modifying the internal level of ascending neuromodulation. This is illustrated in [Figure 5](#pbio-0030141-g005){ref-type="fig"}A--[5](#pbio-0030141-g005){ref-type="fig"}C, showing the firing rate in response to stimuli of variable duration as the ascending neuromodulatory current *I* ~neuromodul~ is changed (we obtained similar results when stimulus intensity, rather than duration, was manipulated). For *I* ~neuromodul~ = −1.0 μA/cm^2^, even stimuli as short as 5 ms can be sufficient to trigger a long-lasting ignited state. For *I* ~neuromodul~ = −0.8 μA/cm^2^, this access threshold is lowered to 15 ms, and for *I* ~neuromodul~ = −0.6 μA/cm^2^, input stimuli of more than 30 ms are needed for global ignition. Interestingly, at this value of *I* ~neuromodul~, there are no longer any ongoing spikes or gamma-band oscillations (see [Figure 2](#pbio-0030141-g002){ref-type="fig"}). Thus, at this intermediate vigilance level, the model retains some capacity for a brief ignition by external stimuli, in the context of an otherwise quiet cortex. For even smaller levels of ascending neuromodulation, ignition is entirely abolished. Even very intense and durable external stimuli fail to elicit global ignition, and the evoked activation remains confined to the first thalamic and cortical steps in area A1. Those simulations demonstrate that a minimal level of spontaneous activity or "wakefulness" is necessary for the processing of external stimuli beyond the first few perceptual stages.
The effect of neuromodulation on the ignition threshold is mediated by at least two mechanisms. First, as described above ("Spontaneous Oscillatory Behavior in a Single Neuron and a Thalamocortical Column"), increasing neuromodulation brings neurons into spontaneous oscillatory states in which their membrane potential periodically approaches or reaches firing threshold, thus enhancing their capacity for prolonged firing. Second, as described in "Facilitation of External Inputs by Spontaneous Activity," increasing neuromodulation also enhanced the neuronal response to an identical input. In the presence of recurrent connections, the two effects combine in a nonlinear manner, making it possible for the very same stimulus to lead either to a major ignition or to a small, bottom-up wave depending on the internal state of vigilance defined by the ascending neuromodulation parameter.
In summary, one role of ascending neuromodulation in the model is to modify, in a nonlinear manner, the capacity of external stimuli to enter into a long-lasting state of distributed representation: The workspace must be in the awake state before global access by any external stimulus can occur. Furthermore, the threshold values of the stimulus parameters beyond which ignition occurs appear very sensitive to changes in either ascending neuromodulatory influences or top-down connection strength. Our simulations therefore predict that measuring the perceptual threshold for brief stimuli should provide a good marker of the integrity and functional state of vigilance of the workspace system.
Ignition as an All-or-None Stochastic Phenomenon {#s2f}
------------------------------------------------
According to the model, the thalamocortical network is under a permanent state of spontaneous activity. Therefore, the processing of an identical external input may change with the local context of ongoing activation. For stimuli close to threshold, this "resonance" of external inputs with internal spontaneous activity plays a determinant role in allowing or blocking ignition. To demonstrate this point, we analyzed the impact of stimuli presented close to the ignition threshold by simulating 100 trials with an identical stimulus (15 ms duration) and a fixed, intermediate intensity of neuromodulation (*I* ~neuromodul~ = −0.9 μA/cm^2^). We observed a considerable variability in neuronal responses, even in the first area, A1 ([Figure 6](#pbio-0030141-g006){ref-type="fig"}). The first peak of bottom-up activation was present in a majority of trials, even in area D, although with some small variance in its exact onset and intensity ([Figure 6](#pbio-0030141-g006){ref-type="fig"}A). However, the bulk of the variability occurred in the second peak, which could be completely absent or very intense and prolonged. In fact, the distribution of firing rates during this second peak was bimodal ([Figure 6](#pbio-0030141-g006){ref-type="fig"}B). Thus, the variability actually betrayed an all-or-none stochastic process, with fluctuations in ongoing activity modifying, on a trial-by-trial basis, the probability of crossing the dynamic threshold for ignition.
::: {#pbio-0030141-g006 .fig}
Figure 6
::: {.caption}
###### All-or-None Stochastic Fluctuations in the Processing of a Fixed Stimulus
\(A) Variability in sample spike trains evoked in area A1 by the same 15-ms stimulus, whose duration was close to ignition threshold (with *I* ~neuromodul~ = −0.9 μA/cm^2^). Each of the twenty lines represents one trial. Note that the first peak is present on a majority of trials, while most of the variability affects the second peak.
\(B) Distribution of the mean firing rate in area A1 during the second peak (75--225 ms poststimulus), showing a bimodal distribution.
\(C) Mean local field potential in area A1, computed separately for trials that lead to ignition (firing peak \> 40 spikes/s in the above time window) and for trials that did not (firing peak \< 15 spikes/s). Note (1) the essentially identical stimulus-induced waves up to the sharp divergence about 110 ms poststimulus; and (2) the presence of a small but significant difference prior to the stimulus, which indicates that ignition is more likely to occur when stimulus presentation coincides with the depolarized phase of spontaneous ongoing oscillations.
:::
:::
Because our simulation is deterministic, ignition success or failure is entirely predictable from the neurons\' current state and firing history. Yet, can it also be predicted from macroscopically observable variables? We classified trials as a function of whether ignition occurred or not, and then examined if those trials already differed in macroscopic variables such as the mean firing rate, mean local field potential, phase, or amplitude of spontaneous oscillations at the peak frequency within 100 ms prior to stimulus presentation. The only reliable predictor was the phase of spontaneous oscillations in area A1 (*p* \< 0.045, Kolmogorov-Smirnov test). As illustrated in [Figure 6](#pbio-0030141-g006){ref-type="fig"}C, ignition was more likely to occur when the first stimulus-evoked spikes reaching area A1 coincided with a period of relative depolarization, thus increasing the probability of spiking activity which could then propagate to higher regions.
Our simulations bear similarity to empirical observations by Fries et al. \[[@pbio-0030141-b42]\], who observed fluctuating, occasionally synchronized gamma-band oscillations in cat area V1 whose phase predicted the latency of firing in response to an external stimulus. Similarly, Super et al. \[[@pbio-0030141-b43]\] observed that the strength of prestimulus multiunit spiking activity in monkey area V1, and the correlation of this activity across multiple electrode sites, partly predicted whether the animal would report seeing a brief visual stimulus on a noisy background. Changes in baseline firing rate were not seen in the present analysis, but can be captured in our model by changes in the neuromodulatory input, which jointly affect baseline firing (see "Spontaneous Oscillatory Behavior in a Single Neuron and a Thalamocortical Column") and ignition probability (see "Ignition as an All-or-None Stochastic Phenomenon").
Fixed Duration of the Ignited State {#s2g}
-----------------------------------
In our simulations, ignition typically lasts 200--300 ms. Why does it ever stop? Spontaneous cessation is due to the spike-rate adaptation current *I* ~SRA~ in pyramidal neurons. This current leads to a progressive buildup of hyperpolarization, eventually overcoming the recurrent excitation provided by surrounded neurons. When the conductance *g* ~SRA~ is forced to 0, thus preventing spike-rate adaptation, long-lasting activity lasts essentially ad infinitum. Choosing intermediate values of this parameter allows modulation of the duration of sustained activity. Thus, the typical duration of 200--300 ms observed in the present results depends on the particular choice of parameters and should only be taken as indicative. What is truly generic is that, for a given choice of parameters, interconnected columns excited by a brief suprathreshold stimulus enter into a temporary metastable state of global activity lasting for an approximately fixed duration, independent of the original stimulus duration.
[Figure 7](#pbio-0030141-g007){ref-type="fig"} illustrates what happens in simulations with stimuli whose duration extends beyond this fixed ignition duration. Two regimes of firing are seen: below 230 ms, activation lasts for a constant value (about 230 ms) independent of stimulus duration. Above 230 ms, activation duration becomes proportional to stimulus duration again, because the neurons are already adapted and are therefore unable to maintain any further sustained activity. These two regimes can be tentatively compared to the psychophysical observations of Efron \[[@pbio-0030141-b44]\], who asked subjects to estimate the onset and offset of brief visual stimuli of variable duration. He observed that below a critical value, subjects judged all stimuli to be of fixed duration (about 130 ms), independent of the actual duration of presentation. Beyond this value, judgments became more accurate and linearly related to stimulus duration ([Figure 7](#pbio-0030141-g007){ref-type="fig"}B). The present simulations show that Efron\'s results are well captured, qualitatively at least, by a simple model, although the extent to which visual persistence is caused by local factors or by global loops as in the present simulation remains to be determined experimentally.
::: {#pbio-0030141-g007 .fig}
Figure 7
::: {.caption}
###### Duration of the Ignited State and Relation to "Visual Persistence"
\(A) Network response to stimuli of increasing duration (format as in [Figure 5](#pbio-0030141-g005){ref-type="fig"}; *I* ~neuromodul~ = −0.9 μA/cm^2^). Note how the duration of the ignition is initially constant and independent of the stimulus, then becomes proportional to it.
\(B) Empirical data on estimation of the subjective duration of a visual stimulus (replotted from data in Efron \[[@pbio-0030141-b44]\]). Both participants judged accurately the duration of stimuli above a critical duration, but judged duration to be constant below this value.
:::
:::
Spontaneous Ignition {#s2h}
--------------------
Up to now we have considered mostly the impact of an external stimulus onto the workspace system (ignition) and its interaction with ongoing oscillations of moderate intensity. However, simulations also revealed that ignition can occur spontaneously. In addition to waxing-and-waning gamma-band oscillations that are already present in the single-column model, the network with global long-distance connectivity occasionally falls into a state of globally synchronous elevated activity analogous to its ignition by external stimuli (see [Figure 4](#pbio-0030141-g004){ref-type="fig"}). For a period of about 200--300 ms, the neurons coding for a given representation become spontaneously active synchronously within each of the four areas, with firing rates approaching 50--100 Hz. Meanwhile, the neurons coding for the other representation are quiescent and oscillations are actively suppressed in the higher areas C and D (see [Figure 4](#pbio-0030141-g004){ref-type="fig"}, top left).
While this type of spontaneous states is highly similar to the ignited state evoked by external stimuli, a key difference is that during spontaneous ignition, elevated activity almost always starts at the highest level (area D), and then propagates downward to areas C, B, and A, where firing rates also tend to be lower. This is the reverse order of activation of that for external stimuli, and it underlines the top-down character of spontaneous ignition.
Another important difference is that spontaneous and evoked ignition tend to occur in different regimes of what we have referred to as vigilance. The above studies of ignition evoked by external stimuli were performed with −1.0 ≤ *I* ~neuromodul~ ≤ −0.6 μA/cm^2^. In this regime, spontaneous ignition is quite rare, and spontaneous activity is mostly characterized by low firing rates and waxing and waning gamma-band oscillations, which do not interfere much with incoming external stimuli. Spontaneous ignition becomes a prominent phenomenon when *I* ~neuromodul~ is of higher intensity (\< −1.0 μA/cm^2^). For *I* ~neuromodul~ \< −1.2 μA/cm^2^, spontaneous activity becomes characterized by a steady stream of successive ignitions. When more than two global assemblies are present, the network successively visits them in random order. Thus, at high vigilance levels, the activity of workspace neurons resembles a "stream" (in the sense of William James) of discrete episodes of spontaneous, metastable, coherent activation separated by sharp transitions.
Extinction of External Stimuli by Spontaneous Ignition {#s2i}
------------------------------------------------------
What happens to external stimuli when the workspace network is in a state of spontaneous global ignition? To study this, we first identified, in the absence of any external stimulation, an isolated period of spontaneous ignition of assembly 1 ([Figure 8](#pbio-0030141-g008){ref-type="fig"}, center graphs; *I* ~neuromodul~ = −1.0 μA/cm^2^). We then took advantage of the deterministic character of our simulation and re-ran the very same simulation, while presenting a second, competing stimulus in the input thalamic neurons of area A2, at various time lags relative to the spontaneous activation.
::: {#pbio-0030141-g008 .fig}
Figure 8
::: {.caption}
###### Competition between Spontaneous Workspace Activity and External Stimulation: A Possible Correlate of Inattentional Blindness
In this simulation, a time period was selected during which the representation of stimulus T1 became spontaneously activated in the absence of any external stimulation (note how the activation starts at the higher areas C and D, and propagates top-down to areas B1, then A1). Starting from the same initial state, the simulation was then digitally reproduced with the addition of an external stimulation of stimulus T2 at various times relative to the peak of spontaneous activity in area D. When T2 occurred before or after the period of spontaneous T1 activity, a normal global reverberant activation of T1 ensued. However, when T2 occurred during spontaneous T1 activity, T2 processing was limited to a short, bottom-up wave through areas A2, B2, and, to a lesser and more variable extent, areas C and D. Such blocking of T2-induced activity, similar to the attentional blink \[[@pbio-0030141-b22]\], may represent the cerebral basis of a state of "inattentional blindness," in which external stimuli fail to be detected during periods of spontaneous thought.
:::
:::
These simulations demonstrated a blocking effect of spontaneous activity. The stimulus presented during spontaneous ignition of a competing assembly does not lead to ignition, but is "extinguished" ([Figure 8](#pbio-0030141-g008){ref-type="fig"}, bottom graphs). Activation caused by stimulus 2 initially propagates in a bottom-up manner through A2, B2, and occasionally even C2 and D2. Only the first peak is present, however: Competition prevents the establishment of a strong reverberating assembly, and thus of long-lasting metastable activity coding for stimulus 2. When this stimulus is presented either before or after the spontaneous ignition, however, no such blocking occurs, and full ignition is seen. Further, if the stimulus is presented just before the spontaneous ignition, the latter fails to occur. In summary, we observe an all-or-none competition between spontaneous and evoked activations, with either one capable of preventing the other. In the discussion, we suggest that this phenomenon may relate to inattentional blindness in humans.
Discussion {#s3}
==========
Predictions and Experimental Test of the Model {#s3a}
----------------------------------------------
We have described the neuronal dynamics that emerge from a simplified model of multiple interconnected thalamocortical columns. This is an artificial and formal network, which by no means is anticipated to exactly reproduce the properties of actual thalamocortical networks, even in non-human species. Furthermore, although we have simulated a hierarchy of areas, each of them is very minimally implemented (one column for each representation), and the model therefore does not capture the increasingly abstract transformations that occur in actual cortical areas, nor the multiplicity of states they can have. Our purpose, however, is to examine to what extent the architectural principles and the cellular and molecular components introduced into the network may confer intrinsic dynamic properties that might be independent of the actual number of neurons and underlying complexity of the network. The correlations identified in the Results section between the present computer simulations and experimental data collected in humans and in non-human species legitimate, in our opinion, a close comparison with actual physiological and psychophysical experiments.
Two main states of the network have been observed: spontaneous gamma-band thalamocortical oscillations under the control of ascending neuromodulator systems; and, within a spontaneously active network, the sudden "ignition" of one out of many possible coherent states of high-level activity amidst cortical neurons with long-distance projections. In this discussion, we examine to what extent the delineation of these two dynamic states can capture empirical data on consciousness and its available neural correlates. We consider first how changes in spontaneous activity relate to the continuum of consciousness states that can be observed in the transitions between the awake state, sleep, anesthesia, or coma. Second, we discuss the all-or-none transition in neuronal activity associated with ignition and attempt to relate it to the transition from subliminal to conscious processing in various perceptual paradigms. Third, we discuss the interactions between vigilance and conscious access, and examine whether the model captures some of the processes underlying inattentional blindness.
Spontaneous Thalamocortical Rhythms and States of Vigilance {#s3b}
-----------------------------------------------------------
The transition between the awake and asleep states is known to be regulated by various diffuse ascending neuromodulatory systems located in the brainstem, hypothalamus, and basal forebrain, and liberating substances such as acetylcholine, noradrenalin, serotonin, and histamine in the cortex and thalamus. In particular, cholinergic neurons in the pedunculopontine nucleus increase their firing prior to awakening and, through their diffuse projections, depolarize thalamic neurons, directly or indirectly, switching them out of the slow bursting mode and into fast gamma-band oscillations \[[@pbio-0030141-b09]\]. Similar effects can be obtained by electrical stimulation of the brain stem or by direct application of acetylcholine \[[@pbio-0030141-b45]\]. One mechanism for this involves the closing of a leaky K^+^ conductance coupled to muscarinic cholinergic and α~1~-adrenergic receptors. Moreover, constitutive gain-of-function mutations in the α~4~- and β~2~-subunit genes of the nicotinic acetylcholine receptor cause autosomal dominant frontal lobe epilepsy \[[@pbio-0030141-b46]\], and deletion of the β~2~-subunit is accompanied by a decrease of "microarousals" that take place during slow-wave sleep \[[@pbio-0030141-b47]\].
The present simulation incorporated only minimal details of these mechanisms. For simplicity, we merely modeled increases in vigilance by changes in a single current *I* ~neuromodul~, summarizing the depolarizing influence of ascending neuromodulation systems onto thalamic and cortical neurons. Despite its extreme simplification, this mechanism was sufficient to generate a dynamic phase transition in our simulation, whose properties bear interesting similarities with actual empirical observations. We observed a threshold value of ascending neuromodulation beyond which structured neuronal activity emerged in the form of spontaneous thalamocortical oscillations in the gamma band (20--100 Hz, with a peak of the power spectrum around 40 Hz). This phenomenon was quite robust, since it was observed with two different mechanisms for spontaneous activity, either using intrinsic cellular oscillators or noisy spike generation (for other demonstrations of spontaneous oscillations in simpler network simulations, see \[[@pbio-0030141-b48],[@pbio-0030141-b49]\]). This robustness appears to arise from the structure and delays inherent to thalamocortical loops, which function as a filter and selectively enhance gamma-band oscillations even when fed with unstructured noise.
The waxing-and-waning synchronous bursts of oscillations that we observed bear similarity with empirical observations. During the waking state, thalamic neurons exhibit fast spontaneous oscillations of their membrane potential at frequencies in the gamma band (20--80 Hz). These oscillations exhibit transient periods of thalamocortical resonance, which are detectable macroscopically as bouts of gamma-band oscillations using electrophysiological recordings, for instance in the cat thalamus and cortex \[[@pbio-0030141-b34],[@pbio-0030141-b35]\], or in humans using electro- and magnetoencephalography \[[@pbio-0030141-b50]\]. When falling asleep, these fast and quickly changing rhythms disappear and are replaced by slower, more globally synchronized states of activity \[[@pbio-0030141-b09]\], including sleep spindles (7--14 Hz), followed in deeper stages of sleep by delta waves (1--4 Hz) and then slow-wave sleep (\< 1 Hz). Our model did not incorporate mechanisms for generating such slow sleep rhythms, which could be usefully added in a more complete future simulation (see, for instance, \[[@pbio-0030141-b36]\]).
Following Llinas et al. \[[@pbio-0030141-b07]\], we propose that the spontaneous oscillatory activity that arises in thalamocortical networks constitutes the neuronal basis of the state of consciousness referred to as vigilance. An original feature of our simulation is to characterize precisely, in terms of a dynamic phase transition, the change in thalamocortical activity that characterizes a change in vigilance. Extending previous work by Wang \[[@pbio-0030141-b29]\], we show that there exists a vigilance threshold around which thalamocortical activity changes suddenly according to a Hopf bifurcation. The combination of continuity and discontinuity that this bifurcation presents may shed interesting light on the nature of the awakening process at the neuronal level. The Hopf bifurcation is continuous in the amplitude of spontaneous activity, which increases steadily from zero as vigilance increases. In that respect, our model incorporates a true continuum of consciousness states, from high vigilance to drowsiness and the various states of sleep, anesthesia, or coma. It is consistent with empirical observations that the ratio of high-to-low frequencies in the scalp EEG changes with the depth of anesthesia and correlates with objective tests of vigilance and reportability and with subjective measures of consciousness \[[@pbio-0030141-b51],[@pbio-0030141-b52],[@pbio-0030141-b53]\].
However, the Hopf bifurcation is also discontinuous in frequency space. As ascending neuromodulation increases, high-frequency activity appears suddenly, at a precise threshold, and within a well-defined frequency band (see [Figure 2](#pbio-0030141-g002){ref-type="fig"}). This discontinuous jump in frequency space may capture the observation that during awakening or returning from anesthesia, there is a definite threshold for regaining of consciousness. According to our model, this threshold should coincide with the threshold for emergence of high-frequency, spontaneous thalamocortical oscillations. Note that, at the vigilance threshold, the thalamocortical high-frequency power drops quickly to zero. Thus, the precise value of the threshold might be difficult to locate empirically in noisy data. However, the Hopf bifurcation also predicts a continuous square-root increase in oscillation amplitude immediately above threshold (linear increase in power), a quantitative prediction that remains to be tested.
All the above properties of the Hopf bifurcation are generic, in the sense that they describe the behavior of a broad variety of dynamical systems \[[@pbio-0030141-b30]\]. Thus, they are likely to be independent of the particular details of the present simulation. Steyn-Ross et al. \[[@pbio-0030141-b54],[@pbio-0030141-b55],[@pbio-0030141-b56],[@pbio-0030141-b57]\], on the basis of a formal mathematical exploration of a simplified cortical column, have also suggested that a phase transition occurs in cortical networks during anesthesia. While their model also predicts a collapse of the high-frequency content of the EEG, it differs from ours in predicting a discrete jump in EEG power. This is a testable difference between the two models.
We close this section by stressing that our model considers a highly simplified cortex in which oscillations are often apparent in the firing trains of single pyramidal cells. This may be considered a limitation of our approach, given that oscillations are rarely obvious in actual recording of awake animals (e.g., \[[@pbio-0030141-b58]\]). However, the basic properties of our model persist even in the absence of intrinsic cellular oscillators. In that case, oscillations are much less obvious in single-neuron firing trains, but are a network property most apparent in the local field potential. Furthermore, our simulations use only a small number of cells. As the number of neurons per cortical area is increased, oscillations become distributed and less conspicuous at the single-cell level. Overall, our simulations, although obviously very simplified, are not incompatible with experimental observations in awake cats, monkeys, and humans, which indicates that oscillations do occur in awake animals and humans, particularly under conditions of focused, effortful processing, and are typically more prominent in the local field potential than in the spiking activity of isolated cells \[[@pbio-0030141-b59],[@pbio-0030141-b60],[@pbio-0030141-b61],[@pbio-0030141-b62],[@pbio-0030141-b63],[@pbio-0030141-b64],[@pbio-0030141-b65],[@pbio-0030141-b66],[@pbio-0030141-b67]\].
It is also important to note that oscillations do not play a computational role in our model, which does not involve temporal coding. Oscillations and long-distance synchrony merely emerge as a consequence of the high level of reverberating activity in the network. Because oscillations are also present during local processing within a single thalamocortical column, high-frequency coherence and synchrony are expected to be partially ambiguous indicators of conscious access, present during both subliminal processing and conscious access, but with a higher intensity and over more distant sites during the latter.
Anatomical Basis of the Awake State {#s3c}
-----------------------------------
Anatomically, our view predicts that, in the awake state, spontaneous activity is present in all areas, but exhibits a higher degree of organization (stream of discrete states) in higher cortical association areas, whose neurons are tightly interconnected at long-distance into a global neuronal workspace and mobilize other low-level areas in a top-down manner. Thus, we would expect cortical areas particularly rich in "workspace neurons" with long-distance connections (i.e., prefrontal, parietal, superior temporal, and cingulate cortices) to show the most intense and consistent spontaneous activity in the awake state. Other areas would also be active, but with more variability, reflecting the changes in contents contemplated by the subject at any given moment.
This prediction fits within a recent line of research that has examined the "resting state," "default mode," or "baseline" activity of the awake human brain at rest. This research has evidenced a broadly distributed network of areas active during rest, including dorsal and ventral medial prefrontal, lateral parietotemporal, and posterior cingulate cortices \[[@pbio-0030141-b12],[@pbio-0030141-b14],[@pbio-0030141-b68]\]. This network is not static and strictly confined, but constantly fluctuates in synchrony with changes in EEG spectral content \[[@pbio-0030141-b16]\]. Furthermore, prefrontal, parietal, and cingulate areas show the greatest drop in metabolism during various types of transitions away from the awake state, whether during anesthesia, sleep, coma, or the vegetative state \[[@pbio-0030141-b15],[@pbio-0030141-b69],[@pbio-0030141-b70],[@pbio-0030141-b71],[@pbio-0030141-b72],[@pbio-0030141-b73]\]. Paus \[[@pbio-0030141-b74]\], reviewing several neuroimaging studies of vigilance in humans, suggests that a distributed neural circuit with nodes in the pontomesencephalic tegmentum, basal forebrain, thalamus, and anterior cingulate shows an identical decrease in activation whenever subjects fall asleep, are anesthetized, or exhibit a spontaneous drop in vigilance. This network fits well with the circuit predicted by our model, namely ascending neuromodulation systems, higher association cortices, and their associated thalamic nuclei.
Interestingly, Balkin et al. \[[@pbio-0030141-b73]\], studying the reestablishment of regional cerebral blood flow during awakening, have observed a two-stage process, whereby the brainstem and thalamus are restored first, later followed by an increase in prefrontal-cingulate activation and functional connectivity. This observations may correspond, in our model, to the two transitions observed as vigilance increases: first, waxing-and-waning LFP oscillations appear in thalamocortical columns, establishing a state of quiet vigilance with moderate spontaneous activity, but with a capacity for ignition by external stimuli; and second, at higher levels of the neuromodulation parameter, a stream of spontaneously ignited states appears, originating from higher cortical areas.
Global Amplification and Conscious Access {#s3d}
-----------------------------------------
Once the network was placed in the appropriate state of vigilance, our simulations examined its reaction to external stimuli. We observed a two-step process. All stimuli, even very brief ones, yielded a brief pulse of bottom-up excitation in the two areas of the model closest to the input side (areas A and B). However, for stimuli whose duration exceeded a threshold, we observed a propagation of firing into higher cortical areas C and D, and the establishment of a long-lasting reverberating state simultaneously involving bottom-up propagation and top-down amplification. We called this sudden, nonlinear self-amplification process "ignition" (as originally applied to Hebbian cell assemblies, see \[[@pbio-0030141-b75],[@pbio-0030141-b76]\]). This term does not imply that the entire network becomes uncontrollably active. Rather, ignition reflects the activation of a topologically restricted assembly of neurons, distributed in many cortical areas, but all coding for a coherent mental object, and with joint inhibition of other neurons coding for competing objects.
Although we acknowledge that nonlinearities can occur at any stage in perceptual processing, including the retina, we speculate that the characteristic psychophysical sigmoidal curves observed when subjects report the perception of short or dim stimuli is related to the nonlinearity created by reverberant workspace interconnections. Psychological and neuronal models of sensory decisions assume a progressive integration of sensory evidence until the latter reaches a fixed threshold \[[@pbio-0030141-b77],[@pbio-0030141-b78]\]. Our model proposes that this threshold is not implemented locally within a single area, but is a distributed property of many distant association areas. One ensuing prediction, recently verified experimentally using a dual-task paradigm, is that two independent integrations should not be able to proceed simultaneously \[[@pbio-0030141-b79]\]. In essence, the neuronal global workspace acts as the "central bottleneck" postulated in psychological models of dual-task processing.
The threshold for workspace ignition again reflects a bifurcation in dynamical system theory, but note that this bifurcation differs markedly from the one underlying the vigilance threshold. Conscious access, according to our model, is characterized by an all-or-none, first-order transition in which the relevant order parameters of the network (firing rates, gamma-band oscillations, synchrony) all show a discontinuous jump to higher values (whereas changes in vigilance generate a continuum of states, as described above).
Experimentally, the model predicts that each content of consciousness should be characterized by (1) activation of specific thalamocortical columns in relevant cerebral processors holding the conscious content (e.g., columns of neurons sensitive to motion in area MT, if the percept is one of motion); (2) joint activation of a broadly distributed subpopulation of neurons in prefrontal, parietal, and cingulate cortices; and (3) long-lasting reverberating activity among those sites, creating an episode of phase synchrony in the gamma band. Importantly, the model also predicts that for external stimuli that remain below the ignition threshold, the same specialized processors should be initially activated, but without leading to the subsequent global ignition.
Only a few studies to date have compared brain responses to comparable stimuli in conscious and nonconscious conditions. Those studies, obtained in diverse paradigms including masking \[[@pbio-0030141-b80]\], binocular rivalry \[[@pbio-0030141-b81]\], change blindness \[[@pbio-0030141-b82]\], attentional blink \[[@pbio-0030141-b83],[@pbio-0030141-b84]\], and perceptual hysteresis \[[@pbio-0030141-b85]\] all suggest that the onset of conscious perception is associated with a sudden coactivation of parietal and frontal areas, often including the anterior cingulate \[[@pbio-0030141-b86]\]. Concomitantly, an amplification of activation in relevant posterior areas \[[@pbio-0030141-b80]\] is also seen during focal attention \[[@pbio-0030141-b87]\]. In EEG, bursts of induced gamma-band activity are observed during conscious access to a visual percept \[[@pbio-0030141-b88],[@pbio-0030141-b89]\].
At the single-cell level, recordings in V1 and inferotemporal cortex in animals trained to report their visual percepts indicate that an initial phasic peak of firing is unaffected by reportability, but that reportable perception is associated with the presence of a late sustained activation that is not seen for nonreportable masked stimuli \[[@pbio-0030141-b90],[@pbio-0030141-b91],[@pbio-0030141-b92],[@pbio-0030141-b93]\]. Thus, the firing histogram shows two successive peaks, with the second being present only during conscious perception, a profile highly similar to that observed in our simulations.
According to our model, the second peak, but not the first, should be affected by disrupting feedback from prefrontal, parietal, and cingulate cortices. This prediction could be tested by recording from visual cortices while reversibly cooling or inactivating distant association cortices. Note, however, that our model postulates a multiplicity of top-down loops, and it might not be possible experimentally to disrupt them all at once. Thus, such inactivation studies might not show a complete abolition of the ignited state, but only an elevation of the ignition threshold.
In our model, ignition places the thalamocortical network in a metastable state that lasts for a relatively fixed duration (e.g., 200--300 ms). Once an ignited assembly loses its support, the network again becomes available for either spontaneous or externally induced ignition. Thus, our simulations predict that multiple episodes of metastability should follow each other in a stream of discrete states, nested within a much slower fluctuation of vigilance. Few experimental studies are relevant this prediction. "Microstate" analyses have suggested that the human EEG consists of a series of quasistable states of 100--200 ms duration \[[@pbio-0030141-b94]\]. Likewise, Freeman has observed, in both awake cats and humans, sharp transitions in the phase of local potentials that occur simultaneously at distant sites across the scalp and delimit metastable periods of 100--300 ms duration \[[@pbio-0030141-b95]\]. At the single-cell level, hidden Markov models have been applied to the firing trains of simultaneously recorded neurons in the frontal cortex of a monkey performing a delayed-response task \[[@pbio-0030141-b96]\]. Those analyses revealed that neurons undergo a sequence of discrete states separated by abrupt changes, perhaps reflecting an internal rehearsal process. At present, however, these experimental forays remain isolated and have not yet been related to conscious reports.
Interactions between Ongoing Spontaneous Activity and External Stimuli {#s3e}
----------------------------------------------------------------------
An original aspect of the present simulations concerns the interactions between ongoing spontaneous activity and external stimuli. These interactions are mostly facilitatory: Higher spontaneous activity brings neurons closer to firing threshold, thus facilitating the detection of weak stimuli. However, very high spontaneous activity (spontaneous ignition) has a blocking role, preventing access to other external stimuli. Those two aspects lead to two distinct sets of predictions, which are considered in turn.
Concerning the first point, our model predicts that the threshold for conscious access (ignition) is not fixed, but decreases as vigilance increases. At one extreme, very low levels of vigilance prevent the possibility of ignition, even by long and intense stimuli. Such stimuli only lead to a short pulse of activation through the thalamus and areas A or sometimes B. Thus, we expect early sensory processing, but a lack of higher cortical processing, for sensory stimuli presented during altered states of consciousness. This prediction is consistent with empirical observations of auditory processing during sleep \[[@pbio-0030141-b97]\] or the vegetative state \[[@pbio-0030141-b98]\], where stimuli activated the thalamus and auditory cortex, but failed to generate the distributed state of correlated prefrontal, parietal, and cingulate activity observed in awake normal subjects. Similar observations have been made with tactile or pain stimuli, suggesting that the lack of prefrontal-parietal-cingulate ignition is quite characteristic of those states \[[@pbio-0030141-b99],[@pbio-0030141-b100]\]. As described above, these areas also show a global decrease in their baseline metabolism even in the absence of stimuli, suggesting a considerable decrease in spontaneous activity. It would be interesting to relate, subject by subject, the resting metabolism and the activation by external stimuli of varying duration, to verify the predicted inverse correlation between spontaneous activity and the conscious access threshold.
We note that our model predicts the existence of an intermediate state in which the network has essentially no spontaneous activity, but is still capable of global ignition by intense and durable stimuli. This may tentatively capture some aspects of the "minimally conscious state," where comatose subjects show fluctuating evidence for perception of environmental stimuli, and in whom distributed cortical activity, including cingulate cortex, can still be elicited by auditory or pain stimuli \[[@pbio-0030141-b101]\].
The inverse relation between vigilance and the conscious access threshold should also be present in the normal awake state, and could be manipulated empirically, for instance by sleep deprivation. Pharmacological agents such as nicotine, which can mimic and potentiate ascending cholinergic systems, might also have an influence on the perceptual threshold in visual masking or other psychophysical tests, which should be measurable both psychophysically and with brain imaging measures of ignition (e.g., prefrontal-cingulate activity in fMRI, P300 in event-related potentials). Finally, some disease processes might alter both spontaneous workspace activity and evoked ignition. In mice, deletion of the β~2~ subunit of the nicotinic receptor, which confers high-affinity acetylcholine or nicotine binding, leaves automatic navigation behavior intact, but interferes selectively with spontaneous exploration \[[@pbio-0030141-b102]\]. In schizophrenic patients, subliminal processing is intact, but the threshold for conscious perception of masked visual stimuli is increased, possibly relating to an impairment of top-down prefrontal-cingulate connectivity \[[@pbio-0030141-b103]\]. In such patients, we predict that nicotine might partially bring the conscious access threshold back toward its normal value. More generally, this threshold is predicted to provide a sensitive indicator of the anatomical integrity of top-down connections, and of the functional integrity of the vigilance system.
The facilitatory effects of spontaneous activity on the responsiveness to external stimuli are brought about by a simple mechanism that was observed in simulations of a single neuron or thalamocortical column: Larger membrane oscillations or impinging synaptic noise bring the neurons closer to firing threshold, where they are more sensitive to even small changes in their inputs. This mechanism has been demonstrated physiologically in cortical slices \[[@pbio-0030141-b104],[@pbio-0030141-b105]\]. Our simulations show, however, that this phenomenon is mitigated by another, more collective phenomenon, which emerges only in the simulation of the globally connected model. At high levels of spontaneous activity, spontaneous ignition of structured global neuronal assemblies frequently happens and, during its occurrence, ignition by external stimuli is prevented ([Figure 8](#pbio-0030141-g008){ref-type="fig"}).
Physiologically, there is some evidence for both of these aspects. First, optical imaging of visual cortex in anesthetized animals has revealed structured states of spontaneous ongoing activity, which have the same global organization as activity patterns evoked by external stimulation \[[@pbio-0030141-b10],[@pbio-0030141-b11]\]. Second, high levels of spontaneous activity have been found to inhibit the sensory responses evoked by external stimuli, for instance by whisker deflection in somatosensory cortex \[[@pbio-0030141-b106]\]. It has been noted that such interactions with ongoing activity can provide an explanation for the large variability in spike trains evoked by the very same sensory stimuli \[[@pbio-0030141-b106],[@pbio-0030141-b107]\]. Indeed, the present model predicts that there should be more variability in evoked spike trains in sensory areas during the normal awake state than during sleep or anesthesia.
The complete blocking of some incoming stimuli that occurs in our model offers a plausible explanation for the psychological phenomenon of "inattentional blindness" \[[@pbio-0030141-b26]\]. In this phenomenon, human observers engaged into an intense mental activity (such as detecting or counting stimuli of a certain type) become totally oblivious to other, irrelevant stimuli, even when they occur within the fovea for a long duration \[[@pbio-0030141-b27],[@pbio-0030141-b28]\]. Although inattentional blindness is typically studied in the laboratory by placing subjects in a predefined task, our simulations suggest that spontaneous trains of thought, unrelated to external stimuli and instructions, may also exert a temporary blocking. Our model predicts that this state should be characterized by (1) an intense prefrontal-parietal-cingulate activation by the distracting thought or object prior to the presentation of the target stimulus; and (2) a proportional reduction of the target-induced activation to a brief bottom-up activation in specialized processors ([Figure 8](#pbio-0030141-g008){ref-type="fig"}).
Only a single fMRI study to date provides direct support for prediction 2 \[[@pbio-0030141-b108]\]. However, there is considerably more evidence from the closely related phenomenon of the "attentional blink," in which the distracting object is presented a few hundreds of milliseconds prior to the target stimulus. In particular, Marois et al. \[[@pbio-0030141-b109]\] obtained data compatible with prediction 1: the intensity of parietofrontal activity by the distracting task predicts the extent to which the target is extinguished. Furthermore, Marois et al. \[[@pbio-0030141-b83]\] supported prediction 2: during the blink, an unseen picture of house still activates the specialized parahippocampal place area, but fails to activate correlated activity in parietal, prefrontal, and cingulate normally observed in nonblinked trials. Similarly, in magnetoencephalography, the long-range synchrony linking frontal, temporal, and parietal areas is reduced for blinked stimuli \[[@pbio-0030141-b84]\]. Sergent et al. \[[@pbio-0030141-b110]\] used psychophysical methods to verify another prediction of the present model, namely the all-or-none character of conscious perception during the blink. Future research should extend those paradigms using time-resolved neuroimaging methods, such as event-related potentials, to test the prediction that early bottom-up activation is preserved, but later activation and top-down recurrent reverberation are suppressed in an all-or-none manner during the blink (see, for example, \[[@pbio-0030141-b111],[@pbio-0030141-b112]\]).
Comparison with Other Theories of Consciousness {#s3f}
-----------------------------------------------
The psychological concept of a "global workspace" was initially proposed by Baars \[[@pbio-0030141-b25]\] as a cognitive model of consciousness. Baars suggested that the global workspace can be neurally related to the nonspecific nuclei of the thalamus, particular the reticular formation, though he mentioned in passing that "it is possible that corticocortical connections should also be included" (p. 123 in \[[@pbio-0030141-b25]\]). The present model provides a much more detailed neurobiological implementation, and proposes that long-distance cortical connections are essential, particularly those linking prefrontal cortex to associative areas of the parietal and cingulate cortices.
A mention of parietoprefrontal networks in relation to consciousness can be found in Shallice \[[@pbio-0030141-b113]\] and Frith et al. \[[@pbio-0030141-b114]\]. In their pioneering efforts to specify the neural correlates of consciousness, Crick and Koch \[[@pbio-0030141-b115]\] also mentioned the importance of connections to and from prefrontal cortex (though more recently, Crick and Koch \[[@pbio-0030141-b38]\] defended the opposite view, that prefrontal cortex contributes to an "unconscious homunculus"). Their work, however, led them to the position that primary visual area V1 could not participate in conscious contents. Contrariwise, we see no fundamental reason why top-down mobilization would exclude primary areas, including V1, for instance in a high-acuity perception or mental imagery task.
Indeed, Lamme and his colleagues \[[@pbio-0030141-b116]\] have produced a wealth of empirical data indicating that late amplification of the local field potential in V1 correlates tightly with visual consciousness and attention in the awake monkey. Lamme suggests that consciousness occurs when recurrent bottom-up and top-down interactions occur between V1 and distant areas. The present model can be seen as a partial implementation of Lamme\'s hypothesis. One point of divergence is that we emphasize the unity of consciousness, involving a single ignited state that connects many distant areas, while Lamme \[[@pbio-0030141-b117]\] proposes that connections local to visual areas may be sufficient for a form of phenomenological awareness without conscious reportability (see also \[[@pbio-0030141-b118]\]).
None of the proposals reviewed so far were specified in terms of formal neural architectures that could simulate the dynamics of conscious versus nonconscious tasks. Tononi and Edelman \[[@pbio-0030141-b119]\] emphasized the role of information integration and of reentrant connections in establishing a "dynamic core" linking distributed cortical and thalamic neurons, yet without proposing an explicit simulation of this process. While the present work rests in part on a formalism established by Lumer et al. \[[@pbio-0030141-b120]\], their simulations focused on early and presumably nonconscious visual processing, with reentrant connections mostly linking cortical neurons in a horizontal manner rather than the present long-distance, top-down scheme for access to consciousness.
Spontaneous Activity and Autonomy of the Organism {#s3g}
-------------------------------------------------
We close by noting that most theories of conscious processing have failed to recognize the important role of spontaneous thalamocortical activity (for exceptions, see \[[@pbio-0030141-b01],[@pbio-0030141-b07],[@pbio-0030141-b120]\]). By contrast, the present computational model proposes an essential role for spontaneous activity within an anatomically distinct set of "workspace neurons" in giving the formal organism an autonomy relative to the external world. Autonomy has several facets. First, in the absence of external signals, an autonomous organism must be capable of generating spontaneous representations and intentions (self-activation). Our present and past work \[[@pbio-0030141-b18],[@pbio-0030141-b19],[@pbio-0030141-b20]\] proposes that spontaneous neuronal activity is an essential component of this "projective style" \[[@pbio-0030141-b121],[@pbio-0030141-b122]\], which departs from the input-output view currently dominant in the neural network community. The present work shows how sources of noise such as spontaneous membrane oscillations and noisy synaptic transmission can be harnessed to generate a stream of highly organized states of self-activation in the complete absence of external inputs. This capacity is expected to play a crucial role in the spontaneous generation of novel, flexible behavior, as evidenced for instance in neuropsychological tests such as the Tower of London test or the Wisconsin card sorting test.
Second, even when submitted to external stimulation, an autonomous organism must be capable of representing internally only those stimuli that are relevant to the present situation. Relevant stimuli must first be selected, based on reward-based evaluation systems not included in the present model (but see \[[@pbio-0030141-b123]\]), and then maintained online over a time period often incommensurate with their actual input duration, all the while resisting distraction by irrelevant stimuli. The present work provides the basic building blocks for such a decoupling of part of the organism\'s internal activity from its current inputs. A crucial role is attributed to the strongly recurrent connectivity of cortical neurons in association areas, which collectively form a conscious workspace, an internal "milieu" buffered from the outside world, and within which mental hypotheses can be entertained and discarded at will. An inevitable consequence of this autonomy, however, is that some stimuli are inappropriately filtered out, thus causing inattentional blindness.
Materials and Methods {#s4}
=====================
{#s4a}
The general architecture of the neural network that forms the basis of the present simulations has been described elsewhere \[[@pbio-0030141-b22]\]. Here we expand upon this description by providing new specifications relevant to the generation of spontaneous activity.
### Model neurons {#s4a1}
Neurons are modeled as single-compartment integrate-and-fire units with membrane potential *V*. When *V* exceeds a threshold θ, a spike is recorded and transmitted to other neurons with a delay, and *V* is reset to −80 mV for a refractory period of 4 ms.
The temporal evolution of *V,* expressed in millivolts, is given by:
where *C* = 1 μF/cm^2^ and currents are expressed in μA/cm^2^. The leak conductance is *g* ~Leak~ = 0.1 mS/cm^2^, so that the membrane time constant *C*/*g* ~Leak~ is 10 ms.
*I* ~GABA~, *I* ~AMPA~, and *I* ~NMDA~ are the total synaptic currents, respectively, from fast GABAergic, AMPA and NMDA glutamatergic synapses \[[@pbio-0030141-b120]\]:
where the sum in each case is over all relevant neurons connected to postsynaptic neuron *i* (inhibitory interneurons for *I* ~GABA~, bottom-up and intra-column excitatory neurons for *I* ~AMPA~, and top-down excitatory neurons for *I* ~NMDA~). The *g~i,j~* are the synaptic strengths and *t* ~delay\ *i,j*~ the transmission delays in milliseconds from neuron *j* to neuron *i*. The functions *h~j~*(*t*) characterize the synaptic inputs from neuron *j* and are given by the convolution of the spike train *s~j~*(*t*) with the synaptic activation profile *a*(*t*):
where *a*(*t*) = α (e^−*t*/τ1^ − e^−*t*/τ2^) / (e^−*t*peak/τ1^ − e^−*t*peak/τ2^), with *t*peak = τ~1~ τ~2~ / (τ~1~ − τ~2~), where τ~1~ and τ~2~ are rise and decay constants characteristic of each channel \[[@pbio-0030141-b120]\]. In the NMDA case only, the effective conductance is scaled by *m* ~NMDA~(*V*), a factor that characterizes the voltage dependence of NMDA channels \[[@pbio-0030141-b37]\]:
*I* ~SRA~ is an additional current used to model spike-rate adaptation \[[@pbio-0030141-b37]\] that is present only in cortical pyramidal neurons:
After each spike, the conductance *g* ~SRA~ is increased by a small quantity (here 0.01 mS/cm^2^); otherwise, it decays exponentially according to τ~SRA~ d*g* ~SRA~ /d*t* = −*g* ~SRA~.
*I* ~input~ is the current applied during stimulus presentation to thalamic neurons of the lowest hierarchical level to simulate a visual or auditory input:
A final current, *I* ~neuromodul~, summarizes the known depolarizing effects of ascending activating systems such as those from cholinergic, noradrenergic, and serotoninergic nuclei in the brain stem, basal forebrain, and hypothalamus \[[@pbio-0030141-b124]\]. This parameter is used to control the level of vigilance.
### Origins of spontaneous activity {#s4a2}
We consider two possible sources for spontaneous activity, both of which are meant as theoretical idealizations on a continuum of possibilities. The first case, hereafter referred to as the "cellular oscillator model," corresponds to our initial publication \[[@pbio-0030141-b22]\] and is a purely deterministic model in which neurons follow simple differential equations incorporating persistent sodium and slowly inactivating potassium currents whose interplay generates intrinsic gamma-band oscillations of membrane potential \[[@pbio-0030141-b29]\], comparable to those recorded experimentally \[[@pbio-0030141-b07]\]. In this cellular oscillator model only, *I* ~NaP~ and *I* ~KS~ are persistent sodium and slowly inactivating potassium currents \[[@pbio-0030141-b29]\] whose simplified expression is
where *m* ~NaP~ = (1 + e^−(*V*+51)/5^)^−1^, and
where τ~KS~ d*m* ~KS~/d*t* = (1 + e^−(*V*\ +\ 34)/6.5^)^−1^ − *m* ~KS~, with τ~KS~ = 6 ms. We also describe another simulation, hereafter called the "random spikes model," in which intrinsic membrane oscillations are absent (*I* ~NaP~ = *I* ~KS~ = 0), but stochastic spontaneous activity arises from fast, random fluctuations in membrane potential, capturing the joint effects of synaptic and postsynaptic noise on spike initiation. Thus, in the random spikes model, on each time step the spike threshold θ varies randomly, independently for each neuron, as a Gaussian with mean −48 mV and standard deviation 5 mV, while in the deterministic cellular oscillator model, the spike threshold θ is fixed at −48 mV.
### Columnar organization and connectivity {#s4a3}
Each thalamocortical column comprised 80 excitatory and 40 inhibitory neurons organized in a three-layered structure, schematizing supragranular, infragranular, and layer IV cortical neurons, and a corresponding thalamic sector (see [Figure 1](#pbio-0030141-g001){ref-type="fig"}C). Within each layer, there were 20 excitatory and 10 inhibitory neurons.
Connections were established with a 60% probability. Connection parameters (synaptic strength γ, expressed in mS/cm^2^; and transmission latency *t* ~delay\ *i,j*~ *,* expressed in ms) were drawn from Gaussian distribution with 10% standard deviation around a fixed mean. All neurons connecting to a given area randomly contacted excitatory and inhibitory neurons with the same probability and strength.
Concerning inhibition, a broad variety of inhibitory cell types are known to exist in cortex. Here, we adopted a simplified scheme in which inhibitory neurons sent only horizontal connections to other neurons within their cortical layer of origin, both within a column (γ~GABA~ = 0.12, τ~delay~ = 2) and toward other competing columns within the same area (γ~GABA~ = 0.60, τ~delay~ = 2).
Concerning the intracolumnar connectivity, we adopted principles and parameter values that captured the major properties of translaminar connections \[[@pbio-0030141-b120],[@pbio-0030141-b125]\], although we did not attempt to capture the possible functional roles of the different layers (e.g., see \[[@pbio-0030141-b126]\]). Latencies were taken from Lumer et al. \[[@pbio-0030141-b120]\], which describes how they must be adjusted to take into account the coarseness of the present three-layer model and the fact that, in reality, some of these pathways are disynaptic. Thalamic excitatory neurons projected to layer IV (γ~AMPA~ = 0.20, τ~delay~ = 3) and, with lesser strength, to infragranular neurons (γ~AMPA~ = 0.10, τ~delay~ = 3). Layer IV excitatory neurons projected to supragranular neurons (γ~AMPA~ = 0.15, τ~delay~ = 2). Supragranular excitatory neurons projected to infragranular neurons (γ~AMPA~ = 0.10, τ~delay~ = 2). Finally, infragranular excitatory neurons projected to layer 4 (γ~AMPA~ = 0.05, τ~delay~ = 7), to supragranular neurons (γ~AMPA~ = 0.05, τ~delay~ = 7), and to the thalamus (γ~AMPA~ = 0.075, τ~delay~ = 8).
Concerning corticocortical projections, supragranular excitatory neurons of each area projected to layer IV of the next area (γ~AMPA~ = 0.05, τ~delay~ = 3). In agreement with physiological observations \[[@pbio-0030141-b127],[@pbio-0030141-b128]\], top-down connections were slower, more numerous, and more diffuse. They connected the supra- and infragranular excitatory neurons of a given column to the supra- and infragranular layers of all areas of a lower hierarchical level (see [Figure 1](#pbio-0030141-g001){ref-type="fig"}B). Strong top-down connections linked columns coding for the same stimulus (γ~NMDA~ = 0.05), while weaker top-down connections projected to all columns of a lower area (γ~NMDA~ = 0.025). Transmission delays increased with cortical distance (τ~delay~ = 5 + 3δ, with δ = 1 for consecutive areas, δ = 2 for areas two levels apart in the hierarchy, etc).
### Parameter values {#s4a4}
All parameter values were fixed and adopted without modification from \[[@pbio-0030141-b29],[@pbio-0030141-b37],[@pbio-0030141-b120]\]: *V* ~Rest~ = −67 m*V*, *V* ~Na~ = 55 mV, *V* ~K~ = −90 mV, *V* ~GABA~ = *V* ~SRA~ = −70 mV, *V* ~AMPA~ = *V* ~NMDA~ = 0 mV; α~GABA~ = 0.175, α~AMPA~ = 0.05, α~NMDA~ = 0.0075; τ1~GABA~ = 1 ms, τ1~AMPA~ = 0.5 ms, τ1~NMDA~ = 4 ms, τ2~GABA~ = 7 ms, τ2~AMPA~ = 2.4 ms, τ2~NMDA~ = 40 ms, τ~SRA~ = 200 ms; and *g* ~input~ = 0.06 mS/cm^2^. To avoid an artificial digital reproducibility of oscillations across neurons, the following parameters were generated randomly for each neuron, from a Gaussian distribution with a 5% standard deviation and the following mean: *g* ~NaP~ = 0.2 mS/cm^2^, *g* ~KS~ = 8 mS/cm^2^. To study variations in the state of spontaneous activity, *I* ~neuromodul~ was treated as the only experimental variable between 0 and −2 μA/cm^2^.
Two main measures were used to describe network activity: the mean number of spikes emitted per millisecond, and the mean membrane potential, both averaged across all pyramidal cells in a given column. For simplicity, we refer to the latter parameter as the LFP. Note, however, that actual electrophysiological recordings of the LFP are much more complex, in at least two respects: (1) LFPs mainly reflect postsynaptic potentials rather than spike activity; and (2) they vary across cortical layers. Given this complexity, we opted for the simplest possible summary measure of the collective evolution of the network; however, we verified that our results still held when other measures were used, for instance the mean synaptic inputs of pyramidal cells within a specified cortical layer.
This work was supported by the Institut National de la Santé et de la Recherche Médicale, Commissariat à l\'Energie Atomique, the Collège de France, the Association Française contre les Myopathies, the Communauté Economique Européenne, and the Association pour la Recherche contre le Cancer. We gratefully acknowledge useful discussions with Joaquim Forget, Lionel Naccache, Christophe Pallier, and Mariano Sigman.
**Competing interests.** The authors have declared that no competing interests exist.
**Author contributions.** SD and JPC conceived and designed the experiments. SD performed the experiments and analyzed the data. SD and JPC wrote the paper.
Citation: Dehaene S, Changeux JP (2005) Ongoing spontaneous activity controls access to consciousness: A neuronal model for inattentional blindness. PLoS Biol 3(5): e141.
EEG
: electroencephalogram
LFP
: local field potential
|
PubMed Central
|
2024-06-05T03:55:55.105401
|
2005-4-12
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074751/",
"journal": "PLoS Biol. 2005 May 12; 3(5):e141",
"authors": [
{
"first": "Stanislas",
"last": "Dehaene"
},
{
"first": "Jean-Pierre",
"last": "Changeux"
}
]
}
|
PMC1074752
|
Have you ever walked smack into a parking meter or tripped over something on the sidewalk? Embarrassing as such incidents may be, they\'re the product of normal brain function. The brain is continuously bombarded with sensory information about the environment but perceives just a fraction of these inputs. The rest---pertinent details or not---is filtered out. It\'s thought that consciousness emerges from the activity of multiple spontaneous neural processors that run in parallel and connect to a higher order cognitive network that mediates the conscious perception. But this higher order network has limited processing capacity. That means if you\'re distracted, your brain can\'t accommodate additional sensory information, like "there\'s a parking meter in front of you, look out!"
To understand how spontaneous brain processing interacts with higher order cognition, Stanislas Dehaene and Jean-Pierre Changeux modeled the dynamic properties of brain activity with computer simulations. Their simulations show that while spontaneous brain activity sometimes facilitates processing, more often it competes with external stimuli for access to consciousness. Intriguingly, the results of the computer simulations very closely match physiological and psychophysical experimental data and thus shed new light on how intrinsic brain activity modulates conscious perception.
Neurons are simulated in their model as single "integrate and fire" units, integrating the signals received from all connected cells and firing an action potential as soon as their threshold is exceeded. These units are nested in columns, which are multiply linked among themselves and thereby form hierarchical assemblies. Lower columns increase their firing activities upon perception of external stimuli. This excitation propagates upwards to higher processing areas in a bottom-up activation process, but the model also includes, critically, top-down connections that can amplify incoming inputs. Eventually, if the input is strong enough, the reverberating excitation results in "ignition" of the global workspace with all areas simultaneously displaying sustained high firing activities. In the ignited network, the information of the stimulus is globally available; in this simplified model ignition corresponds to the access to consciousness. Most interestingly, the ignition threshold is variable and depends on the intensity of spontaneous activity in the network prior to perception of the stimulus.
In their study, Dehaene and Changeux changed the values of only one input parameter: the ascending neuromodulatory current. This model parameter simulates the effect of various diffuse neuromodulatory systems located in lower regions of the brain, which regulate the transition between awake and asleep states by liberating a diversity of neurotransmitters in the cortex and thalamus, the upper brain regions. The dynamics of the neuronal network could clearly be separated into two broad states of activity. Below a certain threshold for the neuromodulatory current, ignition of the global workspace is not possible, no matter how strong the stimulus---this can be compared to the brain states of comatose patients lacking any sign of consciousness. Above the threshold, however, spontaneous neural activity emerges: firing signals are amplified in feedback loops in single columns of the neuronal network.
With higher vigilance states, weaker external stimuli are able to ignite the global workspace. But paying attention to one thing narrows your perceptive capacity. Once ignited by one stimulus, the network cannot consciously process any others. Dehaene and Changeux propose that spontaneous activity---which operates within an "anatomically distinct set of workplace neurons"---offers an organism a measure of autonomy relative to the external world. While this decoupling of internal thought and external stimuli does have its disadvantages---like that pesky parking meter---it also provides the opportunity for introspection and creativity, which the authors argue is likely to "play a crucial role in the spontaneous generation of novel, flexible behavior."
|
PubMed Central
|
2024-06-05T03:55:55.112805
|
2005-4-12
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074752/",
"journal": "PLoS Biol. 2005 May 12; 3(5):e179",
"authors": []
}
|
PMC1074811
|
Harrowing tales of starvation and endurance epitomize Antarctica\'s "heroic age," when men equipped with little more than fortitude struggled against a landscape seemingly designed to thwart their intentions. A menacing sea ice figures prominently in these improbable survival stories. Daunting to early-20th-century explorers---trapping (and ultimately crushing) Ernest Shackleton\'s *Endurance* and derailing Robert Scott\'s 1901 *Discovery* expedition---the seasonal pack ice is the lifeblood of Antarctica\'s marine ecosystem.
But as winter temperatures continue to climb in the Antarctic, the once-forbidding winter sea ice is starting to deteriorate. The ice pack is forming later and retreating earlier---and it\'s having a serious impact on the abundance of krill, the backbone of the Antarctic food chain.
"Sea ice is the engine that drives Antarctic ecosystems," says William Fraser, a principal investigator for the Palmer Long Term Ecological Research program (LTER) and president of Polar Oceans Research Group in Montana. "Many of the key species that govern ecosystem dynamics in the Antarctic have life histories that depend on the availability of winter sea ice. The most important species in most sectors of Antarctica is krill."
A major food source for Antarctic fish, penguins, pelagic seabirds, seals, and whales, krill (Euphausia superba) look like shrimp, but weigh just a gram as adults and measure about six centimeters long ([Figure 1](#pbio-0030127-g001){ref-type="fig"}). Norwegian for "whale food," krill aggregate in super-swarms that can reach a density of 30,000 individuals per square meter, attracting whales, which can eat three tons of krill in a single feeding, and fisheries, which net on average 100,000 metric tons per year.
::: {#pbio-0030127-g001 .fig}
Figure 1
::: {.caption}
###### Krill Abundance Has Dropped 80% in 30 Years
\(A) This gravid female is nearly ready to release her eggs.
\(B) Krill feed on phytoplankton, indicated by the green color in this specimen\'s digestive organ.
(Images: Langdon Quetin and Robin Ross, researchers at the Marine Science Institute, University of California, Santa Barbara, and funded by the Office of Polar Programs, National Science Foundation)
:::
:::
The waters off the Antarctic Peninsula favor high krill concentrations. "Once you get into extreme environments such as the Southern Ocean, diversity will decrease but the number of individuals will increase because the production can be very high," says Scott Gallager, a marine biologist at Woods Hole Oceanographic Institution (Woods Hole, Massachusetts, United States).
Production is particularly high along the sea ice edge, he says, because the ice is thinner, which allows more sunlight to penetrate, and because ocean mixing processes along the continental shelf cause an upwelling of nutrient-rich deep water. Increased nutrients support increased primary production along the ice edge. "If ice forms too late," says Gallager, "you don\'t get this higher production, which impacts zooplankton populations like the larval krill that graze the under-ice surface, feeding on ice algae. The ice edge is an absolutely critical habitat, a nursery, for larval krill."
But this krill hotspot is showing some of the most dramatic changes in sea ice extent. Fraser says the western Antarctic Peninsula has registered the "largest increase in temperatures on the planet"---on the order of 6 degrees Celsius---over the past 50 years. This warming trend, he says, has been particularly pronounced during the winter---"crunch time" for many key Antarctic species.
"There was a time when almost every winter experienced very heavy sea ice," Fraser says, "reaching out into the Drake Passage." Where once heavy sea ice would form on average four out of every five years, he says, now it forms just one or two years out of five ([Box 1](#box1){ref-type="boxed-text"}). In a typical year, ice starts to form in the coldest regions along the southern coast in late March/early April (austral fall), then works its way up the coast. But late ice years are becoming more common, with ice forming two to three weeks later. "These patterns are completely different from patterns that existed as recently as 30 to 40 or 50 years ago," says Fraser. "The whole system is becoming unhinged as a result of this enormous warming."
Box 1. Coupled Ocean-Atmosphere System Controls Sea Ice Extent
==============================================================
What are the chances that the western Antarctic Peninsula will start to see more heavy-ice years? Annual winter sea ice extent depends on air-sea interactions between a recently discovered atmospheric phenomenon called the Antarctic Circumpolar Wave ([Figure 3](#pbio-0030127-g003){ref-type="fig"}) and Circumpolar Deep Water (CDW), which is part of the massive Antarctic Circumpolar Current.
The Antarctic Circumpolar Wave can either bring air up off the Antarctic continent, making the peninsula colder than normal, or down out of the South Pacific or South America, making the peninsula warmer, explains Eileen Hofmann, oceanography professor at the Center for Coastal Physical Oceanography at Old Dominion University (Norfolk, Virginia, United States). "What seems to be happening is that the atmosphere and ocean are transitioning into another state because of the warming in the Antarctic Peninsula, and now there are fewer colder periods," says Hofmann, who also heads Southern Ocean GLOBEC (Global Oceans Ecosystem Dynamics). "It warms up particularly in the winter because you don\'t get the strong winds coming up the continent as frequently. And so less sea ice forms in the winter."
The southern boundary of the Antarctic Circumpolar Current meets the continental shelf along the Antarctic Peninsula. As the current moves back and forth along the shelf, it pumps deep water onto the shelf. "Where you get this upwelling of CDW," Hofmann says, "it preferentially selects for diatoms, the preferred food for Antarctic krill." And embryos released around deep water develop faster and hatch at shallower depths, which means they don\'t have as far to swim to reach the surface, which in turn means less chance of being eaten en route and a better chance of finding food sooner. Once the larvae reach surface waters, the circulation over the continent helps retain them along the western Antarctic Peninsula.
If there\'s no winter sea ice covering the ocean surface, the momentum from the atmosphere can go into the ocean and enhance mixing, which forces heat up to the surface and prevents sea ice from forming, Hofmann says. "If you raise the temperature to -1.7 degrees Centigrade"---salinity causes ocean water to freeze at −1.82 degrees C---"you get no sea ice. It\'s that small difference that makes the system very responsive to climate change."
And melting ice introduces freshwater, which is much lighter than seawater, says Scott Gallager of the Woods Hole Oceanographic Institution. "Freshwater would essentially float on the surface and cap off mixing and heat transfer," he explains, which ultimately speeds up the process of warming---producing an exponential increase in the rate of ice-edge retreat.
One of the potential scenarios with rising temperatures, says Hofmann, is that the Antarctic Circumpolar Current will move farther away from the continental shelf along the peninsula. "That would greatly diminish the supply of CDW, and change the physical and biological structure of the shelf," she says. Alternatively, the Antarctic Circumpolar Current would get pinned against the continental shelf. "If CDW is continually pumped onto the shelf, then it probably would warm up the whole shelf water above freezing," says Hofmann. And that, she says, could lead to a collapse of the sea ice.
Retreating Sea Ice and Krill Declines {#s2}
=====================================
In November 2004, the most comprehensive study to date of krill distribution and abundance in the Southern Ocean reported a catastrophic drop in krill numbers. Angus Atkinson, a marine biologist with the British Antarctic Survey, led the study. "We pulled together all the net samples we could lay our hands on that had been obtained in the Southern Ocean over the last 80 years," Atkinson says, analyzing nearly 12,000 krill summer net hauls taken from 1926--1939 and from 1976--2003.
"The Southern Ocean is an enormous area, and at least half of krill stocks were in this comparatively narrow sector between South Africa and the Antarctic Peninsula," he says. His team found a positive correlation between winter sea ice cover and the abundance of krill the following summer. There\'s evidence of a general decline in winter sea ice extent and duration, Atkinson says, and of a general decline in krill populations---down 80% over the past 30 years---over the entire southwest Atlantic sector.
Though krill populations showed big fluctuations in the early years, their average numbers were higher over a longer period, explains Volker Siegel, a krill biologist with the Sea Fisheries Research Institute in Hamburg, Germany, who worked with Atkinson. "Where in the early days you might have 100 krill per square meter on average---with fluctuations between, say 20 and 300---nowadays you might see 20 per square meter, which goes from 50 to five individuals per square meter."
Neither Atkinson nor Siegel can say for sure what\'s causing the decline, but both say the winter sea ice is clearly playing some role. Krill live about five to six years. During the breeding season, from December to March, embryos are released in the upper water column, and the larvae hatch at depths ranging from 400 meters to 1,500 meters, unlike many fish and other invertebrate larvae, which hatch at the surface. Larvae then have to swim up through the water column to reach the sea surface. Unlike adults, krill larvae don\'t have enough body fat to carry them through food shortages. "They\'ll starve if they have to rely on water-column food distributions, and that\'s where the sea ice comes in," says Atkinson. "The ice may also shelter them from predators, but one way or another, ice in the winter is good for young krill."
"We\'ve got to find out what\'s causing these changes and then we can start to predict what\'s going to happen with future scenarios of climate change," Atkinson says. "The other thing we\'ve got to do is look at alternative things which might be affecting krill. We might find there are other things declining as well as sea ice, such as their food, or there might be a change in the fertilization of the waters."
Accompanying the drop in krill abundance, Atkinson and Siegel found an increase in salps, transparent jelly-like creatures that typically inhabit warmer waters than krill. Expanding into the warmer waters, salp populations are increasing in the southern part of their range and replacing the krill. Most krill-dependent predators do not eat salp.
Krill Declines Ripple up the Food Chain {#s3}
=======================================
At a hearing on climate change impacts before the United States Senate Committee on Commerce, Science, and Transportation in May 2004, LTER\'s Fraser testified that the western Antarctic Peninsula\'s cold, dry polar marine ecosystem is gradually being replaced with a warm, moist maritime climate. While all the major components of the food web are responding to these changes, Fraser said, the clearest evidence comes from studies of two especially sensitive indicators of climate change: krill and penguins. "Trends in penguin populations provided some of the first evidence that sea ice conditions in some areas were deteriorating in response to climate warming," Fraser told the senators.
This evidence came from Fraser\'s own studies, over the past 30 years, of three Antarctic penguin species that share similar life histories (including a penchant for krill) but show striking contrasts in their relationship to the sea ice ([Figure 2](#pbio-0030127-g002){ref-type="fig"}). For Adélie (Pygoscelis adeliae) penguins, the presence of sea ice is absolutely essential for survival. Chinstrap (P. antarctica) and gentoo (P. papua) penguins, on the other hand, require the absence of sea ice. "Adélie penguins have experienced a nearly 70% decrease in their populations at our study sites on Anvers Island \[Palmer Station\] in the western Antarctic Peninsula over the last 30 years," says Fraser, "and there\'s evidence that other krill-dependent predators are beginning to decrease."
::: {#pbio-0030127-g002 .fig}
Figure 2
::: {.caption}
###### Antarctic Penguins Show Striking Contrasts in Their Relationship to the Sea Ice
\(A) Ice-dependent Adélie penguins, at their nesting grounds on Anvers Island, have lost 10,000 breeding pairs since 1975.
\(B) Ice-avoiding chinstrap penguins at Anvers Island are increasing in number and range.
\(C) Breeding gentoo penguins---an ice-avoiding species that has not inhabited Anvers Island sites for at least 800 years---are turning up at Palmer Station.
(Images: Donna Fraser)
:::
:::
That\'s a loss of 10,000 breeding pairs since 1975. The major factors underlying this precipitous decline, Fraser says, are retreating sea ice and increasing snowfall. (The loss of sea ice increases the flow of water vapor from the open ocean to the atmosphere, increasing precipitation.) "When sea ice forms," he explains, "it covers these regions of high production and the birds are just able to plop into the water into very good feeding areas." With a life history accustomed to the formation of sea ice at critical points in their life cycle, Adélies are finding themselves faced with an unpredictable sea ice cycle that outpaces their ability to adapt. "The birds just don\'t have the sea ice when they need it," Fraser says. As if losing critical winter habitat weren\'t bad enough, Adélies must also contend with the effects of increased snowfall. When the snow melts in the spring, it\'s flooding their nesting areas and drowning their eggs and chicks.
The population trends for "ice-avoiding" chinstraps and gentoos are quite different. Both species are increasing their populations and beginning to replace Adélie penguins across a broad range in the western Antarctic Peninsula. "We\'re seeing breeding gentoo penguins at Palmer Station," Fraser says, with a trace of astonishment. "The paleoecological record does not show that species at our study sites in the last 800 years."
Though chinstraps and gentoos also depend on krill, they\'ve dealt with krill declines by eating more fish and squid (which, not incidentally, also eat krill). Fraser believes this dietary flexibility, along with the increased availability of open water and a late breeding schedule (which protects their eggs and chicks from spring snow meltwater), largely explain their range extensions.
"Adélies don\'t seem capable of adjusting anything about their life history," says Fraser. "They\'re hard-wired to their breeding area, returning to an area year after year after year, even though conditions are deteriorating." And so, their numbers continue to plummet as more chicks perish. Another ice-dependent, krill-eating penguin species, the emperor (Aptenodytes forsteri), is not faring any better. "We have just one emperor colony in our study region, and it\'s decreasing very fast," Fraser says. "It\'s gone down from 300 breeding pairs to nine. They\'re on the verge of extinction in our study region."
A Question of Sustainability {#s4}
============================
A major issue in devising strategies to protect krill populations concerns the impact of krill fisheries. Since drastic declines have occurred in the absence of heavy fishing, it\'s especially important to establish the population dynamics of Antarctic krill. Siegel works with the Convention for the Conservation of Antarctic Marine Living Resources (CCAMLR) to develop sustainable fishing regulations. Based on a point estimate of 44 million metric tons of krill in the southwest Atlantic sector in 2000, CCAMLR calculated a potential yield of 4 million tons, far above the average 100,000-metric-ton catch today. Siegel believes the fisheries aren\'t posing a significant threat to krill stocks at this point, but says much remains to be learned about krill population structure.
Gallager is not so sure about the impact of krill fishing. Some krill fisheries operate near the island of South Georgia, east of the Falkland Islands. "We know that these populations are not self-sustaining, but require recruiting of adults from other locations, and probably from regions along the western peninsula of Antarctica," says Gallager. "If we go in and fish out the populations that are not self-seeding, then they could be entirely wiped out." That\'s why it\'s crucial to know which populations, if any, are self-sustaining, he says. "And we don\'t know a lot about that at all."
Toward that end, Gallager is working on an under-ice "zooplankton observatory" that would sit on the bottom of the ocean floor and continuously release a flotation package---"once an hour, for hopefully the next ten years"---outfitted with sonar and optical sensors up to the surface, then winch itself back down again. The plan is to deploy the observatory off Palmer Station in May 2006. The hope is that data gathered on over-wintering larval krill will shed light on the factors influencing krill survival over the long term. "It may be that this coupling between larval krill and ice is actually underpinning the entire question of krill population dynamics," says Gallager---in which case, understanding how the ice moves relative to the water currents and wind shear over the top will be critical.
Many other aspects of krill biology remain obscure as well. Biologists still don\'t understand the precise mechanisms required to enhance larval growth, reproduction, and recruitment (replenishing populations with new individuals), or how temperature fluctuations affect metabolism and larval growth rate. Robust tests of long-held theories of how the under-ice habitat sustains krill larvae require much more quantitative data---over time and over a wide scale---on larval abundance, distribution, and foraging behavior. But netting krill is not easy. Larvae tend to wedge themselves into nooks and crannies, defying divers\' attempts to nab them while protecting their expensive nets--- loaded with even more expensive electronic gear---from the ice.
Whatever is behind the correlation between sea ice decline and krill declines, the future of the Antarctic ecosystem hangs in the balance. A 2001 report by the Intergovernmental Panel on Climate Change predicted that the Antarctic Peninsula will experience some of the largest, most rapid climate changes on earth. If these trends of rising temperatures and decreasing sea ice continue, says LTER\'s Fraser, "what we are going to see in the next ten, 20, 30 years is a system that is completely different from the one that exists now. Adélies will become regionally extinct."
Concluding his testimony on climate change impacts, Fraser warned the US Senate committee that if future warming continues and the cycle of heavy ice years exceeds the life span of krill, the species will face a reproductive crisis. And that, he said, "will have catastrophic consequences to the integrity of this marine ecosystem."
In 1912, for the sake of a few emperor penguin eggs, Apsley Cherry-Garrard and two members of Scott\'s ill-fated polar expedition endured what Cherry-Garrard called "extremity of suffering" from which only "madness or death may give relief." The group believed the eggs might prove that the penguins were the missing link between "birds and the reptiles from which birds have sprung."
Our understanding has advanced light years since then, rendering such notions nearly quaint. Cherry-Garrard and his companions thought the forbidding Antarctic landscape immune to human assaults. Today, with this notion, too, proven false, one wonders if the damage can be reversed. In Atkinson\'s diplomatic phrase, "there are political issues involved" where global warming is concerned. But the clock is ticking. If the Antarctic ecosystem collapses, it won\'t be because scientists were off on a misguided search for penguin eggs.
::: {#pbio-0030127-g003 .fig}
Figure 3
::: {.caption}
###### Antarctic Circumpolar Wave
In 1996, oceanographers Warren White and Ray Peterson identified significant inter-annual variations in the atmospheric pressure at sea level, wind stress, sea surface temperature, and sea-ice extent over the Southern Ocean. They called this system of coupled anomalies the Antarctic Circumpolar Wave. This simplified schematic summarizes the inter-annual variations in sea surface temperature (red, warm; and blue, cold), atmospheric sea-level pressure (bold H and L), meridional wind stress (denoted by τ), and sea ice extent (gray lines), together with the mean course of the Antarctic Circumpolar Current (green). Heavy black arrows depict the general eastward motion of anomalies, while other arrows indicate communications between the circumpolar current and the more northerly subtropical gyres. (Image: Warren White, <http://jedac.ucsd.edu/ACW/index_research.html>)
:::
:::
**Citation:** Gross L (2005) As the Antarctic ice pack recedes, a fragile ecosystem hangs in the balance. PLoS Biol 3(4): e127.
Liza Gross is a science writer for the Public Library of Science. E-mail: <lgross@plos.org>
CCAMLR
: Convention for the Conservation of Antarctic Marine Living Resources
CDW
: Circumpolar Deep Water
LTER
: Palmer Long Term Ecological Research program
|
PubMed Central
|
2024-06-05T03:55:55.113605
|
2005-4-12
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074811/",
"journal": "PLoS Biol. 2005 Apr 12; 3(4):e127",
"authors": [
{
"first": "Liza",
"last": "Gross"
}
]
}
|
PMC1074812
|
The launch of *PLoS Biology*---only 18 months ago---was just a first step for the Public Library of Science. Our initial goal was to create a flagship journal for the broader PLoS mission by providing an open-access alternative to the best subscription journals in the life sciences, and to put open access firmly on the map.
Despite its youth, *PLoS Biology* is already becoming established as a publication of high standing, and *PLoS Medicine* is rapidly heading in the same direction. Submissions to *PLoS Biology* have steadily grown, recently surpassing the 1,000 mark, and every month sees international coverage of *PLoS Biology* articles in the media. With the increasing support for open access in general, the time is now ripe to build on the success of these two flagship journals by taking the next step and launching the PLoS community journals.
The PLoS community journals will give authors an opportunity to publish a greater range of high-quality papers in open-access journals, so that anyone can read, use, and build on their work. But these publications will also serve another function---they will provide examples that other journals can follow, and will increase confidence in the sustainability of open-access publishing as a business model. More publishers are exploring ways to remove existing subscription barriers and edge towards open access. And more funding agencies are expressing their commitment to make the findings of the research they support freely available to the public. The PLoS community journals will lend further weight to this inexorable shift towards open access.
Each PLoS community journal will cover a broad field of research---so the journals serve specific scientific communities. The journals are also run by the community---academic editors-in-chief and associate editors, supported by PLoS staff. And some of these journals will be collaborations with established community groups, such as the International Society for Computational Biology (ISCB), with whom we are partnering for the launch of *PLoS Computational Biology*. The twin strands of open-source software and public databases of biological information converge on this burgeoning discipline, and the case for an open-access journal in computational biology is easily made. The ISCB has taken a bold step, in the best interests of its discipline and membership. We hope that this action by the ISCB will inspire other scholarly societies to follow suit and will be only the first of many such collaborations between PLoS and other organizations.
The editorial teams running the first three PLoS community journals already comprise a group of over 80 researchers, each headed by an editor-in-chief: Philip E. Bourne (University of California, United States) for *PLoS Computational Biology* ([www.ploscompbiol.org](www.ploscompbiol.org)), Wayne N. Frankel (The Jackson Laboratory, United States) for *PLoS Genetics* ([www.plosgenetics.org](www.plosgenetics.org)), and John A. T. Young (Salk Institute, United States) for *PLoS Pathogens* ([www.plospathogens.org](www.plospathogens.org)). The willingness of so many leading researchers to devote precious time to these new journals is testament to the level of commitment to open access that now exists within the research community, and the trust that PLoS has gained as a publisher of science and medicine.
And what of the relationship between the PLoS community journals and *PLoS Biology*? Does *PLoS Biology* still want papers in the areas in which PLoS launches new journals? Will *PLoS Biology* editors be rejecting more such papers in the knowledge that they will find a home in fellow PLoS publications? Let us assure you that we---the editors of *PLoS Biology*---remain committed to publishing the best research across all of biology.
Moreover, our relationship with the PLoS community journals is one of strict editorial independence. There is some overlap of membership on the editorial boards of *PLoS Biology* and the community journals, reflecting a level of dedication to open access, but as with all scientists who serve on multiple editorial boards and reviewers who review for multiple journals, these individuals are governed by confidentiality. That said, if an author would like a manuscript that has been turned down by one journal to be passed on to another, along with the reviewers\' reports and their identities, we are happy to cooperate, subject to the permission of the reviewers. This can help to expedite the review process, saving time for authors, editors, and reviewers. The editors of *PLoS Biology* and all the PLoS journals are committed to offering a peer-review service that is as constructive, transparent, and efficient as possible.
In the world of scientific publishing, there is nothing quite like launching a journal, especially when the case for the journal is as strong as it is for the PLoS community journals. It\'s enthralling, nerve-racking, and relentless work. But when the manuscripts begin to arrive, the editorial process kicks into action, and the production team starts crafting the first accepted articles, it\'s hard to contain the excitement. PLoS is still a relatively small organization, and all our staff have played a part in preparing for the introduction of the new journals. But PLoS is much bigger than the people on the payroll. It\'s the research community that is making PLoS work, as demonstrated most emphatically by the editors-in-chief and editorial board members who have stepped up to launch the first three PLoS community journals. Please join us in making them a success, and enjoy your share of the excitement.
Mark Patterson, Catriona MacCallum, Hemai Parthasarathy, and Caitlin Sedwick are editors for *PLoS Biology*. E-mail: <mpatterson@plos.org>
|
PubMed Central
|
2024-06-05T03:55:55.115793
|
2005-4-12
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074812/",
"journal": "PLoS Biol. 2005 Apr 12; 3(4):e129",
"authors": [
{
"first": "Mark",
"last": "Patterson"
},
{
"first": "Catriona",
"last": "MacCallum"
},
{
"first": "Hemai",
"last": "Parthasarathy"
},
{
"first": "Caitlin",
"last": "Sedwick"
}
]
}
|
PMC1074813
|
"*"Alchemy: a process of transforming something common into something special"---Webster\'s Dictionary*"
A growing number of studies are reporting the isolation of cardiac stem cells from a variety of tissue sources and examining their effects on promoting the repair of the injured heart. In the current issue of *PLoS Biology*, the storyline takes an unexpected, interesting twist, as Neal Epstein and his colleagues report the isolation of a novel cell type from skeletal muscle that can adopt a highly differentiated cardiac muscle cell phenotype in vitro and in vivo \[[@pbio-0030131-b1]\].
In this study, the authors use a differential isolation procedure to remove the skeletal muscle cells and myoblasts (immature muscle cells), and then collect the cells that are negative for a cell-surface marker called Sca-1. Under defined in vitro conditions, these cells adopt a cardiomyocyte phenotype that goes beyond the simple expression of cardiac-restricted biochemical and molecular markers, extending to the types of single-cell physiological functions that are hallmarks of authentic cardiomyocytes, including action potentials, calcium transients, and contractile activity. They call the cells Spoc cells, an acronym for "skeletal-based precursor of cardiomyocytes."
The study goes on to show that these cells can adopt this phenotype without the addition of cytokines or agents such as azacytidine that are known to activate the muscle gene program in nonmuscle cells. The cells also adopt the cardiac phenotype following their in vivo implantation into the ischemic heart following myocardial infarction, suggesting a potential therapeutic utility for these cells. Since Spoc cells were isolated from murine skeletal muscle, they may eventually allow the use of sophisticated conditional genetic tracking techniques to monitor the migration, maturation, and differentiation of the cells in the in vivo context.
Of course, a study with results this unexpected also raises a number of intriguing questions. Identifying the native location of Spoc cells, as well as the in vivo niche that insulates them from entering the differentiated cardiac program, will be valuable. A rigorous exploration of their developmental origin and their relationship to the other well-known cell types in skeletal muscle should be forthcoming. In this regard, a set of skeletal muscle stem cells, distinct from myoblasts, has also been found \[[@pbio-0030131-b2]\], and the question arises as to whether Spoc cells are related to these other skeletal muscle progenitors.
Cardiomyocyte Precursors Abound {#s2}
===============================
Other studies have reported the isolation of cells that can differentiate into cardiac muscle from diverse noncardiac tissues. In vitro and in vivo studies have suggested that these cells can adopt defined features of the cardiomyocyte phenotype when they are implanted into injured heart following myocardial infarction. It is possible that these studies are pointing to the existence of a rare circulating pool of precursor cells, which are able to home to the heart.
To understand the provenance of these cells, it will become critical to identify a set of gene products, or markers, that are restricted to Spoc cells and to other potential progenitor cells. Since the protocols for the isolation of the Spoc cells have been developed in the mouse, it should also become possible to use state-of-the-art spatial and temporal control strategies for triggering irreversible lineage tracers in these cells in the in vivo context without isolating the cells per se. This will make it possible to follow the fate of these cells as they differentiate. Similar approaches have recently been used to identify a subset of rare, native cardiac progenitors (which are positive for the marker Islet-1) in the newborn hearts of mice, rats, and humans ([Figure 1](#pbio-0030131-g001){ref-type="fig"}; \[[@pbio-0030131-b3]\]). Exploring the relationship of the Spoc cells to these native cardioblasts could reveal shared pathways that drive their formation, renewal, and differentiation.
::: {#pbio-0030131-g001 .fig}
Figure 1
::: {.caption}
###### Islet-1-Positive Cardioblast in the Right Ventricular Chamber of the Neonatal Rat Heart
Islet-1 nuclear staining depicted in green, nuclear staining in blue, myocyte marker staining in red. The left panel shows triple staining; the right panel shows single staining for Islet-1 expression.
:::
:::
Defining a True Progenitor {#s3}
==========================
One of the inherent caveats of the present study is the potential for phenotypic drift that occurs with prolonged growth of any cell type. It is possible, for example, that the genetic program of these cells gradually changes with time such that they can more easily adopt the cardiac phenotype. Heart and skeletal muscle share many key gene regulatory factors and downstream genetic programs, and the loss of a few key negative regulatory checkpoints might allow the cells to differentiate towards a cardiac phenotype. Again, documenting the existence of these cells in the in vivo context, particularly during stages of embryonic myogenesis, will be key.
There is already a clear precedent for this type of phenotypic drift between cardiac and skeletal muscle. H9c2 cells were originally derived from rat heart, but actually represent skeletal myoblasts, because they fuse and form myotubes following serum withdrawal. These cells express myogenic factors in the MyoD family, and do not express cardiac-restricted factors. Although the cells were initially isolated on the basis of their expression of cardiac markers, this is an example phenotypic drift---that is, the cells have drifted from a cardiac phenotype during repeated passage rather than simply representing skeletal muscle cells resident within the heart. Thus, the isolation of an immature cell from a heart does not necessarily denote that it is serving as a cardiac progenitor or stem cell. Potentially the same issue might hold for putative cardiac progenitor cells isolated from skeletal muscle, where the phenotype observed may not necessarily reflect its role in vivo within skeletal muscle.
One of the difficulties with alchemy and this new age of cardiac myocyte biology is related to defining a rigorous set of criteria that allow one to make the claim that the cell type of interest is truly a cardiac progenitor or stem cell and is acquiring a fully differentiated phenotype \[[@pbio-0030131-b4]\]. In this regard, the paper by Epstein and colleagues has done an admirable job of scoring for functional phenotypes (for example, having action potentials and calcium transients) that are far beyond the simple expression of cardiac muscle markers, which has been the phenotypic endpoint for many previous papers in the field. In addition, the authors go on to show the acquisition of the differentiated cardiac phenotype in the in vivo state in the absence of fusion with neighboring cardiac muscle cells, which has been a confounding variable in most other studies of this type. If a precursor cell simply fuses with a fully differentiated cell in a tissue, one could easily be fooled into thinking that the precursor has itself differentiated, as has been found for many types of putative cardiac stem cells \[[@pbio-0030131-b5]\]. Fortunately, techniques have been devised to detect such events.
In surveying the growing number of studies and claims of new cells that can acquire some type of heart cell phenotype in vitro or in vivo (for a review see \[[@pbio-0030131-b6]\]), it will become increasingly important to create a set of rigorous criteria that would distinguish between the following: authentic progenitor cells that are already committed to the cardiac lineage; pluripotent stem cells that can infrequently adopt the cardiac phenotype; phenotypic drift of other muscle progenitors with an increased propensity to enter cardiac lineages; and a variety of other cell types that can aberrantly express cardiac markers ectopically or by fusion with neighboring cardiac muscle cells.
Finding cell-type-specific markers for these cells will be critical, as opposed to generalized markers that do not allow the in vivo discrimination of their precise localization, mobilization, and differentiation in the intact muscle. The gold standard would then become the isolation of the cells from the intact organ after differentiation has occurred by creating genetically based or antibody-based approaches to identify and/or purify the already differentiated progeny from the intact muscle. Alternatively, new two-photon confocal microscopy approaches to identify the cells and then to monitor cardiac function in the intact heart should prove valuable \[[@pbio-0030131-b7]\].
For two decades, the bulk of our knowledge of molecular pathways that guide cardiac growth, development, and disease has been gleaned from a combination of in vivo studies in genetically engineered mice and primary cultures of neonatal and adult rat cardiomyocytes. Perhaps the most scientifically exciting aspect of this new age of cardiomyocyte biology is not simply related to cardiac repair. To date, there have been no continuous differentiated cardiac cell lines, a fact that has hampered the field for decades. The development of well-characterized cardiac progenitor cells offers the promise of using real genetic-based approaches to rapidly define the complex pathways that guide cardiac contractility, excitability, and lineage diversification into atrial, ventricular, and conduction system myocyte cell lineages.
**Citation:** Chien KR (2005) Alchemy and the new age of cardiac muscle cell biology. PLoS Biol 3(4): e131.
Kenneth R. Chien is at the University of California at San Diego Institute of Molecular Medicine, La Jolla, California, United States of America. E-mail: <kchien@partners.org>
|
PubMed Central
|
2024-06-05T03:55:55.116547
|
2005-4-12
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074813/",
"journal": "PLoS Biol. 2005 Apr 12; 3(4):e131",
"authors": [
{
"first": "Kenneth R",
"last": "Chien"
}
]
}
|
PMC1074814
|
::: {#pbio-0030136-g001 .fig}
:::
On the basis of his retrospective at the San Francisco Museum of Modern Art, it is evident that Roy Lichtenstein forged a narrow trajectory through the thickets of contemporary art. Unlike some of his protean fellow artists, such as David Hockney or Frank Stella, he found his style early in his career and followed its course with almost scientific precision for his entire life. Through an obsessive focus on the techniques of graphic production, he turned popular imagery into a high form of art and an exploration of the limits of visual perception.
The core concept of Lichtenstein\'s art is the iconic---drastic simplification of line and shadowing to reduce to the Platonic ideal of the object depicted---golf ball, truck tire, composition book, girl in love, or even the basic brushstroke of a daub of paint \[[@pbio-0030136-b1]\]. The basic inspiration of these icons is advertising imagery and comic book frames. Throughout his artistic life, he drew his imagery from the detritus of printed material---gum wrapper comics and small ads from the backs of newspapers---deliberately elevating visual sources at the opposite pole from the high art of the Renaissance. His first comic works were motivated by his sons, who challenged him by saying if he was such a great artist, could he draw Mickey Mouse for them? Lichtenstein was so stimulated by the result that he worked in the comic book style from then on. Interesting, Chuck Close tells a similar story about his own work, although in reverse. In his case a group of school kids were not "getting" his art and challenged him in the same way to prove his expertise. Close passed the test, but continued in his compulsive style, exploring the minutiae of the pixel structure in his grandiose portraits, while Lichtenstein explored the iconographic largesse of the new domain opened up by this challenge.
Lichtenstein\'s cartoon style of representation can be seen as an ironic commentary on the elitism of art, implying that art is merely a selection from the endless variety of images that bombard us. His self-mockery is exemplified by the second picture in the exhibition, a six-foot canvas bearing the letters "ART" \[[@pbio-0030136-b2]\]. In this piece Lichtenstein is explicitly interrogating the artistic community on the boundaries of the concept of art. Is the raw symbol placed in a gallery sufficient to constitute an artistic statement? This was a game that artists had been playing in earnest since the turn of the 20th century---Lichtenstein came up with one more variation on the theme. His later work consists of an extensive series of graphic "reflections" on the themes of classical art, from the second period of Pompeii, to the light explorations of the Impressionists, to Chinese landscapes, all in his egregious comic book pastiche.
The iconic is by now commonplace---the idea that cartoons capture the essence of an image was popularized by Fred Attneave in 1954 \[[@pbio-0030136-b3]\], when he drew attention to the way the outline of an object (and, indeed, the points of maximum curvature of the outline) captures most of the information of the full object image. The rest of the image can be thrown away without significant loss of its import. Nobel-Prize-winning experimental work conducted by David Hubel and Torsten Wiesel in the late 1950s consolidated this notion with the discovery that individual neurons in the visual cortex can be characterized as simple line detectors, i.e., that they are most active when a line of a particular orientation is found in a particular part of the visual world \[[@pbio-0030136-b4]\]. The work of Hubel and Wiesel illustrated that neural processing encodes the most relevant features in complex images. Indeed, they discovered a population of neurons called end-stopped cells that respond to the ends of lines and points of high line curvature in the manner required by Attneave\'s analysis. Interestingly, this confluence immediately predated Lichtenstein\'s entrée into the world of the comic book image, but the relationship seems to be coincidental in view of the iconographic symbolism of his choices. The likelihood that he had heard of these scientific developments seems remote, particularly in view of the fact that Andy Warhol was following the same track of using cartoon material, reputedly in mutual ignorance of Lichtenstein\'s breakthrough.
Beyond the iconographic, Lichtenstein plays with the visual impression derived from enlarging the halftone dots of the gum-wrapper comics. Through time, the dots become progressively larger and more insistent, emerging from their role as background fillers to dominate the entire canvas in a dizzying field of scintillations. In this sense, Lichtenstein seems to go beyond the role of cultural expositor and gadfly to explore the sensory implications of the optical redundancy of the printer\'s screen. The regular dot arrays shimmer and scintillate, ingraining themselves in our neural memory and projecting onto the gallery walls and neighboring paintings in a reminder of our visual fallibility. Such effects represent a resonance with so-called optical art ("op art"), a style promoted in the mid 1960s---by Bridget Riley and Victor Vasarely, in particular---that relies on visual illusion generated at the early levels of the nervous system: the retinal, the receptoral, the oculomotor, and the neural. Despite its name, it is not concerned with strictly optical effects such as diffraction, diffusion, interference, scintillation, polarization, and related optical phenomena. It is concerned with the visual and perceptual effects of dancing grids, jazzy dots, clashing colors, sliding waves, and so on.
Many of Lichtenstein\'s effects are a by-product of the printer\'s screen structure that is enlarged along with the other details of the printed image structure. His work seems to have been a major precursor of the op art movement, although he is not generally identified as a member of it. Indeed, he plays with the dot-screen as a theme in his later works, notably in the vast *Mirror in Six Panels* (1971), which shows nothing but the mirror surface reflecting empty space, apparently rendered in the transparent sheets of Benday dots in common use by graphic artists. Refreshingly, this is one of the few works that does not contain references to other art genres, but jousts with the concept of the image itself, again a reflection of nothing at all.
For the visual scientist, the most compelling painting in the exhibition may be *Rouen Cathedral Set V* (1969) \[[@pbio-0030136-b5]\], a meditation on Claude Monet\'s mediation on Rouen Cathedral, itself a series of impressionistic paintings of the cathedral in different lighting conditions. Here Lichtenstein abandons the cartoon-style bravura of line and text bubbles in a triplet of silk-screen close-ups of Monet\'s painterly impressions, differing only in the choice of colors for the three panels. The dot-screen now plays the role of a muslin or gauze curtain through which the cathedral is glimpsed, forming a vibrant haze that formalizes the image space into a kind of crystallized transparency that never quite settles into known categories of visual experience. While the left and right panels of the triptych are in bold shades of color, the central panel is rendered in accurate red-green isoluminance. As discovered by Richard Gregory, form processing is much weakened when the luminance differences are removed and forms are represented in colors that are accurately equated for their luminance values \[[@pbio-0030136-b6]\]. Although the colors are well seen, the form seems to shimmer and fluctuate, indicating that the shape-processing mechanisms are not well activated by the pure color differences. In Lichtenstein\'s Monet, the shimmer of the isoluminance interplays with the shimmer of the dot-screen to evoke a visual enigma, as we explore the image space to see whether the structure is indeed the same as in the flanking panels. One of the pleasures of art is its ability to slow down our sensory processing so that we become aware of the processes themselves, not just their symbolic role in our goal-oriented lives. Very few artists have played with the power of isoluminance to achieve this role in form processing: Lichtenstein seems to have been on to this property a decade earlier than Gregory, although he soon retreats back to the boldness of his cartoon pop-art style to explore a potpourri of the icons of classic sources.
There is much more that could be analyzed to place Lichtenstein in an art-historical framework, but perhaps one should just enjoy the power of the concrete image, simplified to its high-tone essentials and projected at large visual angle onto our excitable retinas. One comes away from the exhibit with a sense of the power of raw imagery that one may not have felt since the grade-school days of reading illicit comics when one was supposed to be learning the dates of battle sequences through history. But what does Lichtenstein\'s dot obsession reveal about neural processing? Why does repeated fine-grain structure wreak such havoc with our visual stability? This question was raised, in particular, by Donald MacKay with his high-density radial ray figure ([Figure 1](#pbio-0030136-g002){ref-type="fig"}), which generates powerful complementary effects in both current viewing and as an aftereffect \[[@pbio-0030136-b7]\]. Just why high-density dots and lines elicit such powerful responses from our visual apparatus remains unexplained. Indeed, the issue does not seem even to be a topic of current research interest, despite the proliferation of research activities in visual processing in general. Contemplation of an exhibit such as Lichtenstein\'s sparks a realization of the wealth of neural processes still to be studied and explored.
::: {#pbio-0030136-g002 .fig}
Figure 1
::: {.caption}
###### MacKay Ray Figure
Notice the chrysanthemum effect while viewing the image. Fixate for ten seconds on the center of the figure, then transfer gaze to the blank fixation (F) and notice the streaming circular effects in the blank area, roughly orthogonal to the orientation of the rays.
:::
:::
**Citation:** Tyler CW (2005) Traversing the highwire from pop to optical. PLoS Biol 3(4): e136.
Christopher W. Tyler is Head of the Brain Imaging Center at the Smith-Kettlewell Eye Research Institute, San Francsico, California, United States of America. E-mail: <cwt@ski.org>
|
PubMed Central
|
2024-06-05T03:55:55.117780
|
2005-4-12
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074814/",
"journal": "PLoS Biol. 2005 Apr 12; 3(4):e136",
"authors": [
{
"first": "Christopher W",
"last": "Tyler"
}
]
}
|
PMC1074815
|
Jakob von Uexküll coined the term *Umwelt* to describe the subjective world of animals. The world that animals perceive is not an objective, veridical representation of the physical world, he argued, but is instead a product of the particular sense organs that each species has acquired in its evolutionary history \[[@pbio-0030137-b1]\]. Many animals have sensory abilities that humans don\'t, such as a magnetic compass sense in birds \[[@pbio-0030137-b2]\] or sensitivity to electric fields in fish \[[@pbio-0030137-b3]\]. But even within sensory modalities shared by many animals, such as vision, hearing, and olfaction, there are strong differences between species. For example, bees, but not humans, can see UV light \[[@pbio-0030137-b4],[@pbio-0030137-b5]\] and smell carbon dioxide \[[@pbio-0030137-b6]\], and bats can hear ultrasound \[[@pbio-0030137-b7]\]. But what exactly is the structure of the perceptual worlds proposed by von Uexküll? Can we draw them on paper in the form of maps, allowing us to visualize a particular animal\'s subjective view of the world? Will such maps allow us to predict the similarity of two stimuli (e.g., two colors or two scents) by inspecting the distance between the loci they produce in a perceptual space? Does understanding the metrics of such maps help us predict how stimulus mixtures will be perceived?
Even though perceptual sensations may strike us as ethereal, they must be based on patterns of activity in neuronal hardware---thus, we need to look into the brains of animals to see how the neuronal circuitry processes the information from the sense organs. The study by Guerrieri et al. \[[@pbio-0030137-b8]\] on odor space in honeybees in this issue is an excellent example of how behavioral studies can be paired with neurobiological data to access the perceptual space of an animal. Here, we contrast the complexity of scent perception with two examples of simpler perceptual worlds, vertebrate frequency perception and bee color vision.
The Perception of Pitch in the Auditory System {#s2}
==============================================
Sound is a mechanical vibration whose most defining characteristic is neatly arranged along one dimension---frequency (of waves with alternating high and low pressure). The mechanical structure of our inner ear tidily maps the frequency of sound onto different positions of the basilar membrane in the snail-shaped cochlea \[[@pbio-0030137-b9]\]. The width and flexibility of this membrane increases with distance from the oval window (the point of entry of sound). The result is that, when the frequency of sound is high, it will produce a peak of vibration near the oval window ([Figure 1A](#pbio-0030137-g001){ref-type="fig"}); if the frequency is low, the peak of vibration will be nearer the far (apical) end of the cochlea \[[@pbio-0030137-b10]\]. In this sense, the cochlea processes sound similarly to how a prism acts on white light \[[@pbio-0030137-b11]\]: it decomposes mixtures of sound frequencies into their components, and maps them onto different spatial positions within the cochlea. The thousands of mechanoreceptors distributed along the basilar membrane need not be tuned to different frequencies, as in color vision, where each receptor type responds most strongly in a particular wavelength range. It is the mechanoreceptors\' position in the cochlea that determines which sound frequency will maximally stimulate them \[[@pbio-0030137-b10],[@pbio-0030137-b12]\]. This structure is maintained on a higher processing level: fibers in the auditory cranial nerve send the information from the receptor cells to the brain in parallel, i.e., in a first approximation, each receptor cell has its own neuronal "cable" to the cochlear nuclei, and beyond, to the auditory cortex, where we find a complete topographical map of the audible frequency spectrum ([Figure 1B](#pbio-0030137-g001){ref-type="fig"}), mirroring the mapping of frequencies in the cochlea \[[@pbio-0030137-b11],[@pbio-0030137-b13],[@pbio-0030137-b14]\]. The perception of pitch, then, is arranged along one dimension, as on a piano keyboard. Because of the parallel processing of receptor information from the cochlea, we can hear mixtures of different frequencies, and analyze their components accurately, unless mixtures are very complex. We can identify the tones that a chord is made up of \[[@pbio-0030137-b15]\]. There are, of course, cases where mixtures have unique properties, e.g., in the case of the intensities of harmonic overtones (integer multiples of the fundamental frequency) that distinguish a C tone produced by a guitar from that of a flute---but we can still identify the fundamental, whose perceived pitch is not altered by the overtones. We would never perceive a mixture of 400 Hz and 800 Hz as an intermediate frequency (e.g., 600 Hz).
::: {#pbio-0030137-g001 .fig}
Figure 1
::: {.caption}
###### Frequency Coding in the Human Ear and Cortex
\(A) The human ear and frequency mapping in the cochlea. The three ossicles incus, malleus, and stapes transmit airborne vibration from the tympanic membrane to the oval window at the base of the cochlea. Because of the mechanical properties of the basilar membrane within the snail-shaped cochlea, high frequencies will produce a vibration peak near the oval window, whereas low frequencies will stimulate receptors near the apex of the cochlea (locations for three frequencies indicated schematically). Information from the cochlear receptor cells is transmitted to the cochlear nuclei via the 8th cranial nerve, and on through the midbrain to the cortex. (Redrawn from Figure 12.3 in \[[@pbio-0030137-b11]\].)
\(B) Lateral view of the human brain, with the auditory cortex exposed. The primary auditory cortex contains a topographic map of the cochlear frequency spectrum (shown in kilohertz). (Redrawn from Figure 12.15A in \[[@pbio-0030137-b11]\].)
:::
:::
Color Perception in Bees {#s3}
========================
While these observations are seemingly trivial in the world of sound, the perception of mixtures in color vision is fundamentally different. If we mix yellow with red light, we will see orange---not only will we not be able to tell that the orange light has been produced by mixing two lights, but we will also be unable to distinguish the mixture from monochromatic orange light \[[@pbio-0030137-b16]\]. Similar mixture phenomena are well established in bees; for example, orange light (with a wavelength of 590 nm) can be mixed with blue light (440 nm) to produce a mixture that is indistinguishable from monochromatic bluegreen light (490 nm) \[[@pbio-0030137-b17],[@pbio-0030137-b18]\]. In bees, as in humans, the perception of hues is arranged in a circular fashion around achromatic white or gray \[[@pbio-0030137-b19]\]. In this circular arrangement, complementary colors (i.e., those opposite on a circle) can be additively mixed to generate a neutral stimulus, e.g., UV light (350 nm) mixed in the appropriate ratio with bluegreen light (490 nm) will be perceived as white by bees \[[@pbio-0030137-b17]\]. In the auditory system, mixing only two frequencies to produce the perception of white noise is inconceivable! Finally, the circular arrangement of hues also implies that it is possible to mix two ends of the perceptible spectrum to produce a sensation that is not contained in the spectrum: for humans, mixing violet with red light will produce purple, a sensation that has no equivalent in the visual spectrum; a similar unique percept can be produced by facing bees with a mixture of UV and green light \[[@pbio-0030137-b5],[@pbio-0030137-b17]\].
We (and other animals) can only ever see one color in a point; additive color mixtures are perceived as intermediates of their generative sources, and it is not possible to identify the physical components of a mixture (e.g., the perception of white can be generated mixing any two complementary colors). In contrast, we can hear several frequencies at once, and identify the components of at least simple stimulus mixtures such as triads \[[@pbio-0030137-b15]\]. In color vision, but not in hearing, multiple combinations of physical stimuli will generate identical sensations. The reasons for these fundamental differences between the visual and the auditory sensory modalities can be attributed in a straightforward way to the structure of the sense organs as well as post-receptor neural circuits.
In color vision, there aren\'t thousands of receptor cells each responsible for receiving a narrow range of wavelengths (as there are in the auditory system, with its technique of frequency analysis). Instead, both humans and bees have only three color receptor types, each sensitive to a broad range of wavelengths ([Figure 2A](#pbio-0030137-g002){ref-type="fig"}). Human color receptors are typically termed blue, green, and red receptors, while those in bees are most sensitive to UV, blue, and green light \[[@pbio-0030137-b5],[@pbio-0030137-b20]\]. A single receptor cannot analyze the wavelength of the light it receives: it simply acts as a quantum counter, but the information about the wavelength identity of the quantum is lost on absorption. Any single photoreceptor might respond equally to a medium intensity light in the wavelength range of its peak sensitivity, and to a strong intensity light at the periphery of its sensitive range---hence it cannot disentangle wavelength from intensity \[[@pbio-0030137-b5]\].
::: {#pbio-0030137-g002 .fig}
Figure 2
::: {.caption}
###### Neuronal Color Coding and Color Space in Bees
\(A) Frontal view of bee head (scanning electron micrograph) showing essential features of color processing in the brain. Information from the UV, blue, and green receptors is relayed from the first optic ganglion, the lamina, to the second optic ganglion, the medulla, by so-called monopolar cells (LMCs); cell bodies are symbolized by filled circles. These cells feed into color opponent cells (drawn in red and black) found both in the medulla and lobula, either directly or via interneurons. Chromatic opponent cells receive antagonistic input from the different color channels, and project to the protocerebrum. (Image based on \[[@pbio-0030137-b5],[@pbio-0030137-b18]\].)
\(B) Color opponent space for bees, where axes correspond to excitation values of two types of color opponent neurons. Corners correspond to maximum excitation of the UV (lower left), blue (top), and green (lower right) receptors. Color loci of some representative monochromatic lights are shown. Angular position in this space (as measured from the center) corresponds to hue, whereas distance between color loci corresponds to perceived similarity.
:::
:::
This means that the visual system has to compare the signals from receptors differing in spectral sensitivity. In insects as well as in vertebrates, it does this by means of color opponent neurons ([Figure 2A](#pbio-0030137-g002){ref-type="fig"}) \[[@pbio-0030137-b18],[@pbio-0030137-b21],[@pbio-0030137-b22]\]. The minimum equipment for an animal with two color receptor types is one type of opponent neuron, receiving antagonistic inputs from the two types of cells. Using such an opponent mechanism, the visual system can "tell" whether there is a stronger signal from the short wavelength receptor or the long wavelength receptor---hence it can extract information about stimulus spectral quality. Theoretically, an animal with n color receptor types needs *n* − 1 chromatic opponent mechanisms \[[@pbio-0030137-b23]\]. Indeed, behavioral experiments with bees have shown that only two color opponent mechanisms are necessary to explain color discrimination data \[[@pbio-0030137-b18],[@pbio-0030137-b24]\]. It is not clear which ones these are---physiologists have found at least seven different types of color opponent neurons in the bee optic lobes \[[@pbio-0030137-b5],[@pbio-0030137-b18],[@pbio-0030137-b21]\], and modeling has shown that almost any combination of two color opponent neuron types is adequate for color coding \[[@pbio-0030137-b24]\].
Nevertheless, the two-dimensional color opponent space whose axes correspond to excitation patterns of color opponent mechanisms ([Figure 2B](#pbio-0030137-g002){ref-type="fig"}) has proven extremely useful: distances in such a color space correlate well with behavioral color discrimination data. The color opponent space allows us to predict hue (by assessing angular position) and saturation (by measuring distance from the center), and it can be used to predict the perception of color mixtures, which fall between the loci of the colors used to generate the mixture \[[@pbio-0030137-b5]\]. Because information about the actual receptor signals is discarded in the very periphery of the sensory system, perception is only based on derived (color opponency) dimensions, which measure differences in receptor signals rather than absolute signals.
Olfactory Perceptual Space in Bees {#s4}
==================================
Making sense of scents is a considerably messier affair. Odors are hardly presentable on a physical continuum (like the wavelength of light); they are multidimensional entities that can vary from small gaseous molecules to long-chained hydrocarbons \[[@pbio-0030137-b25],[@pbio-0030137-b26]\]. Organic compounds vary in carbon chain length and functional group, i.e., the group of atoms that give substances their characteristic properties (e.g., alcohols, aldehydes, ketones, or alkanes). At any moment, the air around an animal may contain hundreds of different airborne substances, which fluctuate with wind, humidity, and multiple other factors \[[@pbio-0030137-b25]\].
On the receptor level, the olfactory system shows a complexity that is unparalleled in any color vision system: the receptor family discovered by 2004\'s Nobel laureates L. Buck and R. Axel comprises about 1,000 receptor proteins in mammals, each of which only binds a narrow range of airborne molecules \[[@pbio-0030137-b27],[@pbio-0030137-b28]\]. Only one of these proteins is expressed per receptor cell, so that there are indeed about 1,000 different odor receptor cell types in the mammalian olfactory epithelium \[[@pbio-0030137-b29]\]. In insects, the diversity of such receptors is lower, but still impressive: fruit flies appear to have 65 different odorant receptor genes \[[@pbio-0030137-b30]\]. In honeybees, the number of such genes will be accessible pending the publication of the honeybee genome; so far, screens by H. Robertson (pers. comm.) indicate a number \>130. Some 60,000 olfactory receptor cells (of a few dozen different types) are distributed along the honeybee antennae (\[[@pbio-0030137-b31]\]; [Figure 3A](#pbio-0030137-g003){ref-type="fig"}). How can the brain extract biologically useful information from such multidimensional sensory input?
::: {#pbio-0030137-g003 .fig}
Figure 3
::: {.caption}
###### Neuronal Odor Coding and Odor Space in Bees
\(A) Schematic view of odor processing in the honeybee brain. Some 60,000 odorant receptor cells are distributed along the antenna. These belong to several different types (illustrated with different colors), each responsive to a different set of chemicals. Axons from like receptors project to one or a few glomeruli in the antennal lobe. The glomerular map is organized so that similar odors are mapped to nearby spatial locations (yellow and red), while dissimilar odors stimulate glomeruli located further apart (blue). Axonal projections extend from the antennal lobe to higher processing centers, such as the calyces (CAL) of the mushroom bodies (MB). Some such projections might relay relatively unprocessed sensory information to the mushroom bodies (yellow, red, and blue), while others contain processed information based on lateral interactions between glomeruli (orange, between the yellow and red projections).
\(B) Putative three-dimensional odor space for bees. Guerrieri et al. \[[@pbio-0030137-b8]\] trained bees to associate one of 16 odors with a sucrose reward, and then faced bees with the other 15 odors, to see how similarly bees judged these to the training odor. Distances between these substances in a three-dimensional space predict the bee-subjective similarity of the odors. The most important axis corresponds to the carbon chain length of the substances tested; the other two dimensions separate substances according to functional group. Each word illustrates the spatial distribution of a group of substances with like functional group, but varying in chain length. (Image based on Figure 6 in \[[@pbio-0030137-b8]\].)
:::
:::
The first neuronal center of olfactory information processing, the antennal lobe (or its mammalian analogue, the olfactory bulb) achieves order in two ways. First, axons from like receptor cells (i.e., those that express the same receptor protein and therefore bind the same odorants) project to one or a few glomeruli, i.e., globular, anatomically distinct subunits within the antennal lobe \[[@pbio-0030137-b26]\]. The number of glomeruli ranges from a few dozen to several hundred, and corresponds roughly to the number of putative olfactory receptor types \[[@pbio-0030137-b25]\]. The honeybee\'s antennal lobe contains 160 glomeruli \[[@pbio-0030137-b31]\]. Individual chemicals reliably activate sets of identified glomeruli \[[@pbio-0030137-b32]\]. These micro-relays sum up the input from same chemoreceptors, tremendously increasing the signal-to-noise ratio, and thus facilitating reliable odorant detection \[[@pbio-0030137-b26]\].
A second, and perhaps more remarkable, feature of the antennal lobe is that glomeruli coding for similar substances are located close together, while those that code for distinct scents are spatially segregated \[[@pbio-0030137-b29],[@pbio-0030137-b32],[@pbio-0030137-b33],[@pbio-0030137-b34]\]. Carbon chain length, for example, is neatly represented in this glomerular map \[[@pbio-0030137-b32]\]. How the brain achieves such "chemical mapping" is something of a miracle: how would the developing brain "know" which axons belong to receptors that respond to chemically similar substances, so that these can be wired to neighboring glomeruli? In mammals, it was recently found that the receptor proteins that bind odor molecules are also expressed in the axon terminals of the receptor cells \[[@pbio-0030137-b29]\]. If we assume that similar receptor molecules bind similar odorants, then the developing nervous system could use receptor molecule similarity in the receptor cells\' axon terminals to wire up the neural map in the antennal lobe.
But does this neuronal activity map indeed correspond to the olfactory perceptual space? Theoretically, a perceptual space might have as many dimensions as there are distinct receptor types---or it might have as many axes as there are glomeruli with distinct response profiles. Is it possible that olfactory space in bees, then, has several dozen dimensions? To evaluate the structure of olfactory perceptual space, Guerrieri et al. \[[@pbio-0030137-b8]\] trained honeybees to memorize a wide variety of odors, and then tested how well bees could distinguish these odors from others. They then asked: how many axes must the olfactory perceptual space have, so that distances between odors can be used to predict how similar they will appear to bees? They found that the multidimensional receptor space might be collapsed onto very few perceptual axes: much of the similarity judgments between odors can be explained by a three dimensional space ([Figure 3B](#pbio-0030137-g003){ref-type="fig"}). The most important axis spreads out scents according to carbon chain length, whereas the other two axes separate the odors according to functional group, i.e., they separate primary and secondary alcohols, aldehydes, and ketones. Distances between odor loci in this three-dimensional space correlate well with the discriminability of the odors to bees. They also correspond to the similarity of activation patterns of the glomerular map \[[@pbio-0030137-b32]\], although the actual mechanisms that evaluate this similarity remain to be identified. There are a number of complications, however, that indicate that the olfactory perceptual space is unlikely to submit to the relatively simple rules of perception of color and pitch.
Take mixtures of different stimuli, for example. If odor perception followed rules similar to those of color perception, and if it relied exclusively on derived parameters (such as carbon chain length), then mixtures would be perceived as intermediates of their components. A mixture of two molecules differing only in carbon chain length would be perceived as indistinguishable from a single odorant with an intermediate carbon chain length. This remains to be tested, but we conjecture that this is unlikely to be the case. On the other hand, are odor mixtures simply perceived as a compound entity with distinct components (like a triad in vertebrate pitch perception), or are mixtures unique entities that are perceived as fundamentally different from their elements (like "white" in color perception)? Honeybees appear to employ a combination of the two: they can perceive the components of a mixture (and generalize to these components when faced with them individually), but also attach unique properties to the mixture \[[@pbio-0030137-b31],[@pbio-0030137-b35]\]. A further complication is that Guerrieri et al. \[[@pbio-0030137-b8]\] found intriguing asymmetries in the bees\' assessments of odor similarities: bees respond as if they find odor A more similar to trained odor B than they find B to trained odor A. It will be difficult to represent these complex phenomena in a simple map.
These phenomena indicate that odor perception cannot be as easily visualized in a low-dimensionality space as other sensory modalities. However, for psychophysicists, just as for motorists navigating novel territory, even a rough guidance map is better than no map, and so this olfactory space is undoubtedly a useful tool to predict the bee-subjective similarity of scents. While the antennal lobe clearly structures the sensory input so as to extract derived chemical properties, the information about the input from individual receptor types might not be discounted in the periphery of the nervous system, as in visual perception, but relayed on to higher processing centers, such as the mushroom bodies and the protocerebrum \[[@pbio-0030137-b36],[@pbio-0030137-b37]\]. If both modulated and unmodulated sensory information is available to the neuronal centers that ultimately "decide" on odor similarity, and especially if different individuals attend differently to different kinds of input---e.g., because experience modifies interactions between glomeruli \[[@pbio-0030137-b31]\]---then the derivation of a static perception space that can be applied to all individuals of a species may be quite challenging.
Conclusion {#s5}
==========
Philosophers have correctly pointed out that we cannot actually imagine what it is like to perceive the world through other animals\' sense organs \[[@pbio-0030137-b38]\]. But studies such as those by Guerrieri et al. \[[@pbio-0030137-b8]\] show elegantly how to attach scales and numbers to the inner worlds proposed by von Uexküll a century ago. Mapping these worlds quantitatively will allow us to compare them between related species operating in different environments, to see how the architecture of perceptual spaces is adapted to mirror biologically useful information from the real world in each species \[[@pbio-0030137-b39]\]. One of the most exciting future directions is to explore the extent to which such perceptual worlds are not just species-specific, but in fact individual-specific. In the honeybee, learning alters the response patterns of the glomerular map \[[@pbio-0030137-b40]\]. If the antennal lobe has a dual function in creating the primary dimensions of olfactory perception and storing olfactory memories, then one prediction is that individual experience will alter perception. Similar phenomena have been predicted in other sensory modalities \[[@pbio-0030137-b19],[@pbio-0030137-b41]\], but remain to be shown directly by exposing different individuals to different environments during development.
**Citation:** Chittka L, Brockmann A (2005) Perception space---The final frontier. PLoS Biol 3(4): e137.
Lars Chittka is at the School of Biological Sciences, Queen Mary College, University of London, United Kingdom. Axel Brockmann is in the Department of Entomology, University of Illinois, Urbana-Champaign, Illinois, United States of America.
|
PubMed Central
|
2024-06-05T03:55:55.119257
|
2005-4-12
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1074815/",
"journal": "PLoS Biol. 2005 Apr 12; 3(4):e137",
"authors": [
{
"first": "Lars",
"last": "Chittka"
},
{
"first": "Axel",
"last": "Brockmann"
}
]
}
|
PMC1079781
|
Introduction {#s1}
============
When compared against an adult standard, infant sleep is composed of a reduced set of components. Perhaps the most notable difference between the sleep of infants and adults is the absence of clearly differentiated and sleep-related neocortical electroencephalographic (EEG) activity \[[@pbio-0030143-b01],[@pbio-0030143-b02],[@pbio-0030143-b03]\]. Nonetheless, sleep and wakefulness in the infants of altricial species, including rats, are reliably characterized, respectively, by periods of myoclonic twitching expressed against a background of muscle atonia and high-amplitude behaviors (e.g., locomotion or stretching) expressed against a background of high muscle tone \[[@pbio-0030143-b04]\]. Although myoclonic twitching during active sleep (AS) in infants is more prevalent and more intense than that seen during rapid eye movement (REM) sleep in adults, its similarities to the adult behavior and its linkage to periods of atonia suggest developmental continuity between the infant and adult sleep states \[[@pbio-0030143-b04],[@pbio-0030143-b05],[@pbio-0030143-b06]\]. Opinions differ, however, as to whether these similarities provide sufficient evidence that infant sleep is developmentally continuous with that of adults. Specifically, it has recently been argued that sleep during the "pre-EEG" period---that is, before postnatal day 12 (P12) in rats \[[@pbio-0030143-b02]\]---is best considered a primitive state, comprising an amalgam of sleep components, that only gradually differentiates into the distinct sleep states characteristic of adults \[[@pbio-0030143-b01],[@pbio-0030143-b07]\].
Proponents of the "amalgam" theory of sleep development have made strong claims concerning the neural substrates of infant sleep. For example, it has been argued that infant sleep is (i) undifferentiated and (ii) controlled by distinct neurophysiological mechanisms from those that generate sleep in adults. In particular, Adrien \[[@pbio-0030143-b08]\] writes, "In the immediate postnatal period, and furthermore in utero, AS would seem to be very different---even in terms of underlying mechanisms---from what we know of the sleep phenomenon in the adult" (p. 40). Extending this idea, Adrien and Lanfumey \[[@pbio-0030143-b09]\] state that neural activity during infant sleep indicates "a very primitive system of diffuse activation within the whole central nervous system" (p. 7). More recently, Frank and Heller \[[@pbio-0030143-b01]\] write, "Brainstem-midbrain nuclei important in mediating REM sleep expression do not mediate the expression of AS, or AS myoclonia" (p. 64). Although several papers have reported state-dependent neural activity within the brainstem of infant rats \[[@pbio-0030143-b10],[@pbio-0030143-b11]\], they have not proved convincing \[[@pbio-0030143-b07]\].
Despite the absence of well-differentiated EEG activity in rats before P12, it is now apparent that the state characterized by nuchal atonia---with or without the simultaneous occurrence of myoclonic twitching---satisfies many established criteria of sleep \[[@pbio-0030143-b12],[@pbio-0030143-b13],[@pbio-0030143-b14]\]. For example, during infant sleep, sensory thresholds increase, both under normal conditions \[[@pbio-0030143-b15]\] and during sleep deprivation \[[@pbio-0030143-b16]\]; infant sleep is also homeostatically regulated \[[@pbio-0030143-b16]\]. In addition, it was recently shown that there exists an inhibitory area within the ventromedial medulla, the medullary inhibitory area (MIA), that appears functionally identical to the corresponding area that mediates REM-sleep atonia in adults \[[@pbio-0030143-b17]\]. The MIA (i) causes atonia when stimulated; (ii) contains neurons that exhibit atonia-related discharge profiles; and (iii) when lesioned, results in the partial loss of atonia and the decoupling of the components of infant sleep (i.e., "REM sleep without atonia"). The MIA is not sufficient, however, for generating infant sleep, because decerebrations rostral to the MIA, but caudal to the mesopontine region, abolish sleep-wake cyclicity and reduce myoclonic twitching \[[@pbio-0030143-b18]\] (Karlsson and Blumberg, unpublished data). Thus, expression of the components of infant sleep---atonia and twitching---depends on a network that spans the mesopontine-medullary region.
The aims of the current study were to identify the structures that project to the MIA (and therefore may contribute to the generation of atonia) and to identify supraspinal neural correlates of myoclonic twitching. First, we performed retrograde tracing from the MIA as well as the ventrolateral medulla (an area that typically induces motor activity when stimulated \[[@pbio-0030143-b17]\]). Next, using the tracing data as a guide, we conducted neural recording studies to determine the relationship, if any, between the discharge profiles of individual neurons and the specific components of infant sleep. Finally, we lesioned structures containing neurons that exhibit opposing state-dependent discharge profiles and examined the effects on behavioral states. These experiments provide the most comprehensive survey to date of the neural control of sleep in infancy and reveal surprising parallels between the neural substrates of sleep in infant and adult rats. Moreover, these results are not consistent with the perspective that views infant sleep as "diffuse," "undifferentiated," or arising from "distinct neurophysiologic mechanisms" \[[@pbio-0030143-b01],[@pbio-0030143-b07],[@pbio-0030143-b08],[@pbio-0030143-b09]\]. Finally, the current findings provide a guide to future theorizing about the ontogeny and function of sleep.
Results {#s2}
=======
Mesopontine and Medullary Structures Project to the Medullary Inhibitory Area {#s2a}
-----------------------------------------------------------------------------
In order to reveal the afferent structures to the MIA that may be important for the expression of muscle atonia, the retrograde tracer DiI was infused into the medial medulla in 5 P8 rats ([Figure 1](#pbio-0030143-g001){ref-type="fig"}). Medial DiI infusions resulted in light but widespread labeling throughout the pontomedullary axis. Within the medulla, the giant cells of nucleus gigantocellularis (Gi) were consistently labeled ([Figure 2](#pbio-0030143-g002){ref-type="fig"}A), as were neurons in nucleus magnocellularis and nucleus paramedianis. The vestibular nuclei and cholinergic C3 were lightly labeled and the nucleus of the solitary tract was moderately labeled. Light to moderate labeling was also found along the longitudinal axis from Gi through the nucleus pontis caudalis (PC) to the nucleus pontis oralis (PO). Within the pons, clusters of labeled neurons were consistently found ventral to the locus coeruleus (LC) in the subcoeruleus (alpha, dorsal, and ventral parts, hereafter designated SubLC; [Figure 2](#pbio-0030143-g002){ref-type="fig"}B). Neurons associated with the midline raphé system (i.e., raphé obscurus and medial and dorsal raphé, DR) were heavily labeled ([Figure 2](#pbio-0030143-g002){ref-type="fig"}C). Finally, neurons of the dorsal pontine tegmentum, including the laterodorsal tegmental nucleus (LDT; [Figure 2](#pbio-0030143-g002){ref-type="fig"}D), the pendunculopontine tegmental nucleus (PPT), and the periaqueductal gray were lightly, but consistently, labeled.
::: {#pbio-0030143-g001 .fig}
Figure 1
::: {.caption}
###### DiI Infusion Sites
\(A) Halos from 20 nl medial (*n* = 5) and lateral (*n* = 5) DiI infusions recreated from fluorescent photomicrographs on a coronal section of the medullary inhibitory area of a P8 rat. The largest (black) and smallest (gray) halos are shown.
\(B) Light photomicrograph of a representative DiI infusion site in the medial medulla. 4V, fourth ventricle; H, hypoglossal nucleus; Gi, nucleus gigantocellularis; IO, inferior olive; ST, spinal trigeminal nucleus
:::
:::
::: {#pbio-0030143-g002 .fig}
Figure 2
::: {.caption}
###### Representative Labeling of Cell Bodies in Selected Areas after a DiI Infusion into the Medullary Inhibitory Area
The arrows indicate examples of labeled neurons in (A) nucleus gigantocellularis (Gi), (B) subcoeruleus (SubLC), (C) dorsal raphé (DR), and (D) laterodorsal tegmental nucleus (LDT). Inset in (B) depicts enlarged view of the boxed area. 4V, fourth ventricle; AQ, cerebral aqueduct; DT, dorsal tegmental nucleus; Me5, mesencephalic trigeminal nucleus; ST, spinal trigeminal nucleus
:::
:::
Because our previous study \[[@pbio-0030143-b17]\] had indicated that motor facilitatory neurons were typically found lateral to the MIA, DiI was also infused laterally to the MIA in five additional P8 rats. Overall, few differences in labeling were observed between the medially and laterally infused groups, with the exception that neurons associated with the raphé system were labeled only after medial infusions.
After visual inspection of the tracing data, seven medullary and pontine structures known to modulate behavioral state regulation in adults (the Gi, PC, SubLC, LC, PO, LDT, and DR) were selected for analysis. The cell counts for these seven structures are summarized in [Table 1](#pbio-0030143-t001){ref-type="table"}.
::: {#pbio-0030143-t001 .table-wrap}
Table 1
::: {.caption}
###### Mean Number (± Standard Error) of DiI-Labeled Neurons within Seven Medullary and Mesopontine Structures in P8 Rats
:::
DiI was infused into the medial or lateral medulla. See [Figure 1](#pbio-0030143-g001){ref-type="fig"} for location and extent of infusion sites
Gi, nucleus gigantocellularis; DR, dorsal raphé; LC, locus coeruleus; LDT, laterodorsal tegmental nucleus; PC, nucleus pontis caudalis; PO, nucleus pontis oralis; SubLC, subcoeruleus
:::
Extracellular Recordings Reveal State-Dependent Neuronal Activity in Medullary and Mesopontine Structures {#s2b}
---------------------------------------------------------------------------------------------------------
A total of 142 neurons were recorded from 25 P6--P10 rats (1--6 neurons per rat; 1--4 neurons per recording site) with precollicular decerebrations. (Although precollicular decerebrations disconnect the brainstem from rostral structures known to regulate sleep and wakefulness, infants nonetheless exhibit coherent organization of sleep-wake states; the most noticeable change induced by these decerebrations is more rapid cycling between sleep and wakefulness \[[@pbio-0030143-b19]\].) Fifty-two neurons did not exhibit any state dependency and were not analyzed further. Five neurons were excluded due to unclear histology. The remaining 85 neurons were assigned to one of five classes on the basis of the specific relationship of the unit activity to the components of sleep and wakefulness ([Figure 3](#pbio-0030143-g003){ref-type="fig"}A). The interrelations among the classes, and the number of neurons found within each class, are depicted in [Figure 3](#pbio-0030143-g003){ref-type="fig"}B.
::: {#pbio-0030143-g003 .fig}
Figure 3
::: {.caption}
###### Schematic Representation of the Classes of Neurons Identified in the Current Study and Their Prevalence
\(A) Summary of the classes of discharge profiles described in the present study in relation to nuchal EMG activity.
\(B) Venn diagram depicting the interrelation of the classes of state-dependent neurons. Values indicate the number of neurons in each class found in the present study.
:::
:::
The Lateral Medulla Contains Both EMG-On and Atonia-On Neurons {#s2c}
--------------------------------------------------------------
A total of 13 state-dependent neurons were recorded in the medulla, all within the Gi ([Figure 4](#pbio-0030143-g004){ref-type="fig"}A). Eight neurons exhibited significantly lower discharge rates during periods of atonia (indicative of sleep) than during periods of high muscle tone (indicative of wakefulness; discharge rates, 1.74 ± 0.38 Hz during atonia; 3.79 ± 0.60 Hz during high tone), and five neurons exhibited significantly higher discharge rates during atonia than high tone (discharge rates, 1.21 ± 0.40 Hz during atonia; 0.76 ± 0.29 Hz during high tone). Representative single unit activity and concurrently recorded electromyograms (EMGs) are depicted in [Figures 4](#pbio-0030143-g004){ref-type="fig"}C and [4](#pbio-0030143-g004){ref-type="fig"}D. Average spike durations of atonia-on and EMG-on neurons were 1.87 ± 0.01 ms and 1.74 ± 0.07 ms, respectively (a single-factor ANOVA revealed no significant group differences in spike durations between any of the classes of neurons reported in the current study).
::: {#pbio-0030143-g004 .fig}
Figure 4
::: {.caption}
###### State-Dependent Neural Activity within Nucleus Gigantocellularis
\(A) Recording sites of state-dependent neurons reconstructed on a coronal section of the medulla of a P8 rat. Note the anatomical overlap of the different classes of neurons. Arrow corresponds to top arrow in (B).
\(B) Photomicrograph depicting two marking lesions in nucleus gigantocellularis (the two lesions are approximately 80 μm apart). The white arrow indicates the location of the recording site identified by the arrow in (A).
\(C) Lower trace: single unit activity of a representative Gi EMG-on neuron. Upper trace: concurrently recorded nuchal EMG. Far right: Averaged waveform of representative EMG-on neuron.
\(D) Lower trace: single unit activity of a representative Gi atonia-on neuron. Upper trace: concurrently recorded nuchal EMG. Far right: Averaged waveform of representative atonia-on neuron. Gi, nucleus gigantocellularis; Pyr, pyramids; ST, spinal trigeminal nucleus; V, vestibular nucleus
:::
:::
The Subcoeruleus Contains a High Concentration of Atonia-On Neurons {#s2d}
-------------------------------------------------------------------
Of a total of 38 neurons classified as atonia-on, 17 were located within the SubLC (discharge rates, 2.64 ± 0.48 Hz during atonia; 0.75 ± 0.23 Hz during high tone); EMG-on neurons were not found within the SubLC. Within the brainstem, atonia-on neurons were also recorded in the dorsomedial tegmental nucleus, ventral tegmental nucleus, and the medial and lateral parabrachial nuclei (MPB and LPB, respectively; see [Figure 5](#pbio-0030143-g005){ref-type="fig"}A). Spike durations ranged from 1.46 ± 0.75 ms for atonia-on neurons in the LPB to 1.87 ± 0.03 ms for atonia-on neurons in the SubLC. The averaged spike waveform of a representative atonia-on neuron recorded within the SubLC is depicted in [Figure 5](#pbio-0030143-g005){ref-type="fig"}B. Typically, neurons that were classified as atonia-on discharged tonically throughout the atonia period ([Figure 5](#pbio-0030143-g005){ref-type="fig"}C).
::: {#pbio-0030143-g005 .fig}
Figure 5
::: {.caption}
###### State-Dependent Neural Activity within the Mesopontine Region
\(A) Recording sites of state-dependent neurons reconstructed on a coronal section at the mesopontine level of a P8 rat. Note the predominance of atonia-on neurons.
\(B) Averaged waveform of a representative atonia-on neuron.
\(C) Upper trace: multiunit activity. Lower trace: concurrently recorded nuchal EMG. Spike sorting revealed two units that are easily distinguished by their amplitudes. The higher-amplitude unit is atonia-on; note its tonic discharge throughout the atonia period.
\(D) Upper trace: multiunit activity. Lower trace: concurrently recorded nuchal EMG. Spike sorting revealed two units that are easily distinguished by their amplitudes. The higher-amplitude unit is AS-on; note the absence of multiunit activity at the onset of the atonia period and then the increase in activity coinciding with the appearance of nuchal twitches.
\(E) Mean discharge rates of a representative AS-on neuron during bouts of AS and QS as defined, respectively, by the presence or absence of phasic nuchal twitches during periods of atonia. The arrowhead indicates the midpoint of the AS and QS bouts. 4V, fourth ventricle; LC, locus coeruleus; PC, nucleus pontis caudalis; SubLC, subcoeruleus;
:::
:::
A Subset of Atonia-On Neurons Discharge at Higher Rates during Active Sleep {#s2e}
---------------------------------------------------------------------------
In contrast to the neuron depicted in [Figure 5](#pbio-0030143-g005){ref-type="fig"}C, nine of the atonia-on neurons exhibited significantly higher discharge rates during AS, defined as periods of atonia accompanied by bouts of twitches of the tail and nuchal muscle. These neurons were classified as AS-on (see [Materials and Methods](#s4){ref-type="sec"}). A representative recording of multiunit activity that includes an AS-on neuron is depicted in [Figure 5](#pbio-0030143-g005){ref-type="fig"}D. In [Figure 5](#pbio-0030143-g005){ref-type="fig"}E the average discharge rate of a representative AS-on neuron during AS and QS (defined as periods of atonia without myoclonic twitching) is depicted.
Four neurons recorded within the SubLC were AS-on (discharge rates, 1.84 ± 0.67 Hz during AS; 0.73 ± 0.28 Hz during QS); three medullary neurons were AS-on (discharge rates, 1.99 ± 0.60 Hz during AS; 0.91 ± 0.30 during QS); and two neurons within the MPB were AS-on (discharge rates, 0.43 ± 0.02 Hz during AS; 0.09 ± 0.02 Hz during QS). Spike widths of AS-on neurons ranged from 1.95 ± 0.10 ms in the SubLC to 2.03 ± 0.16 ms in Gi.
The Infant Locus Coeruleus Contains EMG-On Neurons {#s2f}
--------------------------------------------------
Neurons exhibiting EMG-on profiles were recorded in the LPB, PC, and the LC. Within the LC ([Figure 6](#pbio-0030143-g006){ref-type="fig"}A), two such units were found. The averaged spike waveform for one of these units is depicted in [Figure 6](#pbio-0030143-g006){ref-type="fig"}B (both neurons exhibited short spike durations, 1.76 ± 0.12 ms). As shown in [Figure 6](#pbio-0030143-g006){ref-type="fig"}C, both LC units increased their discharge rates primarily during periods of high muscle tone and were virtually silent during periods of atonia (discharge rates, 0.38 ± 0.17 during atonia; 7.81 ± 0.06 Hz during high tone), apart from rare bursts of activity during atonia (e.g., as indicated by the arrow in [Figure 6](#pbio-0030143-g006){ref-type="fig"}C).
::: {#pbio-0030143-g006 .fig}
Figure 6
::: {.caption}
###### Activity of LC Neurons across the Sleep-Wake Cycle in a P8 Rat
\(A) Coronal section of the LC indicating the recording site (double arrows).
\(B) Averaged waveform of unit 2 in (C).
\(C) Upper trace: LC multiunit activity. Middle traces: Activity of two isolated units derived from the multiunit activity. Bottom trace: concurrently recorded nuchal EMG. 4V, fourth ventricle; LC, locus coeruleus
:::
:::
The Dorsolateral Pontine Tegmentum Contains EMG-On Neurons {#s2g}
----------------------------------------------------------
In [Figure 7](#pbio-0030143-g007){ref-type="fig"}A, the recording sites within the pontine tegmentum are depicted. Seventeen of 37 EMG-on neurons were located within a small region of the dorsolateral pontine tegmentum that includes LDT, the cuneiform nucleus, and periaqueductal gray; the remainder were distributed within the LC (see [Figure 6](#pbio-0030143-g006){ref-type="fig"}), PC (see [Figure 5](#pbio-0030143-g005){ref-type="fig"}), and Gi (see [Figure 4](#pbio-0030143-g004){ref-type="fig"}). An averaged spike waveform of a representative EMG-on neuron, recorded in LDT, is depicted in [Figure 7](#pbio-0030143-g007){ref-type="fig"}B (mean spike durations of EMG-on neurons ranged from 1.62 ± 0.11 ms in the LDT to 2.36 ± 0.14 ms in the PC) and a representative recording of multiunit activity that includes an EMG-on neuron is depicted in [Figure 7](#pbio-0030143-g007){ref-type="fig"}C. The discharge rates of non-LDT EMG-on neurons ranged from 0.66 ± 0.40 Hz during atonia in the PC to 4.75 ± 1.09 Hz during high tone in the cuneiform nucleus. The relatively high discharge rates (5.67 ± 1.49 Hz during atonia and 11.10 ± 2.22 Hz during high tone) and short spike durations of LDT EMG-on neurons suggest that they may be non-cholinergic \[[@pbio-0030143-b20]\].
::: {#pbio-0030143-g007 .fig}
Figure 7
::: {.caption}
###### State-Dependent Neuronal Discharges within the Pontine Tegmentum
\(A) Recording sites of state-dependent neurons reconstructed on a coronal section of the brainstem. Note the predominance of EMG-on neurons.
\(B) Averaged waveform of a representative EMG-on neuron.
\(C) Upper trace: multiunit activity. Lower trace: concurrently recorded nuchal EMG. One EMG-on neuron was isolated from the multiunit record; note its tonic discharge during the period of high muscle tone.
\(D) Upper trace: multiunit activity. Lower trace: concurrently recorded nuchal EMG.
\(E) Expanded view of the boxed area from (D). Note how multiunit activity precedes the twitch. Asterisks identify a single isolated unit.
\(F) Peristimulus histogram and raster plot for the twitch-on neuron identified in (E) during a 10-min recording session in a P7 rat (83 total twitches). Inset depicts 55 superimposed action potential waveforms for this unit. This unit\'s mean discharge rate peaks 5--10 ms before the twitch (red line).
\(G) Averaged nuchal EMG for all 83 twitches represented in (F). AQ, cerebral aqueduct; DT, dorsal tegmental nucleus; LDT, laterodorsal tegmental nucleus; PO, nucleus pontis oralis
:::
:::
The Laterodorsal Tegmental Nucleus Contains Twitch-On Neurons {#s2h}
-------------------------------------------------------------
The LDT contains neurons that exhibit a burst of action potentials in anticipation of myoclonic twitches (designated as twitch-on neurons). Ten of the 11 recorded neurons that exhibited this activity pattern were found within the LDT ([Figure 7](#pbio-0030143-g007){ref-type="fig"}A); one neuron was located in the parabrachial nucleus (see [Figure 5](#pbio-0030143-g005){ref-type="fig"}A). Six of these neurons were both EMG- and twitch-on (designated as EMG-on/twitch-on); that is, they exhibited higher discharge rates during periods of high muscle tone than during periods of atonia (mean discharge rates, 15.92 ± 5.93 during high tone; 6.80 ± 5.61 during atonia) while also firing in anticipation of myoclonic twitches (see below). In addition, the EMG-on/twitch-on neurons had slightly broader average spike durations than did twitch-on neurons (1.92 ± 0.13 ms and 1.63 ± 0.16 ms, respectively).
Over the 60-ms period before a myoclonic twitch, the mean discharge rate of twitch-on neurons increased dramatically, from 2.81 ± 0.87 Hz to 24.22 ± 6.67 Hz. In [Figure 7](#pbio-0030143-g007){ref-type="fig"}D, a representative recording of multiunit activity that includes a twitch-on neuron is depicted, and the tight coupling of the bursts of multiunit activity and nuchal twitches is apparent. [Figure 7](#pbio-0030143-g007){ref-type="fig"}E depicts an expanded view of the boxed area from [Figure 7](#pbio-0030143-g007){ref-type="fig"}D and identifies (with asterisks) an isolated twitch-on unit. [Figure 7](#pbio-0030143-g007){ref-type="fig"}F depicts a peristimulus histogram and raster plot of an identified unit (the same unit depicted in [Figure 7](#pbio-0030143-g007){ref-type="fig"}E) triggered on nuchal EMG spikes. Immediately preceding the onset of a twitch, the discharge rate increased more than 30-fold, suggesting that this neuron participates in the production of twitching. It is important to note, however, that the relationship between neuronal discharges and twitches was not perfect; that is, there were occasions when the neuron discharged and a nuchal twitch was not detected, and there were occasions when a nuchal twitch was detected but the neuron did not discharge.
Lesions of Subcoeruleus and Nucleus Pontis Oralis Reduce Atonia Durations and Produce "REM Sleep without Atonia" {#s2i}
----------------------------------------------------------------------------------------------------------------
As described above, the SubLC contains a high concentration of atonia-on neurons (see [Figure 5](#pbio-0030143-g005){ref-type="fig"}); in addition, both the present study (see [Figure 7](#pbio-0030143-g007){ref-type="fig"}) and that of Corner and Bour \[[@pbio-0030143-b10]\] suggest that the PO is important for regulating infant sleep. To examine the causal roles played by these areas in sleep and wakefulness, both areas were lesioned here using high concentrations (50 mM) of quisqualic acid, a glutamate receptor agonist ([Figure 8](#pbio-0030143-g008){ref-type="fig"}).
::: {#pbio-0030143-g008 .fig}
Figure 8
::: {.caption}
###### Effects of Brainstem Lesions on the Expression of Nuchal Muscle tone in P8 Rats
\(A) Representations of lesions of the dorsolateral pontine tegmentum (DLPT; blue), pontis oralis (PO; green), and subcoeruleus (SubLC; red) on coronal sections. Outlined areas indicate the extent of all lesions in a group, and filled areas indicate the smallest lesion in a group.
\(B) Mean ± standard error bout durations of atonia and high muscle tone for sham (*n* = 7; black), PO lesion (*n* = 5; green), and SubLC lesion (*n* = 4; red) groups. **†** Significant difference from sham group, *p* \< 0.05. Pie charts beneath each bar graph indicate percentage of time spent in atonia or high muscle tone. \* Significant difference from sham group, *p* \< 0.05.
\(C) Mean ± standard error bout durations of atonia and high muscle tone for sham (*n* = 7; black) and DLPT lesion (*n* = 6; blue) groups. **†** Significant difference from sham group, *p* \< 0.05. Pie charts beneath each bar graph indicate percentage of time spent in atonia or high muscle tone. \* Significant difference from sham group, *p* \< 0.05.
\(D) Representative 4-min recordings of nuchal EMG for a sham pup and for pups with PO, SubLC, and DLPT lesions. DT, dorsal tegmental nucleus; IC, inferior colliculus; LC, locus coeruleus; LDT, laterodorsal tegmental nucleus; PC, nucleus pontis caudalis; PO, nucleus pontis oralis
:::
:::
The extent of the lesions of the SubLC (*n* = 4) and PO (*n* = 5) is depicted in [Figure 8](#pbio-0030143-g008){ref-type="fig"}A. As shown in the bar graphs of [Figure 8](#pbio-0030143-g008){ref-type="fig"}B, these lesions produced significant changes in the mean bout durations of atonia and high muscle tone (atonia, *F*\[[@pbio-0030143-b02],[@pbio-0030143-b13]\] = 10.02, *p* \< 0.01; high tone, *F*\[[@pbio-0030143-b02],[@pbio-0030143-b13]\] = 4.07, *p* \< 0.01). Moreover, as indicated by the pie charts in [Figure 8](#pbio-0030143-g008){ref-type="fig"}B, the percentage of time in atonia was significantly reduced in the lesion groups (*F*\[[@pbio-0030143-b02],[@pbio-0030143-b13]\] = 14.2, *p* \< 0.0005). Representative recordings of nuchal EMG in pups with SubLC and PO lesions are shown in [Figure 8](#pbio-0030143-g008){ref-type="fig"}D.
Lesions of SubLC and PO also disrupted the expression of myoclonic twitching of the limbs and tail. Although the absolute number of myoclonic twitches during the 15-min scoring period did not differ between the lesion and sham groups (*F*\[[@pbio-0030143-b02],[@pbio-0030143-b13]\] = 0.79, not significant), lesions of SubLC and PO resulted in significant increases in the proportion of twitches that occurred against a background of high muscle tone (*F*\[[@pbio-0030143-b02],[@pbio-0030143-b13]\] = 7.91, *p* \< 0.01). Specifically, in the SubLC and PO lesion groups, respectively, 27.1% and 16.1% of all twitches were expressed against a background of high muscle tone, in contrast with only 0.01% in the sham group.
Lesions of the Dorsolateral Pontine Tegmentum Increase Atonia Durations and Suppress Myoclonic Twitching {#s2j}
--------------------------------------------------------------------------------------------------------
Because the DLPT contained a high concentration of EMG-on and twitch-on neurons (see [Figure 7](#pbio-0030143-g007){ref-type="fig"}), and because of difficulties using chemical lesion techniques in this area (see [Materials and Methods](#s4){ref-type="sec"}), electrolytic lesions of the DLPT were performed; the extent of the lesions is shown in [Figure 8](#pbio-0030143-g008){ref-type="fig"}A. As shown in the bar graphs of [Figure 8](#pbio-0030143-g008){ref-type="fig"}C and in contrast to lesions of SubLC and PO, lesions of the DLPT produced a sizable increase in the mean bout duration of atonia (*t* = 6.7, *df* = 11, *p* \< 0.01) and a decrease in the mean bout duration of high muscle tone (*t* = 3.3, *df* = 11, *p* \< 0.01). Consequently, as indicated by the pie charts in [Figure 8](#pbio-0030143-g008){ref-type="fig"}C, the percentage of time in atonia was significantly increased in the lesion group (*t* = 6.4, *df* = 11, *p* \< 0.0001). A representative recording of nuchal EMG in a pup with a DLPT lesion is shown in [Figure 8](#pbio-0030143-g008){ref-type="fig"}D.
In contrast with lesions of SubLC and PO, DLPT lesions significantly reduced the number of twitches. In the sham group, the mean number of limb twitches over the 15-min period was 518.3 ± 34.3, compared to 302.9 ± 51.7 in the lesion group (*t* = 3.3, *df* = 11, *p* \< 0.01). Also, in contrast to the SubLC and PO lesion groups and similar to shams, less than 0.1% of twitches in pups with DLPT lesions were observed during periods of high muscle tone.
Discussion {#s3}
==========
In the present study, we outline, to our knowledge for the first time, the neural substrates of two primary sleep components in week-old rats. First, we confirm that the inputs to the MIA from medullary and mesopontine structures are similar to those reported in adults \[[@pbio-0030143-b21],[@pbio-0030143-b22],[@pbio-0030143-b23]\]. Second, neurons exhibiting atonia-on discharge profiles (indicative of sleep) were found predominantly within the SubLC, and neurons exhibiting EMG-on discharge profiles (indicative of wakefulness) were found predominantly within the LDT. Third, we report the presence of neurons that exhibit a distinct bursting pattern that anticipates myoclonic twitches. Fourth, consistent with the recording data, we demonstrate that atonia durations are decreased after lesions of SubLC and PO, and myoclonic twitching is reduced after lesions of the DLPT despite a dramatic increase in atonia duration. Finally, lesions of SubLC and PO decoupled myoclonic twitching from nuchal atonia, producing a condition that resembles REM sleep without atonia as described in juvenile and adult rats \[[@pbio-0030143-b24],[@pbio-0030143-b25]\].
The present study adds critical new evidence in support of the notion that sleep in infant rats satisfies most conventional criteria for defining sleep, including state-dependent neural activity \[[@pbio-0030143-b13],[@pbio-0030143-b14]\]. This study also supports the contention that the development of sleep in mammals entails the continuous elaboration of components that are identifiable in early infancy \[[@pbio-0030143-b05],[@pbio-0030143-b06],[@pbio-0030143-b26],[@pbio-0030143-b27],[@pbio-0030143-b28]\]. These experiments are particularly significant in light of the fact that they were performed using an altricial species at ages before the onset of sleep-related neocortical activity \[[@pbio-0030143-b01],[@pbio-0030143-b02],[@pbio-0030143-b03]\]. Consequently, these results may prove to be broadly applicable to our understanding of sleep mechanisms in other mammalian species at an analogous time in development, including the period before 120 d postconception in sheep \[[@pbio-0030143-b29],[@pbio-0030143-b30]\], 50 d postconception in guinea pigs \[[@pbio-0030143-b31]\], approximately 32 wk postconception in preterm human infants \[[@pbio-0030143-b32]\], and P13--15 in kittens \[[@pbio-0030143-b28]\].
We did not find state-dependent activity in the DR despite its heavy labeling after infusions of DiI into the MIA. Nonetheless, it is possible that the DR does contain state-dependent neurons at the ages studied here. For example, it is possible that cessation of DR activity during sleep depends on histaminergic input from the tuberomammillary nucleus \[[@pbio-0030143-b33]\], a region that was disconnected by the precollicular decerebrations used here (see [Materials and Methods](#s4){ref-type="sec"}). Conversely, within the sparsely labeled LDT, we found neurons that exhibited state-dependent discharge profiles.
The medulla of adults contains neurons that participate in the generation of both sleep and wake phenomena \[[@pbio-0030143-b34]\]. In adult cats, classes of medullary neurons have been identified that discharge during wakefulness, during REM sleep \[[@pbio-0030143-b35],[@pbio-0030143-b36],[@pbio-0030143-b37]\], and during REM sleep as well as waking movements \[[@pbio-0030143-b35],[@pbio-0030143-b38],[@pbio-0030143-b39]\]. Medullary neurons with similar discharge profiles have also been identified in adults rats \[[@pbio-0030143-b40],[@pbio-0030143-b41]\]. In the present study, we found medullary EMG-on neurons that compare well with the wake-on neurons reported in adult rats \[[@pbio-0030143-b40],[@pbio-0030143-b41]\] and cats \[[@pbio-0030143-b38],[@pbio-0030143-b39]\], atonia-on neurons that compare well with the REM-sleep specific neurons reported by Kanamori et al. \[[@pbio-0030143-b36]\], and AS-on neurons that compare well with the phasic REM sleep neurons reported by Steriade et al. \[[@pbio-0030143-b35]\]. It should be noted that we did not find AS-on neurons in the brainstem that exhibited high, tonic rates of discharge typical of some REM-on neurons in adults \[[@pbio-0030143-b33]\]. More extensive sampling will be necessary to describe fully the activity patterns exhibited by AS-on neurons in infants.
To our knowledge, LC recordings have not been reported previously in unanesthetized infant rats. Moreover, the EMG-on profile of the LC units described here is consistent with their wake-active profile in adult rats \[[@pbio-0030143-b42]\]. Interestingly, this LC activity profile is consistent with the notion that increased olfactory sensory thresholds during infant sleep result from a decrease in excitatory LC input to mitral cells of the olfactory bulbs \[[@pbio-0030143-b15],[@pbio-0030143-b43]\].
In adult cats, REM sleep depends on the integrity of the peri-LC-α, which corresponds to the sublaterodorsal nucleus (SLD) in rats \[[@pbio-0030143-b33],[@pbio-0030143-b44],[@pbio-0030143-b45]\] and is contained within the region designated as the SubLC here. In a recent study aimed at elucidating the neural substrates of sleep in adult rats, Boissard et al. \[[@pbio-0030143-b44]\] demonstrated that REM sleep can be induced by increasing SLD activity, either by decreasing GABAergic inhibition or by increasing glutamatergic excitation. Moreover, it was demonstrated that the SLD projects to neurons in the medial medulla that also stain positively for glycine \[[@pbio-0030143-b44]\].
The current study supports and amplifies the findings of Boissard et al. \[[@pbio-0030143-b44]\]. First, we did not find any neurons within the SubLC that exhibited an EMG-on profile, and the SubLC contained 17 of the 29 atonia-on neurons reported here. Second, using DiI, we demonstrated that the SubLC connects directly with the MIA. Third, chemical lesions of the SubLC significantly reduced atonia durations. Finally, we note that excitation of the infant MIA with a glutamate agonist is sufficient to produce atonia \[[@pbio-0030143-b17]\]. All together, these findings are consistent with the notion that atonia in infant and adult rats is produced in part by glutamatergic excitatory input from the SubLC to glycinergic neurons of the ventromedial medulla that, in turn, hyperpolarize spinal motoneurons \[[@pbio-0030143-b46],[@pbio-0030143-b47]\].
Although lesions of both the SubLC and PO dramatically reduced the duration of atonia periods, they did not reduce rates of myoclonic twitching. It is not surprising, then, that there were many instances when twitches were observed during periods of high muscle tone, a decoupling of tonic and phasic sleep components that resembles REM sleep without atonia produced in adults after pontine \[[@pbio-0030143-b24]\] or SubLC \[[@pbio-0030143-b48]\] lesions.
The increase in atonia and the reduction in twitching after DLPT lesions may have resulted, respectively, from the disinhibition of atonia facilitatory neurons in the reticular formation (perhaps within the PO) and the loss of twitch-on neurons within the LDT. Of course, it is also possible that these electrolytic lesions exerted their effects by destroying nearby structures or fibers of passage. We do know, however, that although the DLPT lesions were large and frequently included the LC and SubLC, the effect of the DLPT lesions on atonia and twitching cannot be attributed to damage to either of those two areas, because electrolytic lesions of the LC \[[@pbio-0030143-b19]\] and chemical lesions of the SubLC (as shown in the present study) reduced mean atonia durations without affecting twitching. Thus, after large DLPT lesions, the PO may contribute to the elongation of atonia periods via direct and mutually facilitatory interactions with the MIA \[[@pbio-0030143-b49]\].
Previous work in adult cats has explored the importance of the cholinergic mesopontine tegmentum in REM sleep generation \[[@pbio-0030143-b50],[@pbio-0030143-b51]\]. The present findings are consistent with this earlier work (although it should be stressed that we have not demonstrated here that these infant mesopontine mechanisms utilize acetylcholine). Webster et al. \[[@pbio-0030143-b50]\] reported significant decreases in REM sleep after large lesions that destroyed the majority of cholinergic neurons within the LDT, PPT, the parabrachial nucleus, and the LC. Shouse and Siegel \[[@pbio-0030143-b51]\] demonstrated that lesions primarily involving the PPT resulted in the loss of phasic sleep components, whereas lesions of the SubLC resulted in REM sleep without atonia. In contrast, lesions primarily localized to the LDT and LC did not alter the expression of REM sleep components. Recently, it was reported that discrete PPT lesions in adult rats did not affect sleep time but did interfere with sleep propensity after sleep deprivation \[[@pbio-0030143-b52]\]; unfortunately, phasic sleep events were not recorded in that study.
In adult rats, REM sleep induced by selective activation of the SLD lacks phasic components (e.g., rapid eye movements) \[[@pbio-0030143-b44]\]. The authors of that study hypothesized that activation of the phasic components of REM sleep requires activation of neurons within the LDT, presumably in the form of burst-pause discharges such as those identified previously in reticular neurons in adult cats \[[@pbio-0030143-b38],[@pbio-0030143-b39]\]. The presence of such discharges in adult rats, however, has been questioned \[[@pbio-0030143-b53]\]. Here we demonstrate that the LDT (and the parabrachial nucleus) contains twitch-on neurons that appear to participate in the generation of myoclonic twitching of the limbs and nuchal muscle. The burst-pause discharge profile of these twitch-on neurons compares well with neurons recorded in the pontine tegmentum of adult cats (located among fibers of the brachium conjunctivum) that exhibit sharp bursts that anticipate pontine-geniculo-occipital waves \[[@pbio-0030143-b54]\]. In this regard, it may be significant that LDT neurons burst after activation of the low-threshold calcium current \[[@pbio-0030143-b55],[@pbio-0030143-b56],[@pbio-0030143-b57]\], thus allowing them to behave as single-cell oscillators \[[@pbio-0030143-b58]\]. We have here, then, a mechanism by which LDT neurons can produce phasic events such as myoclonic twitches. The reduced rates of twitching after DLPT lesions, as shown here, is consistent with this view.
The bursting of a twitch-on neuron was not always associated with a nuchal muscle twitch. One possible explanation for this is that the downstream effects of the activity of twitch-on neurons are gated by the moment-to-moment levels of glycinergic inhibition provided by descending axons from the MIA \[[@pbio-0030143-b17],[@pbio-0030143-b46],[@pbio-0030143-b47]\]. It is also possible that a burst of activity by twitch-on neurons serves to momentarily override the hyperpolarization of spinal motoneurons and thus allows spinal networks to exhibit behaviorally observable spontaneous activity \[[@pbio-0030143-b59]\].
It has recently been suggested that myoclonic twitches in infants contribute to the self-organization of neural circuits within the spinal cord that govern withdrawal responses to aversive stimuli \[[@pbio-0030143-b60]\]. Specifically, it was argued that a limb movement during a twitch provides sensory feedback that helps to calibrate interactions among neurons within a reflex module; according to this hypothesis, the temporal correlations among motor and sensory events modify synapses through a Hebbian-type learning process. Indeed, we note that the information generated by the feedback from a single limb twitch against a background of muscle atonia represents a situation that is well suited for Hebbian learning to occur \[[@pbio-0030143-b61]\]. But this idea may also apply to any myoclonic twitch, whether produced by phasic activation of limb muscles, nuchal muscle, or eye muscles; in fact, phasic activity during sleep in infants is a global phenomenon occurring as temporally coherent bouts of activity in many muscle groups throughout the body \[[@pbio-0030143-b62]\]. Therefore, we believe that myoclonic twitching will prove to play a broad role in early motor and somatosensory development \[[@pbio-0030143-b26],[@pbio-0030143-b63]\].
Whereas the ontogeny of mammalian sleep remains a topic of dispute \[[@pbio-0030143-b07],[@pbio-0030143-b14]\], some aspects of that debate can now be put to rest. First, the extracellular recordings reported here, and previously \[[@pbio-0030143-b17]\], demonstrate that the discharge profiles of multiple medullary and mesopontine nuclei are tightly coupled to behavioral state. Second, it appears that infant sleep is not mediated by distinct neurophysiological mechanisms, but rather by the very medullary and mesopontine nuclei that participate in sleep expression in adults. Such findings are not easily reconciled with numerous previous statements concerning the ontogeny of mammalian sleep and render moot attempts to explain sleep ontogeny as the gradual emergence of a coherent state from a set of undifferentiated, distinct, diffuse, or primitive sleep mechanisms \[[@pbio-0030143-b01],[@pbio-0030143-b07],[@pbio-0030143-b08],[@pbio-0030143-b09]\]. In contrast, we have argued that the development of sleep entails the elaboration of an elementary sleep circuit that is largely in place soon after birth in rats \[[@pbio-0030143-b14]\]; accordingly, sleep components that emerge during ontogeny do so by integrating with those components that already exist. Of course, significant differences exist between infant and adult sleep and should not be discounted; for example, perhaps the most striking difference between infant and adult sleep is the rapidity with which infants cycle between sleep and wakefulness \[[@pbio-0030143-b19],[@pbio-0030143-b64]\].
Establishing the developmental relations between sleep in infants and adults has important implications for theories concerning the functions of sleep. Many such theories can be applied to infants only with difficulty, because they address behaviors or contexts that are specific to adults. In addition, some theories place particular emphasis on the cerebral cortex \[[@pbio-0030143-b65],[@pbio-0030143-b66]\], a part of the brain that, as mentioned above, does not exhibit state-dependent activity at those ages when infants are thought to sleep the most (e.g., before P12 in rats). If infant and adult sleep were as mechanistically distinct as some have argued, then one might imagine that sleep serves qualitatively different functions at different ages. But, as shown here, the basic mechanisms of infant and adult sleep are not distinct---even at ages before state-dependent neocortical activity is expressed---suggesting that there may be fundamental similarities between the functions of infant and adult sleep that have largely been overlooked. Thus, we believe that obtaining a clearer understanding of the mechanisms of sleep in infants is central to the goal of developing a theory of sleep that has applicability across the lifespan and can account for the inordinate amount of sleep during infancy \[[@pbio-0030143-b67]\].
Materials and Methods {#s4}
=====================
{#s4a}
All experiments were performed in accordance with National Institutes of Health guidelines for the care of animals in research and were approved by the Institutional Animal Care and Use Committee of the University of Iowa. All efforts were made to minimize the number of animals used and their suffering.
### Subjects {#s4a1}
A total of 64 P6--P10 Sprague-Dawley Norway rats *(Rattus norvegicus)* of both sexes from 46 litters were used. Litters were culled to 8 pups within 3 d of birth (day of birth = day 0). Mothers and their litters were housed in standard laboratory cages (48 × 20 × 26 cm) in the animal colony at the University of Iowa where food and water were available ad libitum. All animals were maintained on a 12:12 h light-dark schedule with lights on at 07:00 h. All experiments were conducted during the lights-on phase.
### Retrograde tracing {#s4a2}
Ten P8 rats from four litters were used (body weights, 16.1--22.9 g). While the pup was anesthetized with isoflurane, the head was secured in a stereotaxic instrument (David Kopf Instruments, Tujunga, California, United States) and leveled between bregma and lambda. A small burr hole was drilled over the target area and the retrograde tracer DiI (Molecular Probes, Eugene, Oregon, United States; 20 nl of 5% DiI in DMSO) was infused at the rate of 1nl/s using a microsyringe (model 7000.50C, Hamilton, Reno, Nevada, United States). Infusions were localized to the medial medulla (*n* = 5) and immediately lateral to it (*n* = 5), as described previously \[[@pbio-0030143-b17]\]. Coordinates relative to lambda for the medial site were AP, −2.5 mm; ML, 0.0 mm; DV, −12.3 mm; for the lateral site, the coordinates were AP, −2.5 mm; ML, 1.0 mm; DV, −12.3 mm. After the infusion was complete, the syringe remained in place for 5 min, was then retracted slowly, and the wound closed using cyanoacrylate adhesive.
After surgery, pups recovered in a warm, humidified incubator for 1--2 h before being returned to the home cage. After 26--28 h, the pups were overdosed with sodium pentobarbital (approximately 100 mg/kg intraperitoneally) and perfused transcardially with physiological saline followed by 3% formalin. Brains were postfixed in the skull for 1--2 d in a formalin-sucrose solution and then removed from the skull and fixed for at least two more days in a fresh solution. After fixation, the brains were sliced in 50-μm coronal sections with a sliding microtome (model SM 2000 R, Leica, Bensheim, Germany). The slices were mounted and coverslipped with Vectashield (Vector Laboratories, Burlingame, California, United States) and stored at 2--4 °C until inspected and photographed under fluorescent light using a digital microscope (model BX-51, Olympus, Japan).
Sections were visually inspected throughout the pontomedullary axis, and seven areas were selected for cell counting ([Table 1](#pbio-0030143-t001){ref-type="table"}). Next, digital photographs of one representative section of all seven structures were obtained from the same coronal level from each subject. Every effort was made to ensure that the cell counts were performed at the same level in each brain (roughly corresponding to plates 54, 55, 56, 58, and 66 in the adult rat brain atlas of Paxinos and Watson \[[@pbio-0030143-b68]\]). Finally, labeled cells were counted manually from each photo using ImageJ software (National Institutes of Health, Bethesda, MD).
### Electrophysiology {#s4a3}
Twenty-five P6--P10 rats from 20 litters were used (body weights, 13.4--25.7 g). Under isoflurane anesthesia, a precollicular decerebration was performed and bipolar nuchal EMG electrodes were implanted bilaterally. The infant\'s fragile skull was bleached, dried, and then coated with Vetbond (3M, St. Paul, Minnesota, United States) to add strength. Next, a custom-built, T-shaped, stainless steel head-plant, designed to attach to the earbar and nosebar holders of a stereotaxic instrument, was attached to the skull over the pretreated area using cyanoacrylate adhesive gel. After securing the head-plant and leveling the skull, a small hole was drilled over the recording site. Finally, to inhibit movement and calm the pup, it was wrapped in gauze \[[@pbio-0030143-b69]\] without obscuring movements of the tail.
Recordings were performed in a stereotaxic instrument with brain and body temperatures maintained at approximately 37 °C. Within 2--3 h of surgery, pups began exhibiting sleep-wake cycles, as judged by oscillations in muscle tone as well as myoclonic twitches against a background of nuchal atonia \[[@pbio-0030143-b19]\]. When sleep-wake cyclicity was observed, a stainless steel 8-trode (ALA Science, Westbury, New York United States), connected to a unity gain headstage and digital amplifier (Tucker-Davis Technologies, Alachua, Florida, United States), was lowered into the recording site while neurophysiological activity was monitored using an oscilloscope and audio analyzer (FHC, Bowdoinham, Maine, United States). The 8-trode consists of eight individual recording sites distributed about the tip (500--800 kΩ per electrode), with the entire electrode having a diameter of approximately 150 μm. The signals were amplified (10k) and filtered (500--5000 Hz band-pass) before being sampled at 12.5 kHz. When stable units were observed and at least one of these units appeared to exhibit state-dependent activity, the electrode was left in place for 5--10 min before data collection began. Neurophysiological and EMG data were recorded synchronously for 10 min (encompassing 4--12 sleep-wake cycles) to hard disk for off-line analysis. At the end of the experiment, the recording site was marked by passing a 50 μA anodal current through the eight closely spaced electrodes for 3 s (thereby producing a single small lesion).
In the current study, nuchal EMG was used as an indicator of behavioral state. As described previously \[[@pbio-0030143-b19]\], the EMG signal was dichotomized into bouts of high muscle tone and atonia (indicative of wakefulness and sleep, respectively). The sleep state was further categorized based on the presence or absence of phasic events (i.e., based on visual observation of myoclonic twitching of the tail, or based on twitching detected in the nuchal EMG). The reliable relationship between myoclonic twitching as assessed by behavioral observation or nuchal EMG has now been documented \[[@pbio-0030143-b62]\].
Four of the eight continuously recorded channels containing heterogeneous unit activity were selected for off-line analysis. Using Spike2 software (Cambridge Electronic Design, Cambridge, United Kingdom), a threshold was set for each of the four selected channels to extract spike data with a signal-to-noise ratio exceeding 2:1. Using the multichannel data, principal components analysis was then used for spike sorting \[[@pbio-0030143-b70],[@pbio-0030143-b71]\]. The sorted units were assigned to groups using graphical cluster cutting \[[@pbio-0030143-b72],[@pbio-0030143-b73]\]. To remove artificial clusters, spike waveforms were inspected and autocorrelations were constructed for each cluster. A cluster was deemed to contain a single unit only if the autocorrelation analysis indicated a refractory period of at least 2 ms.
Three types of analyses were performed. The first analysis was aimed at classifying neurons as either EMG-on or atonia-on. Periods of nuchal atonia and high muscle tone were identified from the EMG record, as described above. Next, the firing rates of individual sorted units were compared with the behavioral state of the animal across 4--12 pairs of high tone and atonia periods. State-related differences in mean discharge rates were tested using the Wilcoxon matched-pairs signed-ranks test \[[@pbio-0030143-b74]\]; alpha was set at 0.05. Neurons that significantly increased their firing rates during periods of high muscle tone were designated as EMG-on, and neurons that significantly increased their firing rates during atonia periods were designates as atonia-on.
The second analysis was aimed at determining whether discharge rates differed between sleep bouts that did or did not contain myoclonic twitching. Discharge rates during twelve 4-s segments of atonia with less than or equal to one nuchal EMG spike (indicative of sleep without phasic processes; i.e., QS), and discharge rates during twelve 4-s segments of atonia with more than four nuchal EMG spikes (indicative of sleep with phasic processes; i.e., active sleep, AS) were compared using the Wilcoxon matched-pairs signed-ranks test; alpha was set at 0.05.
The third analysis was aimed at revealing the relationship between the nuchal EMG twitches and neuronal burst activity. To this end, an event marker was placed on the center of all nuchal EMG twitches that occurred during a recording session. Next, a peristimulus histogram of unit activity triggered on the event marker was generated. Neurons were classified as twitch-on if the average discharge rate increased at least 4-fold before a twitch (measured 0--20 ms and 40--60 ms before a twitch).
For all experiments presented here, means are presented with their standard errors.
Brains were prepared for histological analysis as described above. After fixation, the brains were sliced in 25--50-μm sections with a sliding microtome and stained with cresyl violet. The locations of the marking lesions were determined by examining serial sections.
### Lesions {#s4a4}
Chemical lesions of the SubLC and PO were performed in 16 P8 nondecerebrated rats from ten litters (body weights, 17.7--23.9 g). A pup was anesthetized with isoflurane and its head was secured in a stereotaxic instrument, and the skull was leveled between bregma and lambda. A small burr hole was drilled over the target area and a microsyringe (model 7000.50C, Hamilton) was lowered stereotaxically either into the SubLC (*n* = 4; coordinates relative to lambda: AP, −1.8 mm; ML, ± 1.0 mm; DV, −11.8 mm) or into the PO (*n* = 5; coordinates relative to lambda: AP, −1.2 mm; ML, ± 1.0 mm; DV, −12.3 mm). Lesions were produced by infusing a high concentration of quisqualic acid (50 mM; 150 nl over 15 s) as described previously \[[@pbio-0030143-b17]\]. Control animals were treated identically but received infusions of Ringers solution only into the PO (*n* = 5) and SubLC (*n* = 2).
Lesions of the DLPT were performed in 13 P8 nondecerebrated rats from 12 litters (body weights, 17.4--21.8 g). Because previous attempts to use ibotenic acid in the brainstem of pups resulted in very high rates of mortality, and attempts to use quisqualic acid to lesion the DLPT were unsuccessful, electrolytic lesions were used here. After the pup was prepared for stereotaxic surgery, as described above, a concentric, bipolar, tungsten electrode (1 MΩ, 3--4 μm at the tip, model TM33CCINS, World Precision Instruments, Sarasota, Florida, United States) connected to a stimulus isolator (model A395, World Precision Instruments) and a stimulus generator (model 48, Grass, Quincy, Massachusetts, United States) was lowered into the PPT/LDT area (coordinates relative to lambda: AP, −0.8 mm; ML, ± 0.8 mm; DV, −4.4 mm). Lesions were produced by applying a 150-μA current for 15--30 s (*n* = 7). Control animals received identical treatment but without the current (*n* = 6).
After a chemical or electrolytic lesion was produced, bipolar nuchal EMG electrodes were implanted bilaterally, the pup was placed on a felt pad in a supine position (to enable observation of twitching in the individual limbs), and lightly restrained as described previously \[[@pbio-0030143-b04]\]. Next, the pups recovered in a humidified incubator (35 °C) for 3 h. After recovery, pups were intubated and infused with 2--4% of their body weight with warm cream and transferred to a humidified recording chamber maintained at 35 °C. A microcamera placed above the lid of the recording chamber allowed recording of sleep-wake behaviors, as described elsewhere \[[@pbio-0030143-b04]\]. Video data were recorded using a digital video system (model DV8, Vetron, Rebersburg, Pennsylvania, United States). After 1 h of acclimation to the chamber, synchronized behavioral and EMG data were collected for 1 h and recorded to digital tape for off-line scoring and analysis.
EMG signals were digitized at 5 kHz using a data acquisition system (BioPac Systems, Santa Barbara, California, United States). Next, the EMG signals were summed, integrated, and full-wave rectified, and mean bout durations of high-tone and atonia periods were measured over 60 min, as described previously \[[@pbio-0030143-b19]\]. Next, a trained observer viewed a 15-min taped segment of the pup\'s behavior and, using an event recorder, pressed a key when myoclonic twitches were detected \[[@pbio-0030143-b04]\]. Myoclonic twitches are phasic, rapid, and independent movements of the limbs and tail; high-amplitude movements comprise such wake-related behaviors as stretching, kicking, and yawning \[[@pbio-0030143-b75],[@pbio-0030143-b76]\]. We and others have used similar scoring procedures in the past and have found them to be highly reliable, with inter- and intrarater reliability coefficients typically exceeding 0.85 \[[@pbio-0030143-b77],[@pbio-0030143-b78]\]. The total number of twitches during the 15-min scoring session was determined, as well as the number of twitches that occurred against a background of high muscle tone.
For the DLPT lesion experiment, group differences in the expression of sleep components were assessed using an unpaired t-test. For the SubLC and PO lesion experiment, group differences were assessed using a single-factor analysis of variance (ANOVA); the post hoc test was Fisher\'s PLSD. For all tests, alpha was set at 0.05.
Histological methods were identical to those described above.
This research was supported by grants from the National Institute of Mental Health (MH50701 and MH66424). We thank Cynthia Shaw and Jessica Middlemis-Brown for technical assistance. We also thank the staff at the Central Microscopy Research Facilities at the University of Iowa.
**Competing interests.** The authors have declared that no competing interests exist.
**Author contributions.** KÆK and MSB conceived and designed the experiments. KÆK, AJG, EJM, and AMHS performed the experiments. KÆK and AJG analyzed the data. KÆK and MSB wrote the paper.
Citation: Karlsson KÆ, Gall AJ, Mohns EJ, Seelke AMH, Blumberg MS (2005) The neural substrates of infant sleep in rats. PLoS Biol 3(5): e143.
AS
: active sleep
DLPT
: dorsolateral pontine tegmentum
DR
: dorsal raphé
EEG
: electroencephalogram
EMG
: electromyogram
Gi
: nucleus gigantocellularis
LC
: locus coeruleus
LDT
: laterodorsal tegmental nucleus
LPB
: lateral parabrachial nucleus
MIA
: medullary inhibitory area
MPB
: medial parabrachial nucleus
P\[number\]
: postnatal day \[number\]
PC
: nucleus pontis caudalis
PO
: nucleus pontis oralis
PPT
: pendunculopontine tegmental nucleus
QS
: quiet sleep
REM
: rapid eye movement
SLD
: sublaterodorsal nucleus
SubLC
: subcoeruleus
|
PubMed Central
|
2024-06-05T03:55:55.122444
|
2005-4-19
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079781/",
"journal": "PLoS Biol. 2005 May 19; 3(5):e143",
"authors": [
{
"first": "Karl Æ",
"last": "Karlsson"
},
{
"first": "Andrew J",
"last": "Gall"
},
{
"first": "Ethan J",
"last": "Mohns"
},
{
"first": "Adele M. H",
"last": "Seelke"
},
{
"first": "Mark S",
"last": "Blumberg"
}
]
}
|
PMC1079782
|
Introduction {#s1}
============
*Prochlorococcus* is the numerically dominant primary producer in the temperate and tropical surface oceans \[[@pbio-0030144-b01]\]. These cyanobacteria are the smallest known photosynthetic organisms (less than a micron in diameter), yet are significant contributors to global photosynthesis \[[@pbio-0030144-b02],[@pbio-0030144-b03]\] because they occur in high abundance (as many as 10^5^ cells/ml) throughout much of the world\'s oceans. They are adapted to living in low-nutrient oceanic regions \[[@pbio-0030144-b04]\] and are physiologically and genetically diverse with at least two "ecotypes" that have distinctive light physiology \[[@pbio-0030144-b05]\], nitrogen \[[@pbio-0030144-b06]\] and phosphorus (L. R. Moore, personal communication) utilization, and copper \[[@pbio-0030144-b07]\] and virus (phage) \[[@pbio-0030144-b08]\] sensitivity. Cyanobacterial phages are also abundant in these environments \[[@pbio-0030144-b08],[@pbio-0030144-b09],[@pbio-0030144-b10],[@pbio-0030144-b11],[@pbio-0030144-b12]\] and have a small, but significant, role in mediating population sizes \[[@pbio-0030144-b09],[@pbio-0030144-b10]\]. Further, cyanophages likely play a role in maintaining the extensive microdiversity within marine cyanobacteria \[[@pbio-0030144-b09],[@pbio-0030144-b10]\] through keeping "competitive dominants" (sensu \[[@pbio-0030144-b13]\]) in check, as well as by carrying photosynthetic "host" genes \[[@pbio-0030144-b14],[@pbio-0030144-b15],[@pbio-0030144-b16]\] and mediating horizontal transfer of genetic material between cyanobacterial hosts \[[@pbio-0030144-b14]\].
Although there are more than 430 completed double-stranded DNA phage genomes in GenBank, only nine phages with published genomes infect marine hosts (cyanophage P60; vibriophages VpV262, KVP40, VP16T, VP16C, K139, and VHML; roseophage SIO1; and *Pseudoalteromonas* phage PM2). Of those nine, only one infects cyanobacteria (cyanophage P60, a member of the Podoviridae). P60 was isolated from estuarine waters using *Synechococcus* WH7803 as a host and appears most closely related to the T7-like phages \[[@pbio-0030144-b17]\]. It contains 11 T7-like phage genes and has no genes with homology to non-T7-like phages. However, it lacks the conserved T7-like genome architecture. Thus, P60 is thought to be only distantly related to the T7-like phages, but still part of a T7 supergroup \[[@pbio-0030144-b18]\] proposed by Hardies et al. \[[@pbio-0030144-b19]\]. The T7 supergroup also contains two other marine phages (roseophage SIO1 and vibriophage VpV262) that show similarity to some (three) T7-like genes. However, these phages lack many T7-like genes including the hallmark T7-like RNA polymerase (RNAP) gene \[[@pbio-0030144-b18]\]. Thus, there is clearly a gradient in relatedness among the T7 supergroup, with these newer marine phage genomes at the distant, less-similar end of the group.
Marine phages are subject to different selection pressures (e.g., dispersal strategies, encounter rates, limiting nutrients, and environmental variability) than their relatively well-studied terrestrial counterparts. Thus, beyond informing phage taxonomy, the analysis of their genomes should unveil "signatures" of these selective agents. For example, genomic analysis of two marine phages, roseophage SIO1 \[[@pbio-0030144-b20]\] and vibriophage KVP40 \[[@pbio-0030144-b21]\], has revealed phosphate-inducible genes. It is thought that these genes play an important regulatory role in the phosphorus-limited waters from which they were isolated. Similarly, some *Prochlorococcus* and *Synechococcus* phages (including the three cyanophage genomes presented here) contain core photosynthetic genes that are full-length, conserved, and cyanobacterial in origin \[[@pbio-0030144-b14],[@pbio-0030144-b15],[@pbio-0030144-b16]\]. They are hypothesized to be important for maintaining active photosynthetic reaction centers---and hence the flow of energy---during phage infection \[[@pbio-0030144-b14],[@pbio-0030144-b15],[@pbio-0030144-b16]\].
With a large collection of phages from which to choose \[[@pbio-0030144-b08]\], we used host range and phage morphology to select strains for sequencing. The selected podovirus (P-SSP7) is very host-specific, infecting a single high-light-adapted (HL) *Prochlorococcus* strain of 21 *Prochlorococcus* and *Synechococcus* strains tested. In contrast, the two myoviruses that were selected cross-infect between *Prochlorococcus* (but not *Synechococcus*) hosts: P-SSM2 can infect three low-light-adapted (LL) host strains, and P-SSM4 can infect two HL and two LL hosts \[[@pbio-0030144-b08]\]. We had no prior knowledge of the gene content of these phages; thus, with regard to their genomes, these phages were selected randomly.
As mentioned earlier, our first survey of these phage genomes led to the surprising discovery of photosynthetic genes in all three *Prochlorococcus* phages \[[@pbio-0030144-b14]\], similar to the findings in *Synechococcus* cyanophages \[[@pbio-0030144-b15],[@pbio-0030144-b16],[@pbio-0030144-b22]\]. In this report, we present a more thorough analysis of these three cyanophage genomes, which, we argue, appear to be T7-like (P-SSP7) and T4-like (P-SSM2 and P-SSM4) phages.
Results/Discussion {#s2}
==================
General Features of the Podovirus P-SSP7 {#s2a}
----------------------------------------
P-SSP7 is morphologically similar to the Podoviridae (tails are short and noncontractile; [Figure 1](#pbio-0030144-g001){ref-type="fig"}A). It also includes a rectangular region of electron transparency ([Figure 1](#pbio-0030144-g001){ref-type="fig"}A) that is similar to the gp14/gp15/gp16 core located at the unique portal vertex found in coliphage T7 \[[@pbio-0030144-b23]\]. Its genome contains 44,970 bp (54 open reading frames \[ORFs\]; 38.7% G+C content; [Figure 1](#pbio-0030144-g001){ref-type="fig"}B), including a T7-like RNAP and a phage-related integrase gene (a more detailed analysis of this feature is discussed later). Thus, the P-SSP7 genome is more T7-like or P22-like than φ29-like among the Podoviridae ([Table 1](#pbio-0030144-t001){ref-type="table"}). Thirty-five percent of the translated ORFs have best hits to phage proteins; nearly all of these are T7-like, whereas none are P22-like ([Figure 1](#pbio-0030144-g001){ref-type="fig"}C). Together, these data suggest that P-SSP7 is most closely related to the T7-like phages. Surprisingly, 11% of the translated ORFs have best hits to bacterial proteins, with well over half of these being cyanobacterial (see later discussion). Roughly half (54%) of the translated ORFs could not be assigned a function ([Figure 1](#pbio-0030144-g001){ref-type="fig"}C).
::: {#pbio-0030144-g001 .fig}
Figure 1
::: {.caption}
###### Features of the *Prochlorococcus* Podovirus P-SSP7
\(A) Electron micrograph of negative-stained podovirus P-SSP7. Note the distinct T7-like capsid and tail structure. Scale bar indicates 100 nm.
\(B) Genome arrangement of *Prochlorococcus* podovirus P-SSP7. The ORFs are sequentially numbered within the boxes, and gene names are designated above the boxes. Gene designations use T7 nomenclature for T7-like genes \[[@pbio-0030144-b24]\] or microbial nomenclature for non-phage genes. Class I, II, and III genes refer to those in T7 \[[@pbio-0030144-b66]\] that belong to gene regions primarily involved in host transcription of phage genes (class I), DNA replication (class II), and the formation of the virion structure (class III). The ORFs are designated by boxes, and in this genome, all ORFs are oriented in the same direction. Although the phage genome is one molecule of DNA, the representation is broken to fit on a single page. Note that the P-SSP7 genome is most similar to genomes of the T7-like phages.
\(C) Taxonomy of best BLASTp hits for P-SSP7. Each predicted coding sequence from the phage genomes was used as a query against the nonredundant database to identify the taxon of the best hit (details in [Materials and Methods](#s3){ref-type="sec"}). Blue slices indicate phage hits, while yellow slices indicate cellular hits.
\(D) Diagrammatic representation of the genomic regions surrounding a putative phage and host integration site. This site consists of a 42-bp exact match between the podovirus P-SSP7 and its host *Prochlorococcus* MED4 located directly downstream of the phage integrase gene and the noncoding strand of a host tRNA gene.
:::
:::
::: {#pbio-0030144-t001 .table-wrap}
Table 1
::: {.caption}
###### Genome-Wide Characteristics of the *Prochlorococcus* Cyanophage P-SSP7 Relative to the Other Recognized Phage Groups within the Podoviridae \[[@pbio-0030144-b105]\]
:::
Y indicates that the feature is present, N indicates that the feature is absent, and a question mark indicates that the presence or absence of the feature is unknown
:::
An examination of the genomes of coliphage T7 and its closest coliphage relatives (T3, gh-1, ΦYe03--12, ΦA1122) revealed that they share 26 genes, which we define as core genes ([Table 2](#pbio-0030144-t002){ref-type="table"}). P-SSP7 has 15 of these 26 core genes and an additional gene *(0.7)* that is common, but not universal, among T7-like phages ([Table 2](#pbio-0030144-t002){ref-type="table"}). Further, only two non-T7-like phage genes were identified in this genome: hypothetical gene *12* from a *Burkholderia* phage, *Bcep1,* of the Myoviridae family, and the phage-related integrase gene discussed later. Strikingly, the T7-like genes found in P-SSP7 are arranged in exactly the same order as in other T7-like phages ([Figure 1](#pbio-0030144-g001){ref-type="fig"}B). The gene content and genome architecture of P-SSP7 contrast with those from the three other sequenced marine podovirus genomes in the T7 supergroup \[[@pbio-0030144-b17],[@pbio-0030144-b19],[@pbio-0030144-b20]\]. SIO1 and VpV262 lack the hallmark T7-like RNAP and contain only three T7-like core genes ([Table 2](#pbio-0030144-t002){ref-type="table"}), whereas cyanophage P60 contains 11 core genes ([Table 2](#pbio-0030144-t002){ref-type="table"}) but clearly lacks the conserved T7-like genome architecture \[[@pbio-0030144-b17]\].
::: {#pbio-0030144-t002 .table-wrap}
Table 2
::: {.caption}
###### Shared Genes in T7-Like Phages
:::
The T7 supergroup contains phages with close similarity to T7 (the T7-like phages T3, gh-1, φYe03-12, and φA1122), as well as more distant relatives (e.g., P60, VpV262, φ-KMV, and SIO1) \[19\]. All T7-like phages are represented as well as the marine phages belonging to the T7 supergroup for comparison. The size (amino acids) of each predicted coding region is presented using gene numbers and function assignments according to T7 terminology \[24\]. For P-SSP7, No e-value is given for ORFs that were assigned using size, domain homology, and synteny. A long dash indicates the lack of a particular gene using standard searches
^a^The best e-value was microbe-related rather than related to the T7-like phages
^b^Putative split genes in cyanophage P60
^c^A putative frameshifted gene in cyanophage P-SSP7
:::
The putative functions of the 16 T7-like genes in P-SSP7 would allow for the majority of host interactions and phage production as follows (T7-like gene designations are shown in parentheses): shutdown of host transcription *(0.7),* phage gene transcription *(1),* degradation of host DNA *(3, 6),* DNA replication *(1, 2.5, 4, 5),* formation of a channel across the cell envelope via an extensible tail *(15, 16)* \[[@pbio-0030144-b24]\], DNA packaging *(19),* and virion formation *(8, 9, 10, 11, 12, 17).* We found two stretches of DNA (frame +1 from nucleotides 9994--10525, then frame +3 from nucleotides 10485--11759) with matches to T7 gp5 (DNA polymerase \[DNAP\]): one corresponding to the 3′-exonuclease and one to the polymerase (nucleotidyl transferase) segments of the T7 enzyme. This region may encode a split variant of T7 family DNAP (V. Petrov and J. Karam, personal communication), an arrangement that has been shown to be functional in archaea \[[@pbio-0030144-b25]\] and some T4-like phages (V. Petrov and J. Karam, personal communication).
As described earlier, we identified only 15 of the 26 core T7-like genes in P-SSP7. What are the functions of the absent gene set? It includes genes that in T7 are involved in ligation of DNA fragments *(1.3),* inhibition of host RNAP *(2),* interactions that are specific to the host cell envelope during virion formation *(6.7, 13, 14),* lysis events *(3.5, 17.5),* small-subunit terminase activity *(18),* and unknown functions *(5.7, 6.5, 18.5)* \[[@pbio-0030144-b23]\]. These same genes are also absent in the marine podovirus genomes in the T7 supergroup (cyanophage P60, vibriophage VpV262, and roseophage SIO1; [Table 3](#pbio-0030144-t003){ref-type="table"}). If we assume a conserved genomic architecture among the T7-like phages, we find hypothetical ORFs in homologous positions to these T7 core genes in P-SSP7 ([Figure 1](#pbio-0030144-g001){ref-type="fig"}B) that may fulfill these core (e.g., *5.7, 6.5, 6.7, 13, 14, 17.5, 18, 18.5*) and common (e.g., antirestriction gene *0.3*) T7-like gene functions. Alternatively, their functions may be unnecessary for this phage.
::: {#pbio-0030144-t003 .table-wrap}
Table 3
::: {.caption}
###### Genome-Wide Characteristics of the *Prochlorococcus* Cyanomyophages P-SSM2 and P-SSM4 Relative to the Other Recognized Phage Groups within the Myoviridae \[[@pbio-0030144-b105]\]
:::
Y indicates that the feature is present, N indicates that the feature is absent, and a question mark indicates that no representative phage genomes have been completely sequenced, so the presence or absence of the character is unknown
^a^ Phage integrates using a transposase rather than a site-specific integrase
:::
The P-SSP7 genome assembled as a circular chromosome, suggesting that it is circularly permuted, thus lacking the terminal repeats that are common among T7-like phages \[[@pbio-0030144-b26]\]. Confirmation of this hypothesis would require direct sequencing of the genome ends (I. Molineux, personal communication), which was not possible in this study because of the difficulty of obtaining significant quantities of purified DNA \[[@pbio-0030144-b27]\].
Hypothesized Lysogeny in P-SSP7 {#s2b}
-------------------------------
One of the more interesting discoveries in the podovirus genome is the presence of a tyrosine site-specific recombinase *(int)* gene ([Figure 1](#pbio-0030144-g001){ref-type="fig"}B), which in temperate phages encodes a protein that enables the phage to integrate its genome into the host genome \[[@pbio-0030144-b28]\]. T7 is a classically lytic phage, and there has been only one other report of *int* genes in a T7-like phage: in an integrated prophage in the Pseudomonas putida KT2440 genome \[[@pbio-0030144-b29]\]. The P-SSP7 *int* contains conserved amino acid motifs previously identified for site-specific recombinases (Arg-His-Arg-Tyr, Leu-Leu-Gly-His, and Gly-Thr \[[@pbio-0030144-b30]\]) suggesting it is functional. Downstream of *int,* we find a 42-bp sequence that is identical to part of the noncoding strand of the leucine tRNA gene in the phage\'s host genome (*Prochlorococcus* MED4) ([Figure 1](#pbio-0030144-g001){ref-type="fig"}D). tRNA genes are a common integration site for phages and other mobile elements \[[@pbio-0030144-b31]\], adding support to the hypothesis that this *int* gene is functional.
P-SSP7 was isolated from surface ocean waters at the end of summer stratification \[[@pbio-0030144-b08]\], when nutrients are extremely limiting. We have hypothesized \[[@pbio-0030144-b08]\] that the integrating phase of the temperate-phage life cycle may be selected for under these conditions; thus, finding the *int* gene in this particular phage is consistent with this hypothesis. None of the complete genome sequences of cyanobacterial hosts reported to date have intact prophages \[[@pbio-0030144-b04],[@pbio-0030144-b32],[@pbio-0030144-b33],[@pbio-0030144-b34]\]. Moreover, temperate phages have not been induced from unicellular freshwater or marine cyanobacterial cultures \[[@pbio-0030144-b09],[@pbio-0030144-b35],[@pbio-0030144-b36]\]. Although some field experiments suggest that temperate cyanophages can be induced from *Synechococcus* \[[@pbio-0030144-b37],[@pbio-0030144-b38]\], prophage integration has not been demonstrated. Thus, experimental validation that P-SSP7 is capable of integration would confirm indirect evidence and establish a valuable experimental system.
General Features of the Myoviruses P-SSM2 and P-SSM4 {#s2c}
----------------------------------------------------
P-SSM2 and P-SSM4 are morphologically similar to the Myoviridae (tails are long and contractile; [Figure 2](#pbio-0030144-g002){ref-type="fig"}). Both have an isometric head, contractile tail, baseplate, and tail fiber structures ([Figure 2](#pbio-0030144-g002){ref-type="fig"}) that are most consistent (but see isometric head discussion later) with the morphological characteristics of the T4-like phages \[[@pbio-0030144-b39]\]. Their genomes also have general characteristics that are fully consistent with T4-like status within the Myoviridae ([Table 3](#pbio-0030144-t003){ref-type="table"}). Both genomes are relatively large: P-SSM2 has 252,401 bp (327 ORFs; 35.5% G+C content; [Figure 3](#pbio-0030144-g003){ref-type="fig"}) and P-SSM4 has 178,249 bp (198 ORFs; 36.7% G+C content; [Figure 4](#pbio-0030144-g004){ref-type="fig"}). An apparent strand bias is noteworthy because only 12 (of 327) and six (of 198) ORFs are predicted on the minus strand in the P-SSM2 and P-SSM4 genomes, respectively. Similar to the lytic T4-like phages, integrase genes were absent. Both genomes assembled and closed, suggesting the circularly permuted chromosome common among the T4-like phages ([Table 3](#pbio-0030144-t003){ref-type="table"}). A large portion of the nonhypothetical ORFs have best hits to phage proteins (14% and 21%, respectively) and bacterial proteins (26% and 21%, respectively; [Figure 5](#pbio-0030144-g005){ref-type="fig"}). The phage hits were most similar to T4-like phage proteins, and about half of the bacterial ORFs were most similar to those from cyanobacteria. As with P-SSP7, most of the translated ORFs from P-SSM2 and P-SSM4 could not be assigned a function (60% and 58%, respectively). The majority of the differences between these two phages are due to the presence of two large clusters of genes (24 total) in P-SSM2 (see [Figure 3](#pbio-0030144-g003){ref-type="fig"}) that are absent from P-SSM4. These clusters contain many sugar epimerase, transferase, and synthase genes that we hypothesize to be involved in lipopolysaccharide (LPS) biosynthesis. The large genome size, collective gene complement, and morphology suggest both P-SSM2 and P-SSM4 are most closely related to T4-like phages.
::: {#pbio-0030144-g002 .fig}
Figure 2
::: {.caption}
###### Electron Micrograph of Negative-Stained *Prochlorococcus* Myoviruses P-SSM2 and P-SSM4
Myovirus P-SSM2 with (A) non-contracted tail and (B) contracted tail, and myovirus P-SSM4 with (C) contracted tail and (D) non-contracted tail. Note the T4-like capsid, baseplate, and tail structure in both myoviruses. Scale bars indicate 100 nm.
:::
:::
::: {#pbio-0030144-g003 .fig}
Figure 3
::: {.caption}
###### Genome Arrangement of the *Prochlorococcus* Myovirus P-SSM2
Gene names are designated above the box representing the ORF where genes were identified; descriptions of genes are in [Table 4](#pbio-0030144-t401){ref-type="table"}. The ORFs located above the centering line are on the forward DNA strand, whereas those below the line are on the reverse strand. Although the genome is one molecule, the representation is broken to fit the page. Colors indicate the putative role for the identified genes as inferred from T4 phage. Gene designations use T4 nomenclature for T4-like genes \[[@pbio-0030144-b104]\] or microbial nomenclature for non-phage genes.
:::
:::
::: {#pbio-0030144-g004 .fig}
Figure 4
::: {.caption}
###### Genome Arrangement of the *Prochlorococcus*Myovirus P-SSM4
Gene nomenclature is as in [Figure 3](#pbio-0030144-g003){ref-type="fig"}.
:::
:::
::: {#pbio-0030144-g005 .fig}
Figure 5
::: {.caption}
###### Taxonomy of Best BLASTp Hits for P-SSM2 and P-SSM4
Each predicted coding sequence from both phage genomes was used as a query against the nonredundant database to identify the taxon of the best hit (details in [Materials and Methods](#s3){ref-type="sec"}). Blue slices indicate phage hits, while yellow slices indicate cellular hits.
:::
:::
::: {#pbio-0030144-t401 .table-wrap}
Table 4
::: {.caption}
###### Shared Genes in T4-like Phages
:::
Table modified from \[22,104\]. The T4 supergroup is divided into T-evens (e.g., T4 and RB69), pseudo T-evens (e.g., RB49 and 44RR2.8t), Schizo T-evens (e.g., Aeh1), and the Exo T-evens (e.g., S-PM2) \[106,107\]. For previously published T4 supergroup phages, only the size (amino acids) of selected predicted coding regions are presented using gene names according to T4 terminology. For P-SSM2 and P-SSM4, the size of each translated gene and the e-value of the best phage--T4-like (or microbe-related see below) e-value is presented; Where no e-value is given, these ORFs were assigned based upon size, domain homology, and synteny except where "Fig.6" is listed, which refers to designations made using tail fiber analyses summarized in Figure 6, and P-SSM2 or P-SSM4 indicates designation made through paralogy. A long dash indicates the lack of a particular gene
^a^The best e-value was microbe-related rather than related to T4-like phages
^b^The gene is split into two segments, often by an intron or homing endonuclease
^c^The gene is fused
:::
::: {#pbio-0030144-t402 .table-wrap}
Table 4
::: {.caption}
###### Continued
:::
:::
The six sequenced T4-like phage genomes (T4, RB69, RB49, 44RR2.8t, KVP40, and Aeh1; available as of 15 May 2004 at <http://phage.bioc.tulane.edu/>) share 75 genes ([Table 4](#pbio-0030144-t401){ref-type="table"}), which suggests a core gene complement required for T4-like phage infection. This core contains 18 genes involved in DNA replication, recombination, and repair, seven regulatory genes, ten nucleotide metabolism genes, 34 virion structure and assembly genes, and six genes involved in chaperonin, lysis exclusion, and other activities. Again, despite cyanobacterial hosts being quite divergent from the hosts of these other T4-like phages, our myoviruses contained 43 and 42 of the 75 T4-like core genes, as well as other noncore T4-like genes in each phage (*uvsX, uvsY,* and possibly *dam, 42,* and *hoc* in P-SSM2; *uvsX, uvsY,* and possibly *dam, 42,* and *denV* in P-SSM4; [Table 4](#pbio-0030144-t401){ref-type="table"}). Furthermore, aside from the low-complexity tail fiber related genes (see "Tail-Fiber-Related Genes in the Myoviruses" below), we found no genes with sequence similarity to any phage type other than T4-like phages.
Slightly fewer than half of the core T4-like genes were absent in both myoviruses P-SSM2 and P-SSM4. P-SSM2 and P-SSM4 lack the genes required for anaerobic nucleotide biosynthesis *(nrdD*, *nrdG*, and *nrdH),* which is perhaps not surprising because these phages were isolated from the well-mixed, oxygenated surface oceans. Both myoviruses also lack homologs to the prohead core-encoding genes *(67* and*68)* of the T4-like phages ([Table 4](#pbio-0030144-t401){ref-type="table"}). However, we note that the capsids of both *Prochlorococcus* myoviruses are isometric (see [Figure 2](#pbio-0030144-g002){ref-type="fig"}), rather than prolate as is often observed for other T4-like phage capsids \[[@pbio-0030144-b39]\]. In T4, mutations in the prohead core proteins (gp67 and gp68) are known to cause a capsid structural defect whereby isometric heads are observed \[[@pbio-0030144-b40],[@pbio-0030144-b41],[@pbio-0030144-b42]\]. Thus, functional homologs of prohead core proteins may not be required for the formation of isometric heads in these *Prochlorococcus* myoviruses.
Other T4-like phage gene functions may be represented by divergent homologs filling the T4-like phage role in these cyanomyophages. P-SSM2 and P-SSM4 lack core T4-like chaperonin genes (*rnlA, 31,* and *57A;* [Table 4](#pbio-0030144-t401){ref-type="table"}) and nucleotide metabolism genes (T4-like pyrimidine biosynthesis: *cd, frd, 1,* and *tk*; [Table 4](#pbio-0030144-t401){ref-type="table"}). However, both P-SSM2 and P-SSM4 contain non-T4-like hsp20-family chaperonins, as well as a non-T4-like gene *(mazG)* that in bacteria is involved in degradation of DNA ([Table 5](#pbio-0030144-t005){ref-type="table"}) \[[@pbio-0030144-b43],[@pbio-0030144-b44]\]. Furthermore, P-SSM2 contains ORFs with high sequence similarity to host-encoded homologs of five genes involved in pyrimidine *(pyrE)* and purine *(purH, purL, purM,* and *purN)* biosynthesis ([Table 5](#pbio-0030144-t005){ref-type="table"}). These non-T4-like genes might compensate for T4-like nucleotide metabolism and/or chaperone genes that are absent. Despite the structural similarities between our myophages (see [Figure 2](#pbio-0030144-g002){ref-type="fig"}) and the T4-like phages, some core virion structural genes (e.g., head genes, *2, 24, 67, 68,* and *inh*; tail/tail fiber genes, *10, 11, 12, 34, 35, 37,* and *wac*) have yet to be identified in these myophage genomes (see [Table 4](#pbio-0030144-t401){ref-type="table"}). Similarly, genes involved in transcriptional regulation *(dsbA, rnlA,* and *pseT),* lysis events *(rIIa* and *rIIb),* and replication, recombination, and repair (DNA ligase, *30;* topoisomerases, *39* and *52;* RNase H, *rnh;* and an exonuclease, *dexA*) also have yet to be identified.
::: {#pbio-0030144-t005 .table-wrap}
Table 5
::: {.caption}
###### Summary Table of Unique Features of *Prochlorococcus* Cyanophage Genomes That Are Uncommon among Known Phages
:::
Non-marine T7-like/T4-like phages completely lack these genes. The size (amino acids) and best BLASTp e-value of each predicted coding region are presented using gene names and function assignments according to their function in cellular organisms. The *hli* genes were assigned using e-value and a signature sequence as reported in Lindell et al. \[[@pbio-0030144-b14]\]. A plus sign indicates that the feature is present in the phage group, otherwise the feature is absent or is yet to be identified. PET, photosynthetic electron transport; PSII, photosystem II reaction center
:::
Tail-Fiber-Related Genes in the Myoviruses {#s2d}
------------------------------------------
Sequence analysis of phage tail fiber genes has revealed extensive swapping of gene fragments between loci \[[@pbio-0030144-b45],[@pbio-0030144-b46]\]. Such exchanges yield phages with altered host ranges \[[@pbio-0030144-b47]\]. Although this mosaic gene construction makes computational identification of tail fiber genes by sequence homology difficult, we have attempted to do so in the two *Prochlorocococcus* T4-like genomes. The analysis is motivated by the belief that understanding mechanisms of attachment and host range is critical for developing assays for studying phage--host interactions in wild populations---one of the underlying motivations of our work with this system.
We identified ORFs as potential tail fiber genes by a three-tiered bioinformatics approach using sequence similarity, repeat analysis, and paralogy (details in [Materials and Methods](#s3){ref-type="sec"}). First, sequence similarity to known tail fiber genes was used to add ORFs to the pool of possible tail fiber genes ([Figure 6](#pbio-0030144-g006){ref-type="fig"}). Seven ORFs in P-SSM2 and three ORFs in P-SSM4 had similarity to known tail fiber genes. In T4, the long tail fiber of T4 is composed of four protein subunits including a proximal-end subunit (gp34) anchoring the fiber to the phage baseplate and a distal-end subunit (gp37) responsible for host recognition and attachment (reviewed in \[[@pbio-0030144-b48]\]). Thus P-SSM2 and P-SSM4 ORFs contained regions similar to T4-like phage distal tail fiber genes (gp37; P-SSM2 orf023, orf033, orf295, and orf298; P-SSM4 orf087) and proximal tail fiber genes (gp34; P-SSM2 orf295 and orf315; P-SSM4 orf026 and orf087). Further, two P-SSM2 ORFs (orf034 and orf315) and a P-SSM4 ORF (orf027) are similar to other known tail fiber genes, albeit with low sequence similarity, and for only a small portion of the ORF.
::: {#pbio-0030144-g006 .fig}
Figure 6
::: {.caption}
###### Bioinformatically Identified Tail Fiber Genes from *Prochlorococcus* Myoviruses
Red bars indicate P-SSM2 ORFs (labeled as M2); blue bars indicate P-SSM4 ORFs (labeled as M4). Due to space constraints, P-SSM2 orf67 and P-SSM4 orf10 are broken as indicated.
:::
:::
Second, ORFs containing repeat sequences were added to the pool of possible tail fiber genes. Both simple (amino acid triplets) and complex (longer amino acid motifs) repeats are associated with phage tail fiber genes \[[@pbio-0030144-b49],[@pbio-0030144-b50]\]. Simple repeats are found in two P-SSM2 ORFs (orf23 and orf28; [Figure 6](#pbio-0030144-g006){ref-type="fig"}), with nearly 49% of orf028 encoding the simple triplet repeat Gly-X-Y (where X and Y are often proline, serine, or threonine). Proteins with extended runs of these collagen-like amino acid motifs are thought to fold into trimeric coiled coils, consistent with a tail-fiber-like structure \[[@pbio-0030144-b50]\]. Complex repeat motifs of 15 to 51 amino acids in length are found in P-SSM2 (orf111 and orf298) and P-SSM4 (orf087; [Figure 6](#pbio-0030144-g006){ref-type="fig"}). Some of these motifs are similar to those found in the long distal tail fiber (gp37) and short tail fiber (gp12) genes in T4, where they encode tandem, beta-strand-rich, supersecondary structural elements that are correlated with the beaded or knobbed shaft structure of these tail fibers \[[@pbio-0030144-b49],[@pbio-0030144-b51]\].
Third, possible tail-fiber-encoding ORFs were identified through paralogy to other *Prochlorococcus* phage tail fiber ORFs already identified ([Figure 6](#pbio-0030144-g006){ref-type="fig"}). This approach follows the observation of homology between three T4 tail fiber genes (gp12, gp34, and gp37) \[[@pbio-0030144-b49]\], which are thought to have arisen via gene duplication events \[[@pbio-0030144-b52]\]. These analyses added four ORFs to the pool of possible tail fiber genes for P-SSM2 (orf021, orf022, orf293, and orf301) and two for P-SSM4 (orf080 and orf082).
After identification of a pool of putative tail fiber genes, we used sequence similarity to known tail fiber and/or baseplate genes as a guideline to annotate ORFs according to the known T4 phage architecture. Three tail-fiber-like ORFs of P-SSM2 (orf111, orf295, and orf298) have N-terminal domains that are similar to T4 baseplate proteins ([Figure 6](#pbio-0030144-g006){ref-type="fig"}). In T4, the N-terminus of the proximal long tail fiber (gp34) is bound to the baseplate via the baseplate protein gp9 and possibly gp10 \[[@pbio-0030144-b53],[@pbio-0030144-b54],[@pbio-0030144-b55]\]. The N-terminus of P-SSM2 orf298 is similar to the P-SSM4 orf081 (a gp9 homolog by sequence), suggesting that P-SSM2 orf298 could be analogous to a T4 proximal long tail fiber subunit (gp34), albeit fused to the baseplate socket in P-SSM2. Although such a fused protein does not appear to exist for the other myophage, P-SSM4, the adjacent reading frame to orf081 encodes a possible tail fiber ORF with significant similarity to C-terminal stretches of P-SSM2 orf298. Thus, it appears that P-SSM4 orf081 and orf082 are orthologous with the PSSM2 orf298 N- and C-terminal regions, respectively. P-SSM2 orf295 also appears to be a tail fiber fused to a baseplate protein, gp10, which, in T4, may also play a role in binding tail fiber proteins, although this role is less clear. Similarly, the very large homologous genes (\>15,000 nt) P-SSM2 orf113 and P-SSM4 orf080 appear fused to baseplate wedge initiator (gp7) homologs, which are not known to bind tail fiber in T4 \[[@pbio-0030144-b53]\]. Regardless of their precise assignments relative to T4 tail fiber genes, these putative fusions likely encode tail fiber subunits that bind directly to the baseplate through incorporation of their N-termini into the baseplate complex. Assuming that the long tail fibers of P-SSM2 or P-SSM4 are composed of more than one kind of protein subunit, as in T4 \[[@pbio-0030144-b48]\], we hypothesize that these baseplate-domain-containing tail fibers are unlikely to determine host specificity, but rather are analogous to the proximal long tail fiber (gp34) or short tail fiber (gp12) of T4.
Thus we identify a pool of 12 and five putative tail-fiber-related genes (awaiting experimental confirmation) in the P-SSM2 and P-SSM4 genomes, respectively. Some are quite large relative to those in T4, whereas others appear fused to baseplate genes, which has not been observed for the T4-like phages.
Metabolic Genes Uncommon among Phages {#s2e}
-------------------------------------
All three cyanophages contained genes that are not commonly found in phages. We have selected the following cyanobacterial genes for discussion because we hypothesize that they could play defining functional roles in the marine cyanophage--cyanobacterium phage--host system.
### Photosynthesis-related genes in cyanophages {#s2e1}
We previously reported photosynthesis-related genes *(psbA* and *hli)* in all three of these *Prochlorococcus* phages, as well as other photosynthesis genes *(petE, petF,* and *psbD)* in one of the two *Prochlorococcus* myovirus genomes \[[@pbio-0030144-b14]\]. In addition, genomic analyses have revealed that P-SSM2 contains *pebA* and *ho1,* whereas P-SSM4 contains *pcyA* and *speD* (see [Table 5](#pbio-0030144-t005){ref-type="table"}). In cyanobacteria these genes are involved in phycobilin biosynthesis *(ho1*, *pebA,* and *pcyA)* \[[@pbio-0030144-b56],[@pbio-0030144-b57]\] and polyamine biosynthesis *(speD)*. Although the phycobilin biosynthesis genes are found in *Prochlorococcus* \[[@pbio-0030144-b04],[@pbio-0030144-b34]\], their function is unclear because *Prochlorococcus* does not have the intact phycobilisomes characteristic of most cyanobacteria. These genes are thought to be a remnant of the evolutionary reduction of the phycobilisome-based antenna to a chlorophyll-b-based antenna \[[@pbio-0030144-b04],[@pbio-0030144-b58],[@pbio-0030144-b59],[@pbio-0030144-b60]\]. Although low levels of phycoerythrin occur in some LL *Prochlorococcus* strains \[[@pbio-0030144-b61]\], they have, as yet, no known function in the host.
The polyamine biosynthesis gene *speD* found in the phage has a homolog in all of the marine cyanobacteria with complete genome sequences. Although its function has not been confirmed in these organisms, SpeD is known to catalyze the terminal step in polyamine synthesis in other prokaryotes, and polyamines affect the structure and oxygen evolution rate of the photosystem II (PSII) reaction center in higher plants \[[@pbio-0030144-b62]\]. Therefore, SpeD, if expressed, may play a role in maintaining the host PSII reaction center during phage infection.
### Nucleotide metabolism genes {#s2e2}
The podovirus P-SSP7 contains an ORF (orf20) with a putative ribonucleotide reductase (RNR) domain (see [Table 5](#pbio-0030144-t005){ref-type="table"}). In prokaryotes and T4-like phages, RNRs provide the building blocks for DNA synthesis through catalyzing a thioredoxin-mediated reduction of diphosphates (e.g., rNDP → dNDP) during nucleotide metabolism \[[@pbio-0030144-b63]\]. Among T7-like genomes, these domains have been observed only in marine phages (see [Table 5](#pbio-0030144-t005){ref-type="table"}) including cyanophage P60 and roseophage SIO1 \[[@pbio-0030144-b17],[@pbio-0030144-b20]\]. An examination of the two genes *(nrdA* and *nrdB)* in P60 that contain homology to RNRs suggests that they represent a split RNR (as described earlier for DNAP): *nrdA* is similar to the 5′-end and *nrdB* is similar to the 3′-end of cyanobacterial class II RNRs (data not shown). When analyzed for the presence of a class II RNR diagnostic motif \[[@pbio-0030144-b64]\], all three marine T7-like phage putative RNRs were found to contain homology to this motif (seven of nine residues in SIO1, P-SSP7; eight of nine residues in P60; as compared to eight of nine residues in the marine cyanobacteria) ([Figure S1](#sg001){ref-type="supplementary-material"}). Furthermore, the putative RNRs are located in the genomes at the distal end of a region homologous to the nucleotide metabolism region in T7 \[[@pbio-0030144-b65]\]. It is plausible that T7-like phage infection in phosphorus-limited environments requires extra nucleotide-scavenging genes.
Both *Prochlorococcus* myoviruses contain the alpha and beta RNR subunits that are found in all known T4-like phages (see [Table 4](#pbio-0030144-t401){ref-type="table"}). The genes have closer sequence homology to those in T4-like phages than cyanobacterial hosts ([Figure S2](#sg002){ref-type="supplementary-material"}). Interestingly, our myoviruses also contain a noncyanobacterial *cobS* gene, which has never been found in phages. This gene encodes a protein that catalyzes the final step in cobalamin (vitamin B12) biosynthesis in bacteria \[[@pbio-0030144-b66],[@pbio-0030144-b67]\], and cobalamin is an RNR cofactor during nucleotide metabolism in cyanobacteria \[[@pbio-0030144-b68]\]. Both physiological assays \[[@pbio-0030144-b69],[@pbio-0030144-b70]\] and genomic evidence \[[@pbio-0030144-b04],[@pbio-0030144-b34]\] indicate that *Prochlorococcus* synthesizes its own cobalamin. It is tempting to speculate that the phage *cobS* gene serves to boost cobalamin production in the host during infection, thus improving the activity of RNRs. However, these phage RNRs clearly contain the α2 and β2 subunits (typical of class I RNRs) and lack the class II motif described earlier. Thus, if the phage *cobS* does increase cobalamin production and if this production increase is important, then either the phage class I RNRs are cobalamin dependent (which is unprecedented) or cobalamin must be useful for some other process.
### Carbon metabolism genes {#s2e3}
In cyanobacteria, the pentose phosphate pathway oxidizes glucose to produce NADPH for biosynthetic reactions (oxidative branch) and ribulose-5-phosphate for nucleotides and amino acids (non-oxidative branch). This pathway (both branches) is particularly important in cyanobacteria for metabolizing the products of photosynthesis during dark metabolism \[[@pbio-0030144-b71]\]. Long ago, it was hypothesized that cyanophages utilize this pathway as a source of energy and carbon when the host is not photosynthesizing \[[@pbio-0030144-b72]\]. Interestingly, genomic sequencing has recently revealed that *Synechococcus* cyanophage S-RSM2 \[[@pbio-0030144-b16]\] and the *Prochlorococcus* cyanophages P-SSM2 and P-SSM4 \[[@pbio-0030144-b14]\] contain a transaldolase gene *(talC)*. In *Escherichia coli,* transaldolase is a key enzyme in the non-oxidative branch of the pentose phosphate pathway \[[@pbio-0030144-b73]\]. It has been suggested that the product of the phage *talC* gene may facilitate phage access to stored carbon pools during the dark period \[[@pbio-0030144-b16]\].
Recent work in E. coli has revealed two genes *(mipB/fsa* and *talC)* that are divergent from the bona fide transaldolases (*talA* and *talB*) \[[@pbio-0030144-b74]\], but encode a structurally similar enzyme \[[@pbio-0030144-b75]\]. Members of this new subfamily (MipB/TalC) of aldolases, which have a striking sequence similarity to each other, can have distinctly different functions, acting either as a transaldolase or fructose-6-phosphate aldolase, but not both \[[@pbio-0030144-b74]\]. All three of the genes previously reported as "transaldolase" genes in cyanophages \[[@pbio-0030144-b14],[@pbio-0030144-b16]\], as well as an ORF in the podovirus P-SSP7, are most similar to these MipB/TalC aldolase genes (see [Table 5](#pbio-0030144-t005){ref-type="table"}; [Figure S3](#sg003){ref-type="supplementary-material"}). The translated cyanophage genes contain 26 (P-SSM2), 28 (P-SSP7 and S-RSM2), and 29 (P-SSM4) of 32 diagnostic (as designated by Thorell et al. \[[@pbio-0030144-b75]\]) amino acid residues ([Figure S4](#sg004){ref-type="supplementary-material"}). In the active site of this enzyme, as inferred from the crystal structure of E. coli fructose-6-phosphate aldolase, eight of 14 residues are not conserved between the MipB/TalC subfamily, varying depending on enzyme specificity (fructose-6-phosphate aldolase versus transaldolase) \[[@pbio-0030144-b75]\]. When aligned with MipB/TalC members of known substrate specificity, the cyanophage putative active site residues match all eight of those enzyme sequences with transaldolase activity ([Figure S4](#sg004){ref-type="supplementary-material"}). Thus, it appears that each of the four cyanophage *talC* genes encodes an enzyme with transaldolase activity. If functional, these genes are likely to be important for metabolizing carbon substrates---which is central to biosynthesis and energy production---during phage infection of cyanobacterial hosts.
### Phosphate stress genes in the myoviruses {#s2e4}
Phosphorus is a scarce resource in the oligotrophic oceans \[[@pbio-0030144-b76],[@pbio-0030144-b77]\]. It is often growth limiting for cyanobacteria \[[@pbio-0030144-b78]\] and is required in significant amounts for phage replication. Thus it is perhaps not surprising that the phosphate-inducible *phoH* gene, which has been found in two marine phage genomes \[[@pbio-0030144-b20],[@pbio-0030144-b21]\], is also found in both *Prochlorococcus* myoviruses (see [Table 5](#pbio-0030144-t005){ref-type="table"}; see [Figures 3](#pbio-0030144-g003){ref-type="fig"} and [4](#pbio-0030144-g004){ref-type="fig"}). Although the *phoH* gene is found widely distributed among both eubacteria and archaea \[[@pbio-0030144-b79]\], including all cyanobacteria, and is known to be induced under phosphate stress in E. coli \[[@pbio-0030144-b80]\], its function has not been experimentally determined. Bioinformatic analyses suggest that these *phoH* genes are part of a multi-gene family with divergent functions from phospholipid metabolism and RNA modification (COG1702 *phoH* genes) to fatty acid beta-oxidation (COG1875 *phoH* genes) \[[@pbio-0030144-b79]\].
Both P-SSM2 and P-SSM4 also contain a phosphate-inducible *pstS* gene---which is also widespread among the archaea and eubacteria, including all known cyanobacteria---that has not been reported in phages. In bacteria, the *pstS* gene encodes a periplasmic phosphate-binding protein involved in phosphate uptake \[[@pbio-0030144-b81]\]. If expressed by the phage, it might serve to enhance phosphorus acquisition during infection of phosphate-stressed cells.
### LPS biosynthesis genes in P-SSM2 {#s2e5}
The myovirus P-SSM2 contains 24 LPS genes that form two major clusters in the genome (see [Figure 3](#pbio-0030144-g003){ref-type="fig"}). Reports of phage-encoded LPS genes have previously been limited to temperate phages \[[@pbio-0030144-b82]\]. Such temperate phage LPS genes are thought to be used during infection and establishment of the prophage state to alter the cell-surface composition of the host, preventing other phages from attaching to the host cell. Although T4-like phages are commonly thought of as lytic phages, the lytic process can be stalled upon infection (sometimes termed "pseudolysogeny") during suboptimal host growth \[[@pbio-0030144-b83]\]. If this phenomenon occurs in marine phages, as has been suggested \[[@pbio-0030144-b22],[@pbio-0030144-b84],[@pbio-0030144-b85]\], then a phage-encoded LPS gene cluster, even in a lytic phage, might maintain a similar functional role.
### Signature genes for oceanic cyanophages? {#s2e6}
Although data are too limited to be conclusive ([Table 6](#pbio-0030144-t006){ref-type="table"}), some of the host genes that appear common in oceanic cyanophages may ultimately represent signature genes for these phages. For example, the genomes of all three cyanophages presented here and five partial genomes (\<5 kb) of *Synechococcus* cyanomyophages presented by Millard et al. \[[@pbio-0030144-b16]\] all contain a *psbA* gene. Further, all three cyanophages presented here contain at least one *hli* and a *talC* gene, and both myoviruses presented here are unique among the phages in that they contain *pstS* and *cobS* ([Table 6](#pbio-0030144-t006){ref-type="table"}). As more phages are sequenced, will we find that these genes are specifically characteristic of oceanic cyanophages? If true, this would provide us with a powerful tool for studying these phages in the wild because quantitative PCR could be used to differentiate between cyanophages and other phages in environmental samples.
::: {#pbio-0030144-t006 .table-wrap}
Table 6
::: {.caption}
###### Signature Cyanophage Genes?
:::
There are genes that are not commonly found in phages, but are commonly found among the limited cyanophage sequences available
^a^ These phage genomes were not completely sequenced, but were part of a study that did targeted analyses of ∼5kb regions surrounding the *psbA* gene. A question mark indicates that the presence or absence of the feature is unknown
:::
Hypothesized Transient Genes {#s2f}
----------------------------
There are genes of interest, found in only one of the myoviruses, that we hypothesize are not functional, but rather were obtained by cyanomyophages through packaging random DNA, probably by illegitimate recombination \[[@pbio-0030144-b86],[@pbio-0030144-b87]\] with DNA from a common phage genome pool \[[@pbio-0030144-b88]\].
### Trytophan halogenase {#s2f1}
P-SSM2 contains a gene *(prnA)* that is known to exist in only nine species of bacteria, in which it encodes a tryptophan halogenase that catalyzes the NADH-consuming first step of four that are involved in converting tryptophan to the antibiotic pyrrolnitrin \[[@pbio-0030144-b89],[@pbio-0030144-b90],[@pbio-0030144-b91]\]. Although this gene is full length ([Figure S5](#sg005){ref-type="supplementary-material"}), *prnA* is part of a unique metabolic pathway missing in most bacteria, including cyanobacteria.
### Archaeal and eukaryotic genes. {#s2f2}
The other myovirus, P-SSM4, contains three grouped genes with homology only to eukaryotic prion-like proteins (orf32), an archaeal protease (orf35), and a hypothetical protein from a eukaryotic slime mold (orf36) (see [Figure 4](#pbio-0030144-g004){ref-type="fig"}). Other eukaryotic and prion-like genes have been predicted in the genomes of mycobacteriophages that infect actinobacterial hosts \[[@pbio-0030144-b92]\], although they have no similarity to those found in P-SSM4.
### Hemagglutinin neuraminidase {#s2f3}
P-SSM4 contains a possible hemagglutinin neuraminidase (HN), which has only been observed in single-stranded RNA (ssRNA) viruses and *Prochlorococcus* MED4 (orf1400). In ssRNA viruses, HN cleaves sialic acid from glycolipids on the host cell surface, which enables these viruses to attach. Protein alignments show, however, that both the MED4 and P-SSM4 HN genes are only partial genes---they are missing the N- and C-termini (approximately 200 amino acids)---relative to other ssRNA HNs ([Figure S6](#sg006){ref-type="supplementary-material"}). It is noteworthy that the HN gene occurs nowhere else in the prokaryotic world except for MED4. Could this gene have been obtained by P-SSP7 through the phage genome pool (sensu Hendrix et al. \[[@pbio-0030144-b88]\]), then transferred to MED4? This postulate is buttressed by the observation that the HN gene in MED4 is found next to three *hli* genes (which encode high-light-inducible proteins)---genes which we have argued earlier are susceptible to horizontal gene transfer in this phage--host system \[[@pbio-0030144-b14]\].
Ecological and Evolutionary Implications of Phages Carrying Host Genes {#s2g}
----------------------------------------------------------------------
*Prochlorococcus* cells are slow-growing (doubling times range from 1 to 10 d), oxygenic phototrophs that thrive in nutrient-poor, aerobic surface waters \[[@pbio-0030144-b01]\]---conditions that are fundamentally different from those of most of the host cells of the phages sequenced to date. Thus, oceanic cyanophages are subject to substantially different selective pressures than most other sequenced phages in the database. The presence in these phages of host genes that are likely involved in the maintenance of photosynthesis, response to phosphate stress, and mobilization of carbon stores during infection may be interpreted as evidence of such unique pressures (see [Table 5](#pbio-0030144-t005){ref-type="table"}).
If phage genomes interact as "local neighborhoods" (sensu Hendrix et al. \[[@pbio-0030144-b88]\]) within a "global phage metagenome" (sensu Rohwer \[[@pbio-0030144-b93]\]), one would expect to find biologically cohesive units akin to species, defined by local gene transfers as proposed for "microbial species" \[[@pbio-0030144-b94]\]. Such cohesive units would be characterized by core genes that determine a general phage infection lifestyle (e.g., T4-like or T7-like), as well as host-specific genes within phages that infect similar hosts. Indeed, 26 and 75 such core genes exist among the T7-like and T4-like phages, respectively (see [Tables 3](#pbio-0030144-t003){ref-type="table"} and [4](#pbio-0030144-t401){ref-type="table"}), and host-specific genes abound among these cyanophages (see [Figures 1](#pbio-0030144-g001){ref-type="fig"}C, [5](#pbio-0030144-g005){ref-type="fig"}A, and [5](#pbio-0030144-g005){ref-type="fig"}B). That these core genes represent mostly morphological and DNA replication genes suggests a T7-like or T4-like lifestyle that would involve a specific means of delivering DNA from host to host (in a tailed, capsid structure) as well as converting the host into a phage factory. Based upon the presence of many such core genes in our *Prochlorococcus* phages, one would predict they would behave as T7-like (P-SSP7; although probably with the ability to integrate into its host) and T4-like phages (P-SSM2 and P-SSM4) during cyanobacterial infection.
Beyond these core genes, our *Prochlorococcus* phages contain many "nonphage" genes that are of greatest sequence similarity to cyanobacterial genes (see [Figures 1](#pbio-0030144-g001){ref-type="fig"}C, [5](#pbio-0030144-g005){ref-type="fig"}A, and [5](#pbio-0030144-g005){ref-type="fig"}B). We speculate that the acquisition and use of some host genes by phages plays an important role in phage ecology, even *shaping* the evolution of the phage host range. The initial host range alterations are likely to occur by phage tail fiber switching \[[@pbio-0030144-b47]\], but beyond that, these co-opted host genes could either shift or expand the phage\'s host range depending upon whether they affect fitness of the phage in the original hosts. Understanding this dynamic fitness landscape will require modeling efforts directed by a thorough knowledge of the mechanisms and relative rates for this complex genetic shuffling---factors that likely underpin the complexity of phage--host interactions in the environment.
Materials and Methods {#s3}
=====================
{#s3a}
### Electron microscopy {#s3a1}
*Prochlorococcus* phages were concentrated using ultracentrifugation. Concentrates were prepared for microscopy by spotting phage lysates onto freshly glow-discharged carbon/formvar--coated copper grids. Grids were negatively stained with 1% uranyl acetate, dried, and viewed in a JEOL (Peabody, Massachusetts, United States) 1200 EXII transmission electron microscope operated at 80 kV.
### Preparation of cyanophages for genome sequencing {#s3a2}
Three *Prochlorococcus* phages were chosen for sequencing based upon their host ranges, which were restricted to *Prochlorococcus* hosts (see Introduction).
Phages were prepared for genomic sequencing as previously described \[[@pbio-0030144-b14],[@pbio-0030144-b95]\]. Briefly, phage particles were concentrated from phage lysates using polyethylene glycol. Concentrated DNA-containing phage particles were purified from other material in phage lysates using a density cesium chloride gradient. Purified phage particles were broken open (SDS/proteinase K), and DNA was extracted (phenol:chloroform) and precipitated (ethanol) yielding small amounts of DNA (\<1 μg). A custom 1- to 2-kb insert linker-amplified shotgun library was constructed by Lucigen (Middletown, Wisconsin, United States) as described previously \[[@pbio-0030144-b95]\]. Additional larger insert (3--8 kb) clone libraries were constructed from genomic DNA by the Department of Energy (Joint Genome Institute, Walnut Creek, California, United States) using a similar protocol to provide larger scaffolds during assembly. Inserts were sequenced by the Department of Energy Joint Genome Institute from all of these clone libraries and used for initial assembly of these phage genomes. The Stanford Human Genome Center Finishing Group (Palo Alto, California, United States) closed the genomes using primer walking.
### Gene identification and characterization {#s3a3}
Protein coding genes were predicted using GeneMark \[[@pbio-0030144-b96]\] and manual curation. Translated ORFs were compared to known proteins in the nonredundant GenBank database (<http://www.ncbi.nlm.nih.gov/BLAST/>) and in the KEGG database (<http://www.genome.ad.jp/kegg/kegg2.html>) using the BLASTp program (<ftp://ftp.ncbi.nih.gov/blast>). Translated ORFs were also analyzed for signal sequences and transmembrane regions using the Web-based software SignalP and TMHMM, respectively (available at the CBS prediction servers; <http://www.cbs.dtu.dk/services/>). Where BLASTp e-values were high (\>0.001) or no sequence similarity was observed, ORF annotation was aided by the use of PSI-BLAST, gene size, domain conservation, and/or synteny (gene order), the last as suggested for highly divergent genes encountered during phage genome annotation \[[@pbio-0030144-b97]\]. Identification of tRNA genes was done using tRNAscan-SE \[[@pbio-0030144-b98]\].
### Taxonomy of best hits {#s3a4}
For global genome comparison, we used BLASTp (e-values \< 0.001) or manual annotation to classify to which group of organisms or phages each predicted coding sequence was most similar. In most cases this was obvious. However, approximately 2% of the coding sequences were less obvious, so we established an operational definition of "most similar" as the query sequence having e-values within four orders of magnitude of the top cluster of organismal types. For example, if a query sequence was similar to noncyanobacterial sequences with e-values of 10^--29^ to 10^--25^ and to cyanobacterial sequences with e-values of 10^--20^ or greater, then, despite sequence similarity to cyanobacterial sequences, the query would be considered noncyanobacterial.
### Tail fiber gene identification {#s3a5}
Tail fiber genes were identified by generating alignments (stand-alone Basic Local Alignment Search Tool, BLAST \[[@pbio-0030144-b99]\], 2.2.8 release) of conceptually translated, computationally identified ORFs from the P-SSM2 and P-SSM4 genomes against a database consisting of 33,270 sequences encompassing all known phage sequences obtained from the NCBI NR database in April 2004. Only ORFs whose alignments to known tail fiber genes were longer than 100 residues and had e-values less than 0.001 were designated as tail-fiber-like. Sequences close to this cutoff were re-aligned using the bl2seq command of BLAST, which computes e-values independently of database size. Tail-fiber-like paralogs were identified by individually aligning the set of tail-fiber-like ORFs with all other ORFs in the genomes. All ORFs with alignments greater than 100 residues and e-values less than 0.001, were designated as tail fiber paralogs. All BLAST searches and alignments were performed with the low-complexity sequence filter and default parameters. Amino acid sequence repeats were identified by self-alignment matrices using the program Dotter \[[@pbio-0030144-b100]\].
### Sequence manipulation and phylogenetic analyses {#s3a6}
Alignments were generated using Clustal X \[[@pbio-0030144-b101]\] and edited manually as necessary. PAUP V4.0b10 \[[@pbio-0030144-b102]\] was used for the construction of distance and maximum parsimony trees. Amino acid distance trees were inferred using minimum evolution as the objective function, and mean distances. Heuristic searches were performed with 100 random addition sequence replicates and the tree bisection and reconnection branch-swapping algorithm. Starting trees were obtained by stepwise addition of sequences. Bootstrap analyses of 1,000 resamplings were carried out. Maximum likelihood trees were constructed using TREE-PUZZLE 5.0 \[[@pbio-0030144-b103]\]. Evolutionary distances were calculated using the JTT model of substitution assuming a gamma-distributed model of rate heterogeneities with 16 gamma-rate categories empirically estimated from the data. Quartet puzzling support was estimated from 10,000 replicates.
Supporting Information {#s4}
======================
Figure S1
::: {.caption}
###### Class II RNR Motif Compared Against Cyanobacterial and Non-T4-Like Phage RNRs
A question mark indicates this sequence data is not known; a period indicates identical residue to the reference sequence; and a dash indicates a gap in the alignment. *Anab, Anabaena; Pro, Prochlorococcus; Syn, Synechococcus*; *Syncy*, *Synechocystis*.
(10 KB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Figure S2
::: {.caption}
###### Distance Tree of RNR Family Proteins, Including Phage Sequences from P-SSM2, P-SSM4, and P-SSP7
Sequences from P-SSM2, P-SSM4, and P-SSP7 are shown in bold. Trees were generated from 900 amino acids. Bootstrap values for distance and maximum parsimony analyses and quartet puzzling values for maximum likelihood analysis, greater than 50%, are shown at the nodes (distance/maximum likelihood/maximum parsimony). Trees were unrooted; abbreviations as in [Figure S1](#sg001){ref-type="supplementary-material"}.
(14 KB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Figure S3
::: {.caption}
###### Distance Tree of Tal Proteins, Including Phage Sequences from P-SSM2, P-SSM4, and P-SSP7
Sequences from P-SSM2, P-SSM4, and P-SSP7 are shown in bold. Trees were generated from 566 amino acids. Bootstrap values for distance and maximum parsimony analyses and quartet puzzling values for maximum likelihood analysis, greater than 50%, are shown at the nodes (distance/maximum likelihood/maximum parsimony). Trees were unrooted; abbreviations as in [Figure S1](#sg001){ref-type="supplementary-material"}.
(14 KB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Figure S4
::: {.caption}
###### Alignment of TalC Subfamily Aldolases, Including Phage Sequences from P-SSM2, P-SSM4, P-SSP7, and S-RSM2
The 32 amino acid residues suggested to be diagnostic by Thorell et al. \[[@pbio-0030144-b75]\] are labeled with an asterisk and shaded where identical to bona fide TalC proteins, whereas the active site residues are labeled with an "at" symbol. Note the active site residues in the cyanophage TalC sequences exclusively match those from enzymes known to have transaldolase activity rather than fructose-6 phosphate aldolase activity.
(14 KB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Figure S5
::: {.caption}
###### Alignment of Tryptophan Halogenase Amino Acid Sequences Deduced from Phage and Cellular Encoded *prnA* Gene Sequences
Note the phage gene appears full-length relative to the other cellular genes. *Bdellovibrio, Bdellovibrio bacteriovorus; Bordtella, Bordetella pertussis; Burkpyrro, Burkholderia pyrrocinia; Caulobacter, Caulobacter crescentus; Myxfulvus, Myxococcus fulvus; Pschloro, Pseudomonas chlororaphis; Pseud\_fl, Pseudomonas fluorescens; Shewanella, Shewanella oneidensis* MR-1; *Xanaxon, Xanthomonas axonopodis; Xancamp, Xanthomonas campestris.*
(35 KB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Figure S6
::: {.caption}
###### Alignment of HN Amino Acid Sequences Deduced from Phage and ssRNA Viral Gene Sequences
Note the *Prochlorococcus* phage and host gene appears to contain only the central region of the gene relative to the other ssRNA viral genes.APMV6, avian paramyxovirus 6; BPIV3, bovine parainfluenza virus 3; Gparamyxovirus, goose paramyxovirus; HPIV1,2,3, human parainfluenza virus 1,2,3; ProMED4, *Prochlorococcus* MED4.
(36 KB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Accession Numbers {#s4a7}
-----------------
The GenBank (<http://www.ncbi.nlm.nih.gov/Genbank/>) accession numbers for the genomes discussed in this paper are MED4 (BX548174), P-SSM2 (AY939844), P-SSM4 (AY940168), and P-SSP7 (AY939843).
We thank David Mead (Lucigen) and Chris Detter (Department of Energy Joint Genome Institute \[DOE JGI\]) for clone library construction from minimal DNA. The sequencing and assembly of the phage genomes was performed by the production sequencing group at the DOE JGI through the Sequence-for-Others Program under the auspices of the US DOE\'s Office of Science, Biological, and Environmental Research Program and the University of California, Lawrence Livermore National Laboratory, under contract number W-7405-ENG-48; Lawrence Berkeley National Laboratory under contract number DE-AC03--76SF00098; Los Alamos National Laboratory under contract number W-7405-ENG-36; and Stanford University under contract number DE-FC02--99ER62873. This research was supported by the US DOE under grant numbers DE-FG02--99ER62814 and DE-FG02--02ER63445, and the National Science Foundation under grant number OCE-9820035 (to SWC). We thank Sherwood Casjens, Drew Endy, Hector Hernandez, and Roger Hendrix for discussions about phage biology, evolution, and RNRs, as well as Virginia Rich, Debbie Lindell, and Erik Zinser for valuable comments on the manuscript.
Particular thanks go to Ian Molineux for providing access to his unpublished T7 Group review chapter and extensive suggestions on the manuscript; the teams of Henry Krisch and Jim Karam for providing data at the T4-like Genome Web site (<http://phage.bioc.tulane.edu>; Jim Karam and Vasiliy Petrov for their analysis of the gp5 DNAP split in P-SSP7; Luke Thompson for analytical assistance with the cyanophage transaldolase family genes; and Anca Segall for finding the 42-bp exact match sequence in P-SSP7 and *Prochlorococcus* MED4 that supported our hypothesis that the P-SSP7 integrase gene might be functional.
**Competing interests.** The authors have declared that no competing interests exist.
**Author contributions.** MBS grew, purified, and extracted DNA from the phages. Non-authors (see Acknowledgments) prepared clone libraries, sequenced the inserts, and assembled the genomes. MBS, MLC, and FR did the majority of the genome annotation, while PW evaluated tail-fiber-related genes and provided electron micrographs of the particles. MBS and SWC wrote the majority of the paper with significant contributions from all authors, as well as non-authors (detailed in the Acknowledgments).
Citation: Sullivan MB, Coleman ML, Weigele P, Rohwer F, Chisholm SW (2005) Three *Prochlorococcus* cyanophage genomes: Signature features and ecological interpretations. PLoS Biol 3(5): e144.
DNAP
: deoxyribonucleic acid polymerase
HL
: high-light-adapted
HN
: hemagglutinin neuraminidase
LL
: low-light-adapted
LPS
: lipopolysaccharide
ORF
: open reading frame
PSII
: photosystem II
RNAP
: ribonucleic acid polymerase
RNR
: ribonucleotide reductase
ssRNA
: single-stranded ribonucleic acid
|
PubMed Central
|
2024-06-05T03:55:55.127403
|
2005-4-19
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079782/",
"journal": "PLoS Biol. 2005 May 19; 3(5):e144",
"authors": [
{
"first": "Matthew B",
"last": "Sullivan"
},
{
"first": "Maureen L",
"last": "Coleman"
},
{
"first": "Peter",
"last": "Weigele"
},
{
"first": "Forest",
"last": "Rohwer"
},
{
"first": "Sallie W",
"last": "Chisholm"
}
]
}
|
PMC1079783
|
Introduction {#s1}
============
Alternative pre-mRNA splicing is a major determinant of the protein functional diversity underlying human physiology, development, and behavior \[[@pbio-0030158-b01]\]. This process combines exonic sequences in various arrangements to generate two or more mRNA transcripts from a single gene. Splicing patterns are inherently flexible, with variations observed in different cells and tissues and at different stages of development \[[@pbio-0030158-b02]\]. Inducible changes in splicing pattern can also occur as a function of cell excitation in neuronal systems, T cell activation, heat shock, or cell cycle changes \[[@pbio-0030158-b03],[@pbio-0030158-b04],[@pbio-0030158-b05],[@pbio-0030158-b06]\]. Thus, a central problem is to understand the combinatorial mechanisms that adjust splicing patterns in different biological systems. A related issue is to understand how splicing errors, including alterations in splicing patterns, arise from inherited mutations or polymorphisms and contribute to human disease \[[@pbio-0030158-b07],[@pbio-0030158-b08],[@pbio-0030158-b09]\].
Splicing decisions occur in the context of the spliceosome, a highly complex molecular machine containing the small nuclear ribonucleoprotein particles U1, U2, and U4/U5/U6, and a host of protein factors \[[@pbio-0030158-b10],[@pbio-0030158-b11],[@pbio-0030158-b12]\]. Spliceosome assembly occurs in a stepwise fashion to recognize the appropriate splice sites, to fashion the small-nuclear-ribonucleoprotein-particle-based catalytic activity, and to couple the splicing process with transcription, 3′ end formation, and nuclear export. Exon definition, or recognition of the exon as a unit, occurs early in spliceosome assembly, and its efficiency depends upon the strengths of the adjacent splice sites, as well as auxiliary splicing regulatory elements.
RNA control elements, which are distinct from the canonical splice sites, include the positive-acting exonic splicing enhancers (ESEs) and intronic splicing enhancers, and the negative-acting exonic splicing silencers (ESSs) and intronic splicing silencers \[[@pbio-0030158-b08],[@pbio-0030158-b13],[@pbio-0030158-b14],[@pbio-0030158-b15],[@pbio-0030158-b16],[@pbio-0030158-b17]\]. In order to achieve 100% inclusion of the exon in the processed mRNA, constitutive exons generally require some combination of ESEs in addition to the adjacent splice sites. Serine-arginine-rich (SR) protein factors are important mediators of splicing enhancement in both constitutive and alternative splicing. These proteins recognize ESE motifs through their RNA binding domains, and recruit splicing factors or interact with splice sites via interactions with their RS domains \[[@pbio-0030158-b18],[@pbio-0030158-b19],[@pbio-0030158-b20]\].
Alternative splicing affects the majority of human protein coding genes \[[@pbio-0030158-b21],[@pbio-0030158-b22]\], but the molecular control mechanisms are poorly understood. Molecular dissection of a handful of prototypical alternatively spliced genes has shown that cassette exons are included at a frequency that depends on their complex arrangement of positive and negative RNA control elements. It is thought that combinatorial control, which involves the integrated actions of multiple RNA control elements and protein regulatory factors, is the basis of tissue-specific patterns of splicing. Many protein factors of the SR protein and heteronuclear ribonucleoprotein (hnRNP) protein families have been implicated in these mechanisms, and some of their expression patterns are tissue-specific. The polypyrimidine tract binding protein (PTB/hnRNP I), for example, plays important roles in mechanisms of negative control important for brain- and muscle-specific splicing events. Current evidence indicates that PTB/hnRNP I takes part in silencing by recognizing RNA elements containing UCUU and related motifs, and, through protein oligomerization, blocks recognition of the exon by the normal splicing machinery \[[@pbio-0030158-b23]\]. The hnRNP A1 protein has also been implicated in a variety of cellular and viral splicing silencing mechanisms through its cooperative recognition of U AGGG\[U/A\] and related motifs \[[@pbio-0030158-b24]\].
The CI cassette exon (exon 19) of the *GRIN1* transcript (NMDA-type glutamate receptor, NR1 subunit) is a valuable model to study mechanisms of regulation because of its striking patterns of tissue-specific splicing and developmental regulation in the rat brain \[[@pbio-0030158-b25],[@pbio-0030158-b26]\]. (Note that the CI exon is referred to as E21 in these previous studies.) The CI exon is prominently included in the forebrain, and prominently skipped in the hindbrain, but the control mechanisms underlying these patterns are poorly understood. The RNA binding protein NAPOR/CUGBP2 is thought to positively regulate this exon since this factor promotes CI cassette exon inclusion in co-expression assays, and because its tissue-specific expression correlates with the spatial distribution of mRNA transcripts containing the CI exon in rat brain \[[@pbio-0030158-b26]\]. In mammals, NMDA-type glutamate receptors are assembled from *GRIN1* (NR1) and *GRIN2A* (NR2) subunits, and they play highly important roles impacting learning and memory functions in the brain. Alternative splicing is used extensively for the generation of the brain-specific *GRIN1* transcripts, and CI exon inclusion affects the trafficking of NMDA receptors to the synapse \[[@pbio-0030158-b27],[@pbio-0030158-b28]\].
In many cases tissue-specific exon inclusion is modulated by combinations of sequence motifs acting cooperatively or antagonistically \[[@pbio-0030158-b29]\]. An understanding of the essential ingredients for splicing silencing should allow de novo identification of skipped exons from genomic sequence. Here molecular approaches were used to identify sequences responsible for silencing the CI cassette exon, and this analysis was extended using computational methods to explore the distribution and functional relevance of the identified motifs in mammalian genomes. It is a paradox that the CI cassette exon undergoes predominant exon skipping in particular regions of the brain, since its adjacent splice sites match well to consensus patterns. In our previous study, the downstream intron was shown to play a role in silencing, but the factors involved were not defined \[[@pbio-0030158-b26]\].
Here we define a ternary sequence code---two exonic UAGGs and a 5′-splice-site-proximal GGGG---that imposes silencing on an inherently strong CI cassette exon. We further extend this analysis to investigate the roles of hnRNP proteins and the generality of this type of mechanism genome-wide using molecular and bioinformatics approaches. The association of exon silencing with a UAGG and GGGG motif pattern in human and mouse exons otherwise unrelated to the CI cassette supports the generality of this mechanism, and this is consistent with the demonstrated flexibility in the spatial positioning of the UAGG components of the code.
Results {#s2}
=======
A 5′-Splice-Site-Proximal GGGG and Two Exonic UAGG Motifs Are Required in Combination for Silencing of a Brain-Region-Specific Exon {#s2a}
-----------------------------------------------------------------------------------------------------------------------------------
The 5′ splice site of the CI cassette exon is atypical because of an adjacent GGGG motif, which is conserved in human, rat, and mouse *GRIN1* genes. GGGG motifs in the first ten nucleotides of human introns are generally infrequent (see below). In the case of the CI cassette exon, the GGGG motif is immediately adjacent to the U1 small nuclear RNA complementary region of the 5′ splice site, and the overall complementarity of the 5′ splice site (6 bp) is typical for mammals (6 to 7 bp), including all of the most highly conserved positions (−1 to +5).
The role of the GGGG motif in splicing silencing of the CI cassette exon was examined by generating site-directed mutations in nucleotides +6, +7, and +8 of the intron. These mutations were designed so as not to disrupt the U1 small nuclear RNA complementary nucleotides, which include the last nucleotide of the CI exon and the first five nucleotides of the adjacent intron. Splicing assays involved transfecting splicing reporters into non-neuronal mouse myoblasts (C2C12 cells), followed by measurement of the levels of the exon-included and exon-skipped products by RT-PCR relative to the wild-type sequence.
Each mutation in the GGGG motif led to a dramatic increase in exon inclusion ([Figure 1](#pbio-0030158-g001){ref-type="fig"}A). The strongest effects were observed when the GGG at +6 to +8 was converted to CCC (mutation 5m2) or AUA (5m4), which resulted in an approximately 4-fold increase in exon inclusion, compared to the wild-type sequence. Even a point mutation (5m9) resulted in a 3-fold increase in exon inclusion. Thus, the GGGG motif plays an important role in the silencing mechanism. Additional sequence changes upstream and downstream of the GGGG motif had only modest effects on splicing. For example, mutations 5m1, 5m13, and 5m14 were designed to test potential RNA secondary structures involving the GGGG motif and complementary intron sequences. The modest changes in the splicing pattern resulting from these mutations do not support a significant role in splicing for these hypothetical structures.
::: {#pbio-0030158-g001 .fig}
Figure 1
::: {.caption}
###### Exonic UAGG and 5′ Splice Site GGGG Motifs Are Required in Combination for Silencing of the CI Cassette Exon
\(A) A GGGG splicing silencer motif at the 5′ splice site. Top: Sequence of the 5′ splice site region (5′ to 3′) with exonic (uppercase) and intronic (lowercase) nucleotides. Numbering is relative to the first nucleotide of the intron. Arrowhead indicates 5′ splice site. A predicted SRp40 motif overlying the last seven bases of the exon is indicated. Engineered mutations and names of splicing reporters are indicated immediately below the affected nucleotides. Effect of mutations on the pattern of splicing is shown in a 5′ to 3′ arrangement (gel panel and graph). All splicing reporter plasmids have a three-exon structure in which CI is the middle exon (in the schematic, vertical bars indicate exons and horizontal lines indicate introns). Splicing reporter plasmids were expressed in vivo in mouse C2C12 cells, and splicing patterns assayed by radiolabeled RT-PCR of cellular RNA harvested from the cells. PCR primers are specific for the flanking exons. Results of multiple experiments are shown graphically as the average percent of exon included in product (y-axis) for each splicing reporter construct (x-axis).
\(B) Analysis of ESE motifs. An exonic UAGG splicing silencer motif overlaps an ASF/SF2 motif. Sequence of the CI exon (5′ to 3′) is shown, with engineered mutations (underscored) and names of splicing reporters indicated immediately below the affected nucleotides (bold). Numbering is relative to the first nucleotide of the exon. Predicted ESE motifs for ASF/SF2 and SC35 are highlighted above the exonic sequence as indicated in brackets. The UAGG motif required for silencing (boxed) is indicated below the overlapping ASF/SF2 motif (asterisk). Effect of mutations on the in vivo pattern of splicing is shown in a 5′ to 3′ arrangement (gel panel and graph).
Error bars in (A) and (B) represent standard deviations.
:::
:::
Other than the GGGG motif at the 5′ splice site, the sequence of this intronic region is devoid of guanosine-rich sequences. Strikingly, introduction of a GGG at intron positions +40 to +42 (5m8) resulted in a 5-fold decrease in exon inclusion. In contrast, two overlapping mutations that did not generate guanosine-rich motifs had little or no effect on the splicing pattern (5m11 and 5m12). Thus, in this context the introduction of a second intronic GGG cluster can shift the splicing pattern toward nearly complete exon skipping.
The possibility that sequences within the CI cassette exon itself might contribute to the silencing mechanism was also explored. Either a scarcity of ESE sequences within the CI cassette exon might weaken exon definition, or the presence of exonic ESS sequences might enforce silencing. A model for the arrangement of ESE motifs in the CI cassette exon was based on the high-affinity sequence-recognition sites for known SR family splicing factors ([Figure 1](#pbio-0030158-g001){ref-type="fig"}B, top). Mutations were then made in the ASF/SF2 ( AGCCCGA, CACCCUG, and CGUAGGU) and SC35 ( CGACCCUA, GGCCUCCA, and GUCCUCCA) motifs to test predictions of this model, anticipating that reduced exon inclusion should result from the disruption of functional ESE motifs.
The results of these experiments show that most of the mutations decreased exon inclusion, consistent with ESE function (mutations E1, E2, E3, E4, E5, and E6; [Figure 1](#pbio-0030158-g001){ref-type="fig"}B). In contrast, a pair of double point mutations in a UAGG sequence beginning at position 93 of the exon generated a substantial increase in exon inclusion, indicative of a silencing role for this sequence (E8 and E9; [Figure 1](#pbio-0030158-g001){ref-type="fig"}B). Note that the overlapping ASF/SF2 motif is disrupted by the E9 mutation, but the E8 mutation generates a different ASF/SF2 motif. An additional six-nucleotide mutation (CAUCGU) that eliminates the ASF/SF2 motif at this position also resulted in a strong increase in exon inclusion (K. H. and P. J. G., unpublished data). These results show that the position 93 UAGG motif functions in C2C12 cells primarily as a silencer rather than as part of an ASF/SF2 motif. These results suggested the possible involvement of the splicing repressor hnRNP A1 based on the similarity of the UAGG motif to the hnRNP A1 high-affinity binding sequence U AGGG\[A/U\] determined previously by SELEX experiments \[[@pbio-0030158-b30]\].
A Motif Pattern for Strong Splicing Silencing: Analysis of Copy Number and Position Effects in Neuronal and Non-Neuronal Cells {#s2b}
------------------------------------------------------------------------------------------------------------------------------
The presence of two natural UAGG motifs in the CI cassette exon raised the question of how silencing might be affected by changes in the number of exonic UAGGs. The number and position of UAGG motifs in the CI cassette exon were altered in the context of the wild-type splicing reporter (wt0) and the effects tested in neuronal (PC12) and non-neuronal (C2C12) cell lines ([Figure 2](#pbio-0030158-g002){ref-type="fig"}). One set of mutations varied the position of the 5′-splice-site-proximal UAGG by disrupting the original motif at position 93 of the exon, and by introducing a new UAGG motif at positions 11, 76, and 100 (splicing reporters E10, E11, and E20). These position variations had small effects on the pattern of splicing, with exon skipping predominating in both cell lines ([Figure 2](#pbio-0030158-g002){ref-type="fig"}, lanes 1--4 and 15--18). The effect of a single UAGG was then examined at different positions of the exon (splicing reporters E8, E13, E14, E15, and E21). The resulting splicing patterns uniformly showed an increase in exon inclusion, and these effects were essentially independent of position ([Figure 2](#pbio-0030158-g002){ref-type="fig"}, lanes 5--9 and 19--23). It was also evident that the level of exon inclusion was higher in C2C12 than in PC12 cells, suggesting that there may be differences in splicing factors that mediate or antagonize silencing in the two cell lines. Nonetheless, each cell line exhibited a similar trend---stronger exon silencing associated with increased copy number of exonic UAGGs. Thus, splicing silencing of the CI cassette exon depends critically on the number of UAGG motifs in the exon, but less so on their relative positions. To further test the prediction that the strength of splicing silencing is linked to the number of UAGGs in the exon, a third UAGG was introduced at position 11 of the exon (splicing reporter E18). As a result, the level of exon inclusion decreased to approximately 0% in both cell lines in agreement with this prediction (lanes 11 and 25).
::: {#pbio-0030158-g002 .fig}
Figure 2
::: {.caption}
###### Effect of Number and Position of CI Cassette Exon Splicing Silencer Motifs
Splicing reporters were constructed with variations in the number and position of UAGG and/or GGGG motifs. Three sets of schematics (boxed at center) illustrate the CI cassette exon and adjacent 5′ splice site region with positions of exonic UAGG (black vertical bars) and 5′ splice site GGGG (grey vertical stripe) motifs. Splicing reporter names are indicated at left. Vertical arrowhead indicates 5′ splice site. Each splicing reporter was generated by site-directed mutagenesis from parent plasmid wt0. Natural UAGG positions 51 and 93 represent the starting position of the motif relative to the first base of the exon. Engineered UAGG positions 11, 76, and 100 are also indicated (see schematic in center box at top). Sequence changes of the mutations are underscored: 11, GUGG→GG; 51, UAGG→GG; 76, CCAG→G; 93, UAGG→GG; 100, UCCAA→U. Representative splicing patterns in PC12 cells (left gel panels) and C2C12 cells (right gel panels) are shown together with average percent exon inclusion values. The correlation between motif pattern and strength of splicing silencing is summarized (bottom). Exon-included (double arrowheads) and exon-skipped (single arrowheads) products are indicated.
:::
:::
The role of the 5′-splice-site-proximal GGGG motif was examined independently by generating exons lacking the two natural UAGG motifs in the presence and absence of the GGGG motif (splicing reporters E17 and T8, respectively; [Figure 2](#pbio-0030158-g002){ref-type="fig"}, lanes 10, 12, 24, 26). The GGGG motif had a small silencing effect in both cell lines in the absence of the exonic UAGGs (compare E17 and T8; lanes 10 versus 12, and 24 versus 26). By contrast, silencing was reduced substantially when the GGGG motif was disrupted by mutation in the presence of intact UAGGs: exon inclusion increased significantly in PC12 cells (from 25% to 67%; compare wt0 and D0: [Figure 2](#pbio-0030158-g002){ref-type="fig"}, lanes 13 and 14), and a similar trend was observed in C2C12 cells (from 32% to 87%; [Figure 2](#pbio-0030158-g002){ref-type="fig"}, lanes 27 and 28). Note that mutant D0 contains two intact UAGGs, but lacks the GGGG motif. Thus, the GGGG motif acts cooperatively with the exonic UAGGs in both of these cell lines. Together these results show that, for the CI cassette, multiple exonic UAGGs combined with a 5′-splice-site-proximal GGGG function cooperatively to specify silencing of an otherwise strong exon.
The 5′-Splice-Site-Proximal GGGG Motif Is Involved in Silencing by hnRNP A1 and Anti-Silencing by hnRNP H {#s2c}
---------------------------------------------------------------------------------------------------------
Next we sought to identify protein factors that interact directly with the UAGG and GGGG motifs in order to guide empirical tests for their roles in splicing silencing. GTP-labeled RNA substrates were subjected to UV crosslinking in HeLa nuclear extracts under in vitro splicing conditions. These experiments showed pronounced crosslinking to a protein doublet in the vicinity of 50 kDa for RNA substrates containing the intact GGGG motif (cs1 and 3h1; [Figure 3](#pbio-0030158-g003){ref-type="fig"}A, lanes 1 and 3). By contrast, a point mutation in the GGGG motif largely disrupts protein binding (cs3 and 3h3; [Figure 3](#pbio-0030158-g003){ref-type="fig"}A, lanes 2 and 4). Because the apparent molecular weights of these proteins and the guanosine-rich binding specificity \[[@pbio-0030158-b31]\] suggested the involvement of hnRNP H/H′ and F proteins, relevant antibodies were obtained for immunoprecipitation experiments. These results identified the bottom band of the doublet as hnRNP F ([Figure 3](#pbio-0030158-g003){ref-type="fig"}A, lanes 5--7), whereas the upper band corresponded to hnRNP H/H′ ([Figure 3](#pbio-0030158-g003){ref-type="fig"}A, lanes 8 and 9). Although the hnRNP F antibody is highly specific, the H/H′ antibody crossreacts with hnRNP F, which is 95% identical to H/H′ at the protein sequence level. Control reactions ([Figure 3](#pbio-0030158-g003){ref-type="fig"}A, lanes 10 and 11) show the background level precipitated with preimmune serum (lane 10).
::: {#pbio-0030158-g003 .fig}
Figure 3
::: {.caption}
###### Identification and Functional Roles of Protein Factors That Bind to GGGG and UAGG Motifs
\(A) Detection of protein binding to the 5′ splice site GGGG motif by UV crosslinking in HeLa nuclear extract. Wild-type (cs1 and 3h1) and mutant (cs3 and 3h3) RNA substrates were internally labeled at guanosine nucleotides; mutations are underscored. Pattern of UV crosslinking is shown following RNase digestion and SDS-PAGE (lanes 1--4). Immunoprecipitation reactions (lanes 5--11) contained the 3h1 substrate together with antibody specific for hnRNP F or H/H′; control samples contained preimmune rabbit serum. Gel panel shows the pellet (P), supernatant (S), and input (I) of the immunoprecipitation reactions following SDS-PAGE. The positions of hnRNP H/H′ and F (arrowheads) and protein molecular weight standards (in kilodaltons) are indicated. The hnRNP F and H/H′ antibodies were a gift of C. Milcarek.
\(B) UV crosslinking of exonic position 93 UAGG motif in HeLa nuclear extract. RNA substrates were prepared with a single radiolabeled nucleotide as indicated by the asterisk; sequences are shown (bottom). The wild-type (wt3) and mutant (mt3) substrates are identical except for the underscored mutation. The A1winner substrate corresponds to the high-affinity hnRNP A1 binding sequence previously identified by SELEX. The position of hnRNP A1 is indicated (arrowhead). Monoclonal antibody 9H10 was a gift of G. Dreyfuss.
\(C) Exon inclusion is enhanced by co-expression of hnRNP F or H. Gel panel shows splicing pattern resulting from co-transfection of wild-type (wt) or mutant (5m2) splicing reporter with hnRNP F or H expression plasmid; splicing reporters are identical to those shown in [Figure 1](#pbio-0030158-g001){ref-type="fig"}A. Control samples were transfected with empty vector; grey wedge indicates two levels (4 and 6 μg) of protein expression plasmid. Arrowhead indicates 5′ splice site. For immunoblot verification of transfected protein expression (bottom), nuclear extracts from transfected cells were separated by SDS-PAGE, transferred to nylon membrane, and developed with an antibody specific for the Xpress tag at the N-terminus of each pcDNA--protein sample. Raw percent exon inclusion values are shown below gel image.
\(D) Silencing effect of hnRNP A1 requires the intact 5′ splice site GGGG motif and full-length downstream intron. Structures of chimeric splicing reporters are shown in which the CI cassette exon and intron flanks were introduced into an unrelated splicing reporter containing sequences from the GABA~A~ receptor γ2 transcript: rGγCI-wt0 (both introns truncated), -up (full-length upstream intron, truncated downstream intron), and -dn (truncated upstream intron, full-length downstream intron). Numbers above indicate length of each intron segment in nucleotides. Arrowhead indicates 5′ splice site. The splicing reporters rGγCI-dn5m2 and -dn5m4 contain the full-length downstream intron with 5′ splice site mutations of [Figure 1](#pbio-0030158-g001){ref-type="fig"}A. Gel panel shows splicing pattern resulting from co-transfection of splicing reporter with hnRNP A1 expression plasmid or vector control. Immunoblot verification of transfected protein expression (bottom) is as described in (C).
:::
:::
Proteins that interact directly with the exonic UAGG motif were identified similarly, except that the RNA substrates contained a single radioactive label in the middle of the UAGG. Even with a single radioactive label, multiple proteins were observed to crosslink to the wild-type substrate, wt3, under splicing conditions ([Figure 3](#pbio-0030158-g003){ref-type="fig"}B, lane 4). To examine hnRNP A1 binding, the SELEX-derived consensus sequence, A1winner, was also tested in parallel. A low efficiency of UV crosslinking of hnRNP A1 has been observed previously \[[@pbio-0030158-b30]\]. The A1winner contains two U AGGGA sequences, and was found to crosslink to hnRNP H/H′ and F, in addition to A1 ([Figure 3](#pbio-0030158-g003){ref-type="fig"}B, lane 1; data not shown). These results show that A1 is immunoprecipitated as an approximately 35-kDa protein from the wt3 sample, as was the case for the A1winner ([Figure 3](#pbio-0030158-g003){ref-type="fig"}B, lanes 1--8). A control substrate, mt3, with a dinucleotide mutation in the UAGG showed little or no immunoprecipitation of crosslinked A1 ([Figure 3](#pbio-0030158-g003){ref-type="fig"}B, lanes 9--11). Thus, these results confirm that hnRNP A1 binds directly to the UAGG motif in the context of the CI cassette exon sequence.
In order to investigate the functional roles of hnRNPs F, H, and A1 in the silencing mechanism, each protein was co-expressed with splicing reporters containing the CI cassette exon, and effects on the splicing pattern were monitored. For the wild-type splicing reporter containing an intact GGGG motif, overexpression of hnRNP F or H was found to enhance CI exon inclusion relative to the pcDNA control ([Figure 3](#pbio-0030158-g003){ref-type="fig"}C, lanes 1--5). These effects were reduced but not eliminated in the presence of the 5m2 splicing reporter, which lacks the GGGG motif ([Figure 3](#pbio-0030158-g003){ref-type="fig"}C, lanes 6--10). These results rule out a role in silencing of the CI exon for hnRNP F and H, and instead support an anti-silencing role for these factors.
Next we asked whether the silencing role of the GGGG motif is mediated through hnRNP A1, since the 5′ splice site of the CI cassette exon is related to the A1 consensus binding motif (ACG:GU AAGGGGAA \[colon defines 5′ splice site\] versus U AGGG\[A/U\]). These experiments also examined the effects of portions of the flanking introns, since our previous study demonstrated a role for the downstream intron in this silencing mechanism. Chimeric splicing reporters contained the CI cassette exon and various portions of the flanking introns inserted between exons 1 and 3 of the GABA~A~ receptor γ2 subunit ([Figure 3](#pbio-0030158-g003){ref-type="fig"}D). When the complete downstream intron was present, co-expression of hnRNP A1 reduced exon inclusion from 78.8% to 29.1%, nearly a 3-fold effect ([Figure 3](#pbio-0030158-g003){ref-type="fig"}D, lanes 5 and 6). In this context, the silencing effect of hnRNP A1 depends upon the intact downstream intron, since the silencing effect was substantially reduced when most of the downstream intron was removed (rGγCI-wt0 and rGγCI-up; [Figure 3](#pbio-0030158-g003){ref-type="fig"}D, lanes 1--4). The role of the 5′ splice site GGGG motif was then examined in the context of the rGγCI-dn reporter by introducing mutations 5m2 and 5m4, which destroy the guanosine cluster. The ability of hnRNP A1 to induce splicing silencing was reduced significantly by these mutations, suggesting that A1 is involved in mediating the cooperative effects of the GGGG motif (rGγCI-dn5m2 and rGγCI-dn5m4 [Figure 3](#pbio-0030158-g003){ref-type="fig"}D, lanes 7--10).
Combinations of UAGG and GGGG Motifs Are Associated with cDNA- and EST-Confirmed Skipped Exons in the Human and Mouse Genomes {#s2d}
-----------------------------------------------------------------------------------------------------------------------------
We next sought to determine the extent to which the CI cassette silencing motif pattern is associated with exon skipping (partial or complete) in the human and mouse genomes. For this analysis, over 90,000 human and mouse orthologous exon pairs were divided into two datasets based on the presence or absence of one or more UAGG motifs at any position in the exon (but not overlapping the splice sites) and a GGGG motif within bases 3--10 of the adjacent downstream intron ([Figure 4](#pbio-0030158-g004){ref-type="fig"}). The percentage of alternatively spliced (skipped) exons in each of these datasets was then determined by use of large-scale, high-stringency alignments of available cDNAs and ESTs to the corresponding genomic loci (see [Materials and Methods](#s4){ref-type="sec"}). If the motif pattern functions generally in splicing silencing, the frequency of exon skipping should be higher in the group of exons containing the UAGG and GGGG motif pattern, compared to those without.
::: {#pbio-0030158-g004 .fig}
Figure 4
::: {.caption}
###### Computational Analysis of UAGG and GGGG Motif Patterns Reveals Association with Exon Skipping Genome-Wide
At the top is a flow chart for the computational analysis used to illustrate the procedure used to identify human exons with and without the CI cassette silencer motif pattern (≥1 exonic UAGG and a 5′-splice-site-proximal GGGG), followed by the determination of the percentage of confirmed skipped exons in each group. The reciprocal pattern (≥1 exonic GGGG and a 5′ splice site UAGG) and related variants were analyzed for comparison as indicated in the graph and table. The graph (middle) shows exons with the motif pattern on the left and the remaining exons without the pattern (w/o) on the right; x-axis, 5′ splice site motif; y-axis, percent confirmed skipped exons; z-axis, exonic motif. Confirmed skipped exons were defined as those skipping events supported by 20 or more individual cDNA and/or EST entries. Exonic motifs were allowed at any position within the exon, but not overlapping the splice sites, and the 5′ splice site motif was restricted to bases 3--10 of the intron. Only exons of 250 nucleotides or fewer were considered. The table (bottom) shows, for each motif pattern, the percentage of confirmed skipped exons within that group (as shown in the graph) and the number of exons in the group (in parentheses). The CI cassette silencer motif pattern is boxed.
:::
:::
In these searches we considered exons of typical size (≤250 bases), and we required each component of the motif pattern to be conserved in sequence and position in the human and mouse orthologous exons. Using these stringent criteria, 16 exons (0.018%) contained the motif pattern, and of these, three were confirmed skipped exons (18.75%). The remaining 90,175 exons (99.98%) lacked the conserved motif pattern, and of these, 4,173 (4.63%) were confirmed skipped exons. The difference in the percentage of skipped exons in these two datasets was significant (*p* \< 0.05). When exon length was not constrained, the fraction of skipped exons with the motifs was slightly lower (15.8%), but still significant (*p* \< 0.05). When this analysis was repeated without requiring conservation of the motif pattern, 227 exons (0.24%) contained the motif pattern, and of these, 18 (7.9%) were confirmed skipped exons (*p* \< 0.05). The remaining 96,292 exons (99.76%) lacked the motif pattern, and of these 4,441 (4.61%) were confirmed skipped exons.
Variations of the CI cassette motif pattern were also analyzed. The reciprocal pattern, one or more GGGG motifs in the exon and a UAGG motif in bases 3--10 of the intron, also showed enrichment for confirmed skipped exons (8.4%) compared to those without this pattern (4.6%) (*p* \< 0.001). Moreover, the occurrence of a 5′ splice site GGGG by itself was found to be associated with exon skipping: exons containing the GGGG motif in bases 3--10 of the intron but lacking UAGG and GGGG within the exon showed a significantly higher rate of exon skipping (7.8%) compared to those without the GGGG intronic motif (4.6%) (*p* \< 0.001). Moving the position of the GGGG motif slightly downstream to bases 11--20 of the intron reduced the fraction of skipped exons observed to background levels (4.6%). Taken together, these data suggest that the close proximity (or overlap) of the GGGG motif to the 5′ splice site may be generally important in silencing, perhaps by limiting binding of U1 or U6 small nuclear ribonucleoprotein particles.
Underrepresentation of UAGG in Constitutive Exons, and Overrepresentation in Skipped Exons {#s2e}
------------------------------------------------------------------------------------------
Underrepresentation of UAGG in constitutively spliced exons and overrepresentation in skipped exons would be expected if this motif frequently plays a role in splicing silencing. To test this idea, approximately 5,000 known human cDNAs were downloaded from Ensembl ([www.ensembl.org](www.ensembl.org), and those containing a full-length ORF were shuffled 50 times using the program CodonShuffle. CodonShuffle randomizes the nucleotide sequence by swapping synonomous codons, preserving the encoded amino acid sequence, codon usage, and base composition of the native mRNA \[[@pbio-0030158-b32]\]. Consequently, the program controls for constraints on the protein coding function of the mRNA, and for constraints on codon usage. Since the ORF is preserved by this type of shuffling, codon arrangements forbid the UAG portion of the UAGG motif to occur in-frame. The occurrence of UAGG was reduced by 1.5-fold in authentic coding sequences as compared to CodonShuffled control sequences (*p* \< 0.001). Thus, the correlation of the motif with exon skipping is statistically significant, and there is modest selection against UAGG sequences for constitutive exons. Next we asked whether UAGG is overrepresented in skipped human exons. As expected, both UAGG and GGGG were found to be significantly overrepresented in skipped exons as compared to constitutive exons in human (χ^2^ = 436 and 87, respectively; *p* \< 10^−5^ for both).
More rigorously, when all possible 5-mers were examined for overrepresentation in orthologous exons that are skipped in both human and mouse, a significant enrichment for U AGGC and U AGGG motifs was found (χ^2^ = 15 and 13, respectively; *p* \< 10^−4^) compared to orthologous pairs of constitutive exons. U AGGA and UAGGU were not significantly overrepresented, but this may be explained by the small dataset used for the analysis (approximately 240 exons), or to functional overlap with ESE sequences. Nonetheless, the appearance of the UAGG motif in two 5-mers indicates the importance of the motif in conserved skipped exons. Overrepresentation of UAGG in skipped exons has also been found for mRNAs expressed in brain and testes, which are enriched for regulated splicing events \[[@pbio-0030158-b33]\].
Identification of Skipped Exons with Conserved UAGG and GGGG Motif Patterns across the Human and Mouse Genomes {#s2f}
--------------------------------------------------------------------------------------------------------------
To identify exons unrelated to the CI cassette that might be silenced by a similar motif configuration, we focused in more detail on the UAGG and GGGG motif pattern by searching for these motifs singly and in combination in the database of approximately 96,000 human and mouse orthologous exons. Exons containing a GGGG in bases 3--10 of the intron and one or more exonic UAGGs were identified in the human and mouse subsets of the database and at the intersection of these datasets. These data are presented as Venn diagrams, and specific examples selected from the intersection dataset are shown to illustrate the motif patterns that are conserved in human and mouse orthologous exons ([Figure 5](#pbio-0030158-g005){ref-type="fig"}). We included in the intersection dataset only exons in which the motif pattern is conserved in sequence and position in the human and mouse orthologous exons.
::: {#pbio-0030158-g005 .fig}
Figure 5
::: {.caption}
###### Genome-Wide Identification of Exons with UAGG and GGGG Silencing Motifs
A database of 96,089 orthologous human and mouse exon pairs was searched for TAGG located anywhere in the exon and GGGG in bases 3--10 of the intron. Venn diagrams indicate the number of exons containing either or both sequence motifs in the human subset and the mouse subset of the database. The number of exons (19) in which UAGG and GGGG silencer motifs are conserved in orthologous human and mouse exons is also shown (intersection). The motif patterns are shown in the context of the exon (uppercase) and 5′ splice site region (lowercase) for 12 examples from the intersection dataset (human sequences are shown). Colon indicates 5′ splice site. The conserved TAGG and GGGG motifs are highlighted in red to illustrate variations in their positions. Gene name (HUGO ID) and exon number within the gene are indicated at far right. For one uncharacterized transcript, the GenBank accession is given instead (NM\_018469\_8).
:::
:::
As expected, the CI cassette exon of the *GRIN1* gene was found in all three of the overlap datasets. Of the 19 exons containing the motif pattern in the intersection dataset, 16 exons of 250 or fewer bases in length were considered for further study based on the observation that skipping of longer exons is quite rare \[[@pbio-0030158-b34]\]. This dataset contained the genes for two well known splicing factors, hnRNP H1 and H3 *(HNRPH1* and *HNRPH3).* Although human hnRNP H1 contains 14 exons and H3 contains ten exons, the UAGG and GGGG motif pattern was found associated with a single exon in each of these genes. As hnRNP H proteins are known to bind to guanosine-rich sequences, the presence of a conserved GGGG motif in the 5′ splice sites of these hnRNP H exons suggests the possibility of autoregulation at the level of splicing.
The hnRNP H exons and additional candidates in the intersection dataset (total of 12) were selected for experimental analysis of splicing patterns by RT-PCR, and to investigate the tissue specificity of the splicing patterns in human tissues ([Figure 6](#pbio-0030158-g006){ref-type="fig"}; [Table 1](#pbio-0030158-t101){ref-type="table"}). The CI cassette exon was included in the analysis as a positive control *(GRIN1).* Skipping of the candidate exons for both human *HNRPH1* and *HNRPH3* was confirmed in several tissues. Exon skipping was also confirmed for candidate exons of *UTRN* and an uncharacterized hypothalamus-expressed gene, and tissue-specific exon skipping was evident for *HNRPH1* exon 5, *HNRPH3* exon 3, and *UTRN* exon 5. To our knowledge, these tissue-specific patterns have not been characterized previously. The results of [Figure 6](#pbio-0030158-g006){ref-type="fig"} were confirmed by DNA sequence analysis of the gel-purified products of the RT-PCR reactions. Although the candidate exon in the *ANXA8* gene was not experimentally validated in our analysis, EST and mRNA evidence confirms that the exon is skipped in cDNA libraries derived from choriocarcinomas ([Table 1](#pbio-0030158-t101){ref-type="table"}). An important caveat is that the true number of skipped exons could be significantly higher than that confirmed by RT-PCR because our sampling of human tissues in these experiments was not exhaustive.
::: {#pbio-0030158-g006 .fig}
Figure 6
::: {.caption}
###### RT-PCR Confirmation of Exon Skipping Patterns in Human Tissues
Eleven orthologous exons (≤250 nucleotides in length) were selected from the analysis of [Figure 5](#pbio-0030158-g005){ref-type="fig"} for RT-PCR analysis in a panel of eight human tissues. These exons are derived from the intersection dataset, in which conserved TAGG and GGGG motifs are present in combination in the human and mouse orthologous exons. Additional cDNA and EST evidence for these skipping events are summarized in [Table 1](#pbio-0030158-t101){ref-type="table"}. Specific primer pairs were designed for each test exon to amplify the exon-included (double arrowhead) and exon-skipped (single arrowhead) products by RT-PCR. Each gel panel shows the products of reactions for a single test exon resolved on agarose gels in the arrangement given in the inset. Gene name, exon number, and Ensembl number (in parentheses) are provided above each gel panel. The far left and far right lanes of each gel panel contain DNA molecular weight markers.
:::
:::
::: {#pbio-0030158-t101 .table-wrap}
Table 1
::: {.caption}
###### Human and Mouse Orthologous Exons Containing TAGG and GGGG Motif Patterns
:::
Entries 1--19 correspond to the intersection dataset of [Figure 5](#pbio-0030158-g005){ref-type="fig"}A, with the human exon listed above the mouse exon. The human subset, h1--h13, and mouse subset, m1--m12, are also shown. ^a^ The Ensembl ID prefixes are ENSG00000- for human and ENSMUSG00000- for mouse. ^b^ Uppercase indicates exonic and lowercase indicates intronic nucleotides. ^c^ Sources: <http://genome.ucsc.edu> and <http://www.ncbi.nlm.nih.gov/>.ND, not determined. DOI: [10.1371/journal.pbio.0030158.t001](10.1371/journal.pbio.0030158.t001)
:::
::: {#pbio-0030158-t102 .table-wrap}
Table 1
::: {.caption}
###### Continued
:::
:::
The mouse orthologs of *HNRPH1* exon 5 and *HNRPH3* exon 3 were chosen for further analysis of their splicing patterns ([Figure 7](#pbio-0030158-g007){ref-type="fig"}, "1 TAGG + GGGG exons"). These splicing patterns were determined using RNA derived from mouse heart and brain tissue, as well as from the mouse C2C12 cell line. For each RNA sample, radioactive RT-PCR reactions were performed for a set of three serial dilutions of the input RNA. Good consistency in the percent exon inclusion values for each set of serial dilutions was evident. Sequence alignments showed that exon 3 of both the human and mouse *HNRPH3* genes contained an additional exonic GGGG motif not found in the orthologous *HNRPH1* exon 5 sequences ([Figure 7](#pbio-0030158-g007){ref-type="fig"}, bottom), which might explain the higher rate of exon skipping observed. *HNRPH1* exon 8 and β-actin exon 2 served as control exons, since these exons do not contain UAGG or GGGG motifs ([Figure 7](#pbio-0030158-g007){ref-type="fig"}, "0 TAGG, 0 GGGG exons"). As expected, the "0 TAGG, 0 GGGG" control exons showed 100% exon inclusion in each case.
::: {#pbio-0030158-g007 .fig}
Figure 7
::: {.caption}
###### Analysis of Splicing Patterns in Mouse Tissues for Variations in the Number of Exonic UAGGs
Splicing patterns were determined by radiolabeled RT-PCR for selected mouse exons. Control reactions include *β-actin* exon 2 and *HNRPH1* exon 8, which were selected because they lack the silencing motifs studied ("0 TAGG, 0 GGGG exons"). *HNRPH1* exon 5 and *HNRPH3* exon 3 are representative of the one TAGG plus GGGG motif pattern ("1 TAGG + GGGG exons"). *Hp1bp3* exon 2, *GRIN1* CI cassette exon, and *NCOA2* exon 13 are examples of tissue-specific exon skipping associated with the two TAGG plus GGGG motif pattern ("2 TAGG + GGGG exons"). *MEN1* exon 8 is also shown. Each gel panel shows splicing patterns tested in RNA samples from mouse heart and brain tissue and mouse C2C12 cells. Gene name, exon number, and Ensembl gene ID (in parentheses) are provided above each gel panel. Curly brackets point to the average percent exon inclusion and standard deviation for each set of serial dilutions; raw values are given immediately below each lane. Sequence alignments (bottom left) of the corresponding human and mouse orthologs illustrate the patterns of silencer motifs (orange). Bold indicates an additional exonic GGGG motif.
:::
:::
The observation that multiple UAGGs are associated with an increased strength of splicing silencing of the CI cassette exon (see [Figure 2](#pbio-0030158-g002){ref-type="fig"}) prompted us to examine several exons with these characteristics that were identified in our searches. From the dataset of 213 human exons containing UAGG and GGGG, 13 exons with two or more UAGGs were identified, and from the dataset of 200 mouse exons containing UAGG and GGGG, 12 exons with two or more UAGGs were identified ([Table 1](#pbio-0030158-t101){ref-type="table"}). Exons within these datasets that had lengths typical for internal coding exons (≤250 bases) were chosen for RT-PCR analysis of their splicing patterns. RNA derived from mouse heart and brain and C2C12 cells confirmed the skipping of *Hp1bp3* exon 2 and *NCOA2* exon 13 and trace levels of skipping for *MEN1* exon 8 (see [Figure 7](#pbio-0030158-g007){ref-type="fig"}). Additional cDNA evidence was found in the databases in support of these splicing patterns ([Table 1](#pbio-0030158-t101){ref-type="table"}). In the case of *Hp1bp3,* sequence alignments showed that two TAGGs and the 5′ splice site GGGG motif were conserved in the human and mouse orthologs, but these exons were not found in the intersection dataset of [Figure 5](#pbio-0030158-g005){ref-type="fig"} because the human exon corresponds to the first exon in the transcript, and consequently was not annotated as an internal exon in the Ensembl dataset. Sequence alignments for the more weakly skipped exons, *NCOA2* exon 13 and *MEN1* exon 8, showed that one or more segments of the motif pattern was imperfect in each set of orthologs (see [Figure 7](#pbio-0030158-g007){ref-type="fig"}, bottom).
Generality of the UAGG and GGGG Motif Pattern for Exon Silencing and Differential Regulation by hnRNP Proteins {#s2g}
--------------------------------------------------------------------------------------------------------------
To test whether the silencing motif pattern identified above for the CI cassette exon is sufficient for exon silencing in vivo, this pattern was introduced into the middle exon of a heterologous splicing reporter, *SIRT1* ([Figure 8](#pbio-0030158-g008){ref-type="fig"}). This middle exon corresponds to the constitutively spliced exon 6 of the human *SIRT1* gene, and lacks any features of the silencing motif pattern to be examined. In these experiments the generality of the motif pattern, as well as the regulatory roles of hnRNP A1 and H were tested. When the GGGG motif was introduced by itself at intron positions 6--9 or 8--11 of the *SIRT1* splicing reporter (substrates *SIRT1*-G6--9 and *SIRT1*-G8--11, respectively), no change in the splicing pattern was observed relative to the parent substrate *SIRT1* ([Figure 8](#pbio-0030158-g008){ref-type="fig"}, lanes 1, 4, and 7). These results indicate that in the *SIRT1* context, the GGGG motif alone is not sufficient to induce exon skipping. However, when two UAGGs were introduced into the middle exon (ESS19), the splicing pattern was shifted substantially, from 100% to 29% exon inclusion ([Figure 8](#pbio-0030158-g008){ref-type="fig"}, lane 10). When the GGGG motif was subsequently introduced into the ESS19 substrate at intron positions 6--9 (ESS19-G6--9), exon inclusion was further reduced to 18% ([Figure 8](#pbio-0030158-g008){ref-type="fig"}, lane 13), showing the combined effects of the motif pattern. In this context, the effect of the intronic GGGG motif was position dependent, since no additional silencing was observed when the GGGG was moved to positions 8--11 of the intron (ESS19-G8--11).
::: {#pbio-0030158-g008 .fig}
Figure 8
::: {.caption}
###### Analysis of UAGG and GGGG Motif Pattern in a Heterologous Context and Effects of hnRNP A1 and H Co-Expression
At the top is a schematic of the heterologous splicing reporter *SIRT1* (pZW8) that contains exon 6 of the human *SIRT1* gene and flanking intron sequences as described previously \[[@pbio-0030158-b17]\]. The intron/exon lengths (in nucleotides) are as follows: exon 1, 308; intron 1, 340; exon 2, 95; intron 2, 287; and exon 3, 436. The silencing motif pattern was introduced sequentially into the middle exon and adjacent 5′ splice site region as highlighted in red. GGGG mutations were introduced by site-directed mutagenesis at positions 6--9 or 8--11 of the second intron. Exonic UAGG motifs were introduced into the middle exon by replacing a HindIII-KpnI restriction fragment TCGAATTC, with GT **TAGG**TA TA (restriction sites are underscored) as described \[[@pbio-0030158-b17]\]. The percent exon inclusion values (average of three repeats) were determined from co-expression assays with vector backbone (vbb) or with a 1:4 ratio of hnRNP A1 or H expression plasmid (same as experiment of [Figure 3](#pbio-0030158-g003){ref-type="fig"}). Below are shown splicing assays following expression in C2C12 cells in the absence and presence of hnRNP A1 ("A1" lanes) or hnRNP H ("H" lanes) protein expression vector. Control reactions contained vector backbone plasmid ("−" lanes). Exon-included (double arrowheads) and exon-skipped (single arrowheads) products are indicated.
:::
:::
Based on the effects of hnRNP A1 and hnRNP H on the level of CI cassette exon inclusion described above (see [Figure 3](#pbio-0030158-g003){ref-type="fig"}), we also tested the effects of these factors with the new splicing reporter substrates in co-expression assays. Relative to the vector backbone controls, the co-expression of hnRNP A1 down-regulated exon inclusion, consistent with the presence of the complete silencing motif pattern or exonic UAGGs, and co-expression of hnRNP H had the opposite effect (see [Figure 8](#pbio-0030158-g008){ref-type="fig"}, lanes 10--18). The differential effects of hnRNP A1 and H were both dependent upon the presence of exonic UAGGs, since no change in the splicing pattern was observed for substrates *SIRT1, SIRT1*-G6--9, or *SIRT1*-G8--11 (see [Figure 8](#pbio-0030158-g008){ref-type="fig"}, lanes 1--9). Interestingly, these results suggest that hnRNP H can exert its anti-silencing effect through the exonic UAGGs.
To further investigate the generality of exon silencing by UAGG and GGGG motifs, we examined a subset of the exons identified by bioinformatics to assess their splicing patterns and sensitivity to regulation by hnRNP A1 and hnRNP H in the *SIRT1* heterologous context. Exons containing the silencing motif pattern should be skipped exons, and regulation by these splicing factors would generally be expected for exons that contain the silencing motif pattern. For the convenience of testing new exons in this context, the *SIRT1* splicing reporter was modified to introduce restriction sites 12 nucleotides upstream and 12 nucleotides downstream of the middle exon. Test exons with 12 nucleotides of flanking intron on each side were then cloned from mouse genomic DNA and inserted in place of the *SIRT1* exon 6 between the restriction sites ([Figure 9](#pbio-0030158-g009){ref-type="fig"}). As controls, the middle exon of the *SIRT1* splicing reporter and the middle exon of ESS19 were reinserted in this context to generate new splicing reporters identical to those tested above except for the added restriction sites. The splicing patterns of these modified substrates, *SIRT1*a and ESS19a, were found to be essentially identical to those of *SIRT1* and ESS19 shown above, which shows that the restriction sites have no effect on the splicing pattern in these assays.
::: {#pbio-0030158-g009 .fig}
Figure 9
::: {.caption}
###### Exons Identified by Bioinformatics to Contain UAGG and GGGG Motif Patterns: Analysis of Splicing Patterns in a Heterologous Context and Effects of hnRNP Co-Expression
Exons were cloned from mouse genomic DNA with 12 nucleotides of flanking intron sequence on each side. Each fragment was inserted into the *SIRT1*a context between the NdeI and XbaI restriction sites (Test Exon). *SIRT1*a is identical to *SIRT1* except for the NdeI and XbaI sites located 12 nucleotides upstream and downstream of the middle exon, respectively. The segment between the NdeI and XbaI restriction sites represents the region of *SIRT1*a replaced by test exons. ESS19a is the same as ESS19 except for the presence of the indicated restriction sites. Test exons included the CI cassette exon of rat *GRIN1* (GRIN1\_CI), exon 8 of *MEN1* (MEN1\_8), and exon 2 of *Hp1bp3* (Hp1bp3\_2). Splicing reporters were expressed in C2C12 cells in the presence of vector backbone (vbb), or with hnRNPA1 or hnRNPH protein expression vector at a ratio of 1:4. Exon-included (double arrowheads) and exon-skipped (single arrowheads) mRNA products were quantified from the gel shown, and used to calculate the percent exon included values (top right).
:::
:::
Next we replaced the test exon of *SIRT1*a with the CI cassette exon of the rat *GRIN1* (*GRIN1*\_CI), exon 8 of *MEN1* (*MEN1*\_8), and exon 2 of *Hp1bp3* (*Hp1bp3*\_2). In the absence of protein co-expression, exon skipping was observed in every case, although the extent of skipping varied over a wide range ([Figure 9](#pbio-0030158-g009){ref-type="fig"}, "Control \[vbb\]"). For the CI cassette exon, hnRNP A1 induced 2.7-fold more skipping, whereas hnRNP H induced 3.2-fold more exon inclusion compared to the control sample ([Figure 9](#pbio-0030158-g009){ref-type="fig"}, compare lanes 3, 8, and 13). Co-expression of hnRNP H increased the inclusion of exon 2 of *Hp1bp3* by a factor of 7.4, but no effects of hnRNP A1 were observed ([Figure 9](#pbio-0030158-g009){ref-type="fig"}, lanes 5, 10, and 15). The latter may have been precluded by the extreme skipping pattern of this exon (0.8% inclusion), which contains three exonic UAGG motifs and a GGGG motif in the 5′ splice site. Thus, for these three exons, the regulation mediated by these hnRNP proteins is specified locally---that is, by sequences limited to the exon and adjacent splice sites. We cannot rule out the possible contributing roles of unknown sequence control elements in splicing silencing. However, sequence alignments show that these exons are highly diverse, and lack shared sequences longer than a few bases.
Co-expression of hnRNP A1 and hnRNP H was also observed to regulate exon 8 of *MEN1,* but with different results. Whereas exon skipping decreased as expected in the presence of hnRNP A1 (74% to 57% exon inclusion), exon skipping decreased to an even greater extent in the presence of hnRNP H (43% exon inclusion), indicating that both of these factors can silence the exon ([Figure 9](#pbio-0030158-g009){ref-type="fig"}, lanes 4, 9, and 14). Because the *MEN1* exon contains two guanosine-rich ASF/SF2 motifs, 5′- GGGAGGA3′ and 5′- AGGAGGG-3′, capable of binding hnRNP H, the observed silencing effect of hnRNP H in this case is not surprising, and is likely explained by the disruption of exon enhancement.
Finally, the results observed for the ESS19 splicing reporter prompted another computational search to determine whether exon skipping is associated with two or more exonic UAGGs genome-wide. Similar to the analysis of [Figure 4](#pbio-0030158-g004){ref-type="fig"}, exons containing two or more conserved UAGGs were identified from a large database (\>94,000) of human and mouse exons and the cDNA/EST-confirmed skipped exons in that group were determined. From this analysis 163 human exons were found to contain two or more exonic UAGGs that are conserved in sequence and position in the orthologous mouse exons, and 16 of these (9.8%) were confirmed skipped exons ([Figure 10](#pbio-0030158-g010){ref-type="fig"}). This was a significant enrichment of exon skipping (*p* \< 0.002) compared to the remaining exons (90,028) lacking UAGGs, of which 4,160 (4.6%) were confirmed skipped exons. When the analysis was repeated for a single UAGG in the exon, a larger number of exons was identified (3,602), but a smaller percentage of confirmed skipped exons, 229 (6.4%), was associated with this group (*p* \< 0.002). The list of 16 human exons with two or more conserved UAGGs and transcript evidence for skipping is shown in [Figure 10](#pbio-0030158-g010){ref-type="fig"}, since these are novel candidates for alternative splicing regulation. Of particular interest are Elongator protein 2, NCOA2, Pumilio homolog 2, and RNA binding protein S1, which are implicated in RNA metabolism.
::: {#pbio-0030158-g010 .fig}
Figure 10
::: {.caption}
###### Computational Analysis of Exonic UAGG Motifs and Exon Skipping Patterns Genome-Wide
Computational searches were performed to identify exons with two or more UAGGs and to determine the association of confirmed exon skipping events with this group. Exons with a single UAGG were analyzed for comparison. The following constraints were applied: (1) exon lengths of 250 bases or fewer and (2) both UAGG motifs conserved in sequence and position in the orthologous mouse exons. The graph illustrates the percentage of confirmed exon skipping events associated with one UAGG or two or more UAGGs (blue bars), or with the remaining exons lacking these motifs (red bars). The list of 16 human exons identified with two or more UAGGs is shown with the Ensembl ID, exon number, 5′ splice site sequence, and gene name. It is not unexpected to find exon 19 of the glutamate NMDA receptor *GRIN1* and exon 13 of *NCOA2,* which have a 5′-splice-site-proximal GGGG, since the sequence of the 5′ splice site was not specified in the search.
:::
:::
Discussion {#s3}
==========
A Combinatorial Code for Exon Silencing {#s3a}
---------------------------------------
Here we use molecular approaches to define a ternary combination of UAGG and GGGG motifs required for silencing the *GRIN1* CI cassette exon, and show that a class of skipped exons in the human and mouse genomes can be identified through bioinformatics searches that maintain the sequence and spatial configuration of the silencing motifs. We also illustrate, using the CI cassette model system, how the combined sequence motifs work cooperatively to determine the strength of exon silencing, with similar trends in neuronal and non-neuronal cell types. While a single exonic UAGG or 5′-splice-site-proximal GGGG motif specifies weak exon skipping, multiple UAGGs in the exon together with the GGGG motif at the 5′ splice site specifies predominant exon skipping (see [Figure 2](#pbio-0030158-g002){ref-type="fig"}). This conclusion is strengthened by the complementary results observed when these RNA signals are systematically disrupted in a skipped exon or introduced into a constitutive exon. The CI cassette exon is converted into a constitutive exon by interrupting all three components of the motif pattern, whereas strong exon skipping results when the same components are introduced into constitutive exon 6 of the human *SIRT1* splicing reporter. In both contexts, hnRNP A1 co-expression mediates silencing and hnRNP H mediates anti-silencing in concert with all three components of the motif pattern ([Figure 11](#pbio-0030158-g011){ref-type="fig"}).
::: {#pbio-0030158-g011 .fig}
Figure 11
::: {.caption}
###### Model for Differential Regulation of the CI Cassette Exon by the Interplay of hnRNP A1 and H and a Ternary Motif Pattern
At the top is a schematic of intron/exon structure and prominent splicing patterns observed in the forebrain (top) and hindbrain (bottom) of rat brain. Below is a summary of splicing regulatory motifs functionally defined in this study depicted on an expanded version of the *GRIN1* CI cassette exon (yellow). ESEs are indicated above the exon. Nucleotides complementary to U1 small nuclear RNA and the interaction of the positive regulator NAPOR/CUGBP2 with the downstream intron are indicated (‡; as determined in \[[@pbio-0030158-b26]\]). UAGG and GGGG splicing silencing motifs defined in this study are highlighted in red. The working model for splicing silencing, based on the results shown here, proposes that the CI cassette is a strong exon silenced by a combination of two exonic UAGG motifs and a 5′-splice-site-proximal GGGG. HnRNP A1 mediates silencing and hnRNP H mediates anti-silencing via these RNA signals.
:::
:::
In this study, bioinformatics searches show that the combination of exonic UAGG and 5′-splice-site-proximal GGGG motifs is relatively rare, since only 0.2% of a large database of human and mouse exons (approximately 200 out of approximately 96,000) harbor UAGG and GGGG motifs together in the correct arrangement. Nonetheless, based on cDNA and EST evidence a significantly higher frequency of exon skipping is associated with the set of 16 exons in which the motif pattern (≥1 exonic UAGGs and a 5′-splice-site-proximal GGGG) is conserved in the human and mouse orthologs (see [Figure 4](#pbio-0030158-g004){ref-type="fig"}). For 14 of the newly identified exons we experimentally determined a rate of approximately 57% exon skipping based on RT-PCR analysis in a variety of human and mouse tissues (eight of 14, not counting the CI cassette). We would expect an imperfect correlation between the presence of the motif pattern and confirmed exon skipping, since the approximately 8--10 exonic enhancer motifs in a typical 140 base exon \[[@pbio-0030158-b13],[@pbio-0030158-b15]\] may override the effects of UAGG and GGGG silencer motifs. This may be due not only to the arrangement of ESE and intronic splicing enhancer motifs in and around a target exon, but also to tissue-specific variations in splicing factors. Evidence was also shown for an increased association of confirmed exon skipping events genome-wide with the presence of two or more conserved exonic UAGGs, as a variation of the original motif pattern. Our functional analysis showed that the presence of multiple UAGGs in the same exon was an important parameter for a predominant exon skipping pattern.
The question of the relative 5′ splice site strengths of those exons containing or lacking the UAGG and GGGG motif pattern was also addressed. When the relative splice site strengths of the two groups were compared using a rank sum statistical test, no significant difference in the distributions was found. In fact the median score for splice site strength was found to be higher (9.31) for the group of exons containing the motifs than for those without (8.68). The close proximity of the motifs to, or their overlap with, the 5′ splice site, however, remains an unresolved issue. While a detectable effect of the (intronic) position of the GGGG motif was observed in the context of the *SIRT1* splicing reporter, the general rules for such position effects were not determined. GGGG or UAGG motifs in the 5′ splice site region may interfere with base pairing interactions involving U1 and/or U6 small nuclear RNAs, and these effects may have a high degree of position dependence.
Numerous ESE motifs have been functionally identified in concert with the regulatory roles of SR proteins, but far less is known about sequence motifs and factors that control silencing. Evidence for exonic UAG and UAGG motifs has been previously reported for splicing silencing mechanisms mediated by hnRNP A1. These include the K-SAM exon of human *FGFR2* \[[@pbio-0030158-b35]\], *SMN2* exon 7 (U AGACA) \[[@pbio-0030158-b36]\], HIV *Tat* exon 2 (U AGACU) \[[@pbio-0030158-b37],[@pbio-0030158-b38]\], *CD44* exon v5 (U AGACA) \[[@pbio-0030158-b39]\], *protein 4.1* exon 16 \[[@pbio-0030158-b40]\], *c-src* exon N1 (UAG: GAGGAAGGU) \[[@pbio-0030158-b41]\], and exons in the hnRNP A1 transcript itself (UAG and U AGAGU) \[[@pbio-0030158-b24],[@pbio-0030158-b42]\]. Taken together with structural evidence that hnRNP A1 recognizes TAGG motifs directly \[[@pbio-0030158-b43]\], A1 is a likely mediator of many if not all of these silencing events. In contrast to the previous studies, however, the 5′-splice-site-proximal GGGG motif is a novel and integral component of the silencing mechanism of the CI cassette exon. While the silencing effect of the GGGG motif by itself is slight, its function with exonic UAGGs is synergistic. Our computational analysis using the CodonShuffle algorithm extends these previous studies by showing genome-wide that the UAGG motif is significantly underrepresented in constitutive exons and overrepresented in skipped exons. Because the CodonShuffle analysis forbids in-frame UAG stop codons, these results are in good agreement with the idea that exonic UAGG motifs function widely as splicing silencers.
In previous studies guanosine-rich motifs have been shown to regulate splicing in diverse ways. Guanosine triplets are generally enriched in short mammalian introns \[[@pbio-0030158-b44],[@pbio-0030158-b45]\], and these sequences have been shown to enhance inclusion of an unusually small exon of *cardiac troponin T* \[[@pbio-0030158-b46],[@pbio-0030158-b47]\], as well as additional exons of human *α-globin* \[[@pbio-0030158-b48]\] and chicken *β-tropomyosin* \[[@pbio-0030158-b49]\], transcripts. Moreover, a disease-related point mutation in a guanosine cluster at position 26 of the intron has been shown to disrupt the normal pattern of splicing of the human *pyruvate dehydrogenase E1α* transcript \[[@pbio-0030158-b50]\]. In some cases, hnRNP H has been implicated in splicing control together with guanosine-rich sequences. A guanosine-rich ESS in *β-tropomyosin* exon 7 is required for exon skipping, and the degree of hnRNP H binding correlates with exon 7 skipping \[[@pbio-0030158-b51]\]. The *c-src* transcript contains a complex intronic enhancer downstream of the neuron-specific NI exon in which multiple guanosine-rich tracts are found that bind to hnRNP H and F and that are required for normal patterns of NI exon inclusion \[[@pbio-0030158-b52],[@pbio-0030158-b53],[@pbio-0030158-b54]\]. In addition, hnRNP H has been shown to bind to the 5′ splice site of *NF-1* exon 3, where it is thought to induce exon skipping when the splice site is weakened by a guanosine to cytosine mutation at position +5 of the intron \[[@pbio-0030158-b55]\]. In this study we show that hnRNP H has a positive effect on exon inclusion for three unrelated exons harboring the UAGG and GGGG motif pattern in the context of a heterologous splicing reporter. Because these exons have no other sequence relatedness, these results suggest that antagonism with hnRNP A1 might be a frequent property of hnRNP H in this type of silencing mechanism (see [Figure 9](#pbio-0030158-g009){ref-type="fig"}).
Model for Splicing Regulation Mediated by a UAGG and GGGG Code: Differential Roles of hnRNP A1 and H {#s3b}
----------------------------------------------------------------------------------------------------
A full understanding of CI cassette exon regulation will require explanations for the complex spatial and temporal variations observed in vivo. Based on functional evidence, we proposed in a previous study that NAPOR/CUGBP2 enhances CI exon inclusion in the rat forebrain, where its expression is enriched. It would be reasonable to predict, however, that the CI cassette exon is inherently a strong exon and should not require a positive regulator, since its splice sites match well to consensus sequences. Here we confirmed this prediction by experimental manipulations of the UAGG and GGGG motif pattern that converted the CI cassette exon into a constitutive exon in the absence of NAPOR/CUGBP2 (splicing reporter T8; see [Figure 2](#pbio-0030158-g002){ref-type="fig"}). These results clearly demonstrate differential roles of hnRNP A1 and H proteins. Furthermore, we showed that the mechanism by which these proteins regulate the CI cassette can be controlled locally through sequences in the exon and adjacent 5′ splice site independent of any distal downstream intron sequences from the *GRIN1* transcript.
Six ESE motifs within the CI cassette exon were functionally identified in this study, and a seventh, an ASF/SF2 motif, overlaps with the exon position 93 UAGG silencer (see [Figure 11](#pbio-0030158-g011){ref-type="fig"}). Predominant exon skipping was retained even when both of the natural UAGGs were carefully repositioned in the exon without destroying or creating any known ESEs. That is, in two distinct cell lines, the six functional ESE motifs did not overpower the silencing function of the three-part UAGG and GGGG code. We observed that UAGG motifs are embedded in 32 ESE motifs reported in the ESEFinder database \[[@pbio-0030158-b56]\], suggesting that the occurrence of overlapping ESE and ESS signals might be quite frequent. In the case of the CI cassette exon, such an arrangement of opposing splicing signals would predict that competition between ASF/SF2 and hnRNP A1 may provide additional options to fine-tune splicing patterns in different tissues or stages of development. However, in comparison to hnRNP H, the co-expression of an ASF/SF2 expression plasmid had only a mild positive effect on exon inclusion in the cell lines tested (K. H. and P. J. G., unpublished data).
Here we show evidence for combinatorial regulation by two different types of RNA elements (UAGG and GGGG) together with differential roles of hnRNP A1 and H (and F), but not all of the combinatorial interactions were experimentally defined. Although the intronic GGGG motif and A1 are involved in silencing, site-specific UV crosslinking of A1 to the GGGG motif was not observed (K. H. and P. J. G., unpublished data). This may be due to limitations of the assay, since UV crosslinking of A1 to its high-affinity site is inefficient \[[@pbio-0030158-b30]\]. Alternatively, the intronic GGGG may play a structural role, or contact an additional protein factor involved in the assembly of the putative silencing complex. We speculate that a silencing complex is formed by the interactions of hnRNP A1 monomers with individual UAGG and GGGG sites together with cooperative interactions between these monomers. We also speculate that hnRNP H and, to a lesser extent, F function principally as anti-silencing factors in the CI cassette mechanism by binding to the GGGG and/or UAGG motifs in a way that disrupts the cooperative binding of A1. In our view this is the simplest model to account for our experimental results, but more complex mechanisms cannot be ruled out at this point. Future studies will be required to establish how the various isoforms of hnRNP H carry out anti-silencing, and whether accessory factors are involved.
Substantial evidence exists in support of models involving competition between hnRNP A1 and SR proteins in modulating 5′ splice site selection or exon inclusion \[[@pbio-0030158-b24],[@pbio-0030158-b57],[@pbio-0030158-b58],[@pbio-0030158-b59],[@pbio-0030158-b60],[@pbio-0030158-b61]\]. The involvement of hnRNP A1 in the CI cassette mechanism is also consistent with previous demonstrations of the cooperative binding of hnRNP A1 to pre-mRNAs \[[@pbio-0030158-b62],[@pbio-0030158-b63],[@pbio-0030158-b64],[@pbio-0030158-b65]\]. Based on the analysis of microarray data \[[@pbio-0030158-b66],[@pbio-0030158-b67]\] documenting considerable variations in the ratios of hnRNP A1 transcripts to hnRNP F and H transcripts in human and mouse \[[@pbio-0030158-b33]\], we suggest that such variations may be involved in directing tissue specificity of exons that are regulated by UAGG and GGGG motifs.
Implications of Genome-Wide Analysis {#s3c}
------------------------------------
Since the CI cassette exon skipping pattern of the *GRIN1* transcript is brain-region-specific, we wished to determine the splicing characteristics of other exons with a similar arrangement of these motifs in the human and mouse genomes. Other transcripts harboring skipped exons that were identified by bioinformatics searches, however, were found to be involved in a variety of cellular functions, such as RNA processing, chromatin structure/function, cell signaling, and regulation of transcription. These include hnRNP H1 and H3 *(HNRPH1* and *HNRPH3),* menin *(MEN1),* nuclear receptor co-activator 2 *(NCOA2),* heterochromatin protein 1 binding protein 3 *(Hp1bp3),* and an uncharacterized hypothalamus transcript ([Table 1](#pbio-0030158-t101){ref-type="table"}). A high proportion of the exon skipping patterns identified were found to be tissue-specific.
The observation that exon 5 of *HNRPH1* and exon 3 of *HNRPH3* contain conserved UAGG and GGGG motifs is intriguing, since hnRNP H proteins crosslink specifically to the GGGG motif adjacent to the CI cassette exon. These exon skipping patterns were confirmed by RT-PCR analysis in this study, and there is additional supporting cDNA and EST evidence in the databases. The RT-PCR analysis shows that these exon skipping patterns are relatively weak, but this is consistent with a motif pattern containing a single exonic UAGG and 5′ splice site GGGG motif. Skipping of exon 5 of *HNRPH1* or exon 3 of *HNRPH3* would result in a shift in the reading frame and introduction of a premature termination codon. Thus, silencing of these exons at the level of splicing is expected to reduce protein expression via either nonsense-mediated mRNA decay or premature termination of protein synthesis. The results shown here suggest a model in which hnRNP H proteins may provide a buffering effect against negative control by hnRNP A1. Autoregulation by a negative feedback loop was recently demonstrated for the splicing factor PTB, which induces skipping of the 11th exon of its cognate pre-mRNA \[[@pbio-0030158-b68]\]. Similarly, hnRNP A1, SRp20, SC35, TIA1, and TIAR proteins are all involved in mechanisms that regulate the splicing patterns of their cognate transcripts \[[@pbio-0030158-b69],[@pbio-0030158-b70]\].
Prospects {#s3d}
---------
If alternative splicing events are as prevalent as recent studies suggest \[[@pbio-0030158-b21],[@pbio-0030158-b22],[@pbio-0030158-b71],[@pbio-0030158-b72]\], it will be important to understand on a global scale the biochemical language that determines tissue-specific patterns, and tunes these patterns in response to physiological stimuli \[[@pbio-0030158-b73],[@pbio-0030158-b74]\]. Here we show that UAGG and GGGG motifs function in combination to silence the CI cassette exon and also serve more generally as patterns to recognize other skipped exons in the human and mouse genomes. Combinatorial splicing control mechanisms are not well understood, and previous studies have not addressed the brain-region-specific splicing switch that is characteristic of the CI cassette exon. Our results suggest that, in general, it might be a useful strategy to use motif pattern searches, together with information about spatial constraints, to identify co-regulated exons. The observation that UAGG and GGGG motif patterns are generally predictive of exon skipping may also be useful in interpreting the effects of mutations underlying certain genetic diseases. Future work will be needed to more fully understand the roles of hnRNP proteins in this type of silencing (and anti-silencing) mechanism, and to further advance the understanding of the complex biochemical language responsible for the regulation and coordination of splicing events genome-wide.
Materials and Methods {#s4}
=====================
{#s4a}
### Plasmid construction and mutagenesis. {#s4a1}
All splicing reporter plasmids except for those in the experiments of [Figure 3](#pbio-0030158-g003){ref-type="fig"}D were derived from the parent plasmid wt (previously called E21wt), in which the CI cassette exon is flanked by full-length introns and adjacent exons \[[@pbio-0030158-b26]\]. Site-directed mutations were introduced into the CI cassette exon or downstream intron using the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, California, United States), and mutations were confirmed by DNA sequencing. The splicing reporters wt and wt0 are identical except that wt has a point mutation at position 78 (C to G change) of the CI exon, which creates a XhoI site. Chimeric splicing reporters were derived from parent plasmid rGγ25 \[[@pbio-0030158-b75]\], in which the CI cassette exon and 164 and 103 bp of the flanking introns (upstream and downstream, respectively) were introduced as a NotI-BamHI fragment. The full-length upstream intron was introduced by replacing the XbaI-NotI fragment of rGγCI-wt0 with the XbaI-NotI PCR product containing *GRIN1* exon 18 and 1,092 bp of adjacent intron (plasmid, rGγCI-up). The full-length downstream intron was introduced by replacing the BamHI-EcoRI fragment of rGγCI-wt0 with the BamHI-EcoRI fragment containing *GRIN1* exon 20 and 1,810 bp of adjacent upstream intron. All splicing reporter plasmids were constructed in a pBS vector followed by transfer into the vector pBPSVPA+ \[[@pbio-0030158-b76]\], in which expression is driven by the SV40 promoter. Expression plasmids for hnRNP proteins F, H, and A1 were generated by subcloning the complete open reading frames into the BamHI site of pcDNA4/HisMax vector (Invitrogen, Carlsbad, California, United States). Open reading frames were obtained from the following plasmids: hnRNP F from plasmid pFlag-F \[[@pbio-0030158-b53]\], hnRNP H/H′ from pFlag-DSEF-1 \[[@pbio-0030158-b77],[@pbio-0030158-b78]\], and hnRNP A1 from plasmid Myc-A1 \[[@pbio-0030158-b79]\]. All plasmid constructs were confirmed by DNA sequencing, and protein expression was verified by Western blot analysis.
### Transient expression and analysis of RNA splicing patterns. {#s4a2}
Growth of C2C12 cells, transfection, and RT-PCR analysis were performed as described \[[@pbio-0030158-b26]\]. Briefly, transfections were performed in 60-mm plates at approximately 70% cell confluency using Lipofectamine (Invitrogen). Transfections contained 3.5 μg of total plasmid DNA made up of splicing reporter plasmid with empty vector and/or protein expression plasmid at the DNA ratios specified. PC12 cells were grown to approximately 85% confluency in RPMI1640 supplemented with 10% fetal bovine serum and 5% horse serum on poly-D-lysine-coated six-well plates. PC12 cell transfections were carried out with Lipofectamine 2000 and a total of 1.25 μg of plasmid DNA (0.25 μg of splicing reporter and 1 μg of protein expression plasmid or vector backbone). After 48 h, cells (C2C12 and PC12) were harvested and total RNA was purified, DNase I treated, and ethanol precipitated. For analysis of splicing patterns, 1 μg of RNA was reverse transcribed with random hexanucleotide primers, and 1/20th of the reaction volume was then amplified for 20--24 PCR cycles in a 10-μl reaction containing 0.2 μM specific primers, two units of Taq polymerase, 0.2 mM dNTPs, and 1 μCi of \[α^32^P\]dCTP in reaction buffer. Under these conditions approximately 1% of the C residues in each product molecule are radiolabeled. Primers used to amplify the CI-cassette-exon-included and -skipped mRNA products were specific for the flanking exons. Sequences from Ensembl were used to design primers for the experiments of [Figure 4](#pbio-0030158-g004){ref-type="fig"}. Primer sequences are available upon request. For gel analysis, 25% v/v of each PCR reaction was resolved on 6% polyacrylamide/5 M urea sequencing gels. Electrophoresis was performed for 1 h at 30 W. Gel images were obtained and results quantitated using a Fuji (Tokyo, Japan) Medical Systems BAS-2500 phosphorimager and Science 2003 ImageGauge software. For the experiment of [Figure 6](#pbio-0030158-g006){ref-type="fig"}, PCR reaction products were resolved on 1% agarose gels in 1× TBE buffer.
### Transcription and site-specific RNA labeling. {#s4a3}
Radioactive RNA substrates were prepared for UV crosslinking analysis as follows. RNAs containing the GGGG motif were prepared by in vitro transcription in 25-μl reactions containing T7 RNA polymerase, 0.4 mM each of ATP, UTP, and CTP, and 0.3 mM GTP plus 25 μCi of \[α^32^P\]GTP, 0.5 mM GpppG, and 0.1 μg of DNA template in standard T7 reaction buffer. DNA templates were prepared by annealing complementary oligonucleotides with the top strand containing the T7 promoter sequence at its 5′ end, followed by the RNA test sequence; bottom strands were complementary to the test sequence. RNAs were purified after DNase treatment by Sephadex G25 chromatography, phenol extraction, and ethanol precipitation. Site-specific labeling of RNA substrates containing the exonic UAGG motif was performed essentially as described \[[@pbio-0030158-b80]\]. Transcription (nonradioactive) of the downstream RNA half was performed as above except that reactions were larger (125 μl) and contained 2 mM guanosine instead of GpppG. After gel purification, the 5′ end of the downstream-half RNA was labeled by polynucleotide kinase with 25 pmol of the purified RNA and 25 pmol of \[γ^32^P\]ATP (6,000 Ci/mmol). After removal of ATP by Sephadex G25 chromatography, the upstream and downstream RNA halves were annealed to a complementary DNA splint covering 16 bases on either side of the desired ligation position. Ligation reactions were performed in 10-μl reactions with 15 Weiss Units of T4 DNA ligase for 4 h at 16 °C, followed by DNase treatment and gel purification. The concentrations and integrity of the RNA preparations were verified by electrophoresis on 10% polyacrylamide/7M urea gels.
### UV crosslinking and immunoprecipitation analysis. {#s4a4}
UV crosslinking reactions (12.5 μl) were performed under splicing conditions as described \[[@pbio-0030158-b81]\] with 100,000 dpm radiolabeled RNA transcript and HeLa nuclear extract (4 mg/ml final concentration). Following UV treatment, samples were digested to completion with RNase A (1 mg/ml, 20 min at 30 °C), and held on ice for immunoprecipitation or SDS-PAGE analysis. For immunoprecipitation reactions, 25 μl of protein A beads (Sigma, St. Louis, Missouri, United States) were equilibrated in Buffer A (10 mM Tris/HCl \[pH 7.5\], 100 mM NaCl, and 1% TritonX100), and antibody was bound to the beads for 1 h on ice (5 μl of R7263 or R7264 for analysis of hnRNP F and H, respectively \[[@pbio-0030158-b82]\], or 1 μl of 9H10 for analysis of hnRNP A1). Equivalent concentrations of rabbit preimmune serum or purified mouse IgG were used for control reactions. Antibody beads were washed three times with Buffer A, and added to UV crosslinking reactions (25 μl) for 20 min on ice. Bound samples were washed four times with Buffer A, and centrifuged to separate pellet and supernatant. Each reaction component was boiled in SDS sample buffer, and resolved on discontinuous 12.5% polyacrylamide gels.
### Generation of datasets and computational analysis. {#s4a5}
Human and mouse genes that were annotated as orthologs were obtained from Ensembl release 16 (<http://www.ensembl.org>). Human--mouse exons were aligned by BLAST (requiring percent identity ≥85 and bit score ≥20), and genes were checked for consistency in terms of orthologous exon order. A total of approximately 94,000 conserved human--mouse exons were retained for further analysis (<http://genes.mit.edu/burgelab/Supplementary/han04>). In a separate analysis, approximately 14,600 internal exons from human genes were designated as skipped exons based on stringent alignments of cDNA and EST sequences to cDNA-verified genomic loci using the genome annotation script GENOA (<http://genes.mit.edu/genoa>). Mapping these exons to the conserved human--mouse Ensembl set identified 4,455 skipped internal human exons that are conserved in mouse. For the codon shuffling analysis, the first 30 bases and the last 60 bases of the original sequences were removed prior to shuffling to simulate removal of the first and last exons. Each sequence was shuffled 50 times using the CodonShuffle program \[[@pbio-0030158-b32]\]. The number of occurrences of each oligonucleotide, e.g., UAGG, divided by the number of occurrences of all possible oligonucleotides of equal length, was compared to the corresponding frequency of occurrence in the shuffled sets. The final fold underrepresentation was computed by taking the mean of the fractions computed over the shuffled sets, and dividing by the observed (true) fraction. The *p*-value for the reduced occurrence of UAGG in authentic coding sequences was determined by counting the number of 4-mers that were greater than 1.488-fold reduced relative to the average of 100 shuffles. None were found for each of the ten shuffles. Thus the *p*-value is 0/256, or *p* \< 0.001.
We thank members of the Grabowski and Burge labs for helpful discussions and critical reading of the manuscript. We gratefully acknowledge Gideon Dreyfuss, Christine Milcarek, and Zefeng Wang for providing antibody and plasmid reagents. This work was supported by a grant from the National Institutes of Health to PJG (GM068584). GY was supported by the Lee Kuan Yew fellowship from Singapore. Support from the Howard Hughes Medical Institute for the initial stages of this project (PJG) is also acknowledged.
**Competing interests.** The authors have declared that no competing interests exist.
**Author contributions.** CBB and PJG conceived and designed the experiments. KH, GY, and PA performed the experiments. KH, GY, PA, CBB, and PJG analyzed the data, contributed reagents/materials/analysis tools, and wrote the paper.
¤a Current address: University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, United States of America
¤b Current address: Crick-Jacobs Center for Computational and Theoretical Biology, Salk Institute, La Jolla, California, United States of America
Citation: Han K, Yeo G, An P, Burge CB, Grabowski PJ (2005) A combinatorial code for splicing silencing: UAGG and GGGG motifs. PLoS Biol 3(5): e158.
ESE
: exonic splicing enhancer
ESS
: exonic splicing silencer
SR
: serine-arginine-rich
hnRNP
: heteronuclear ribonucleoprotein
|
PubMed Central
|
2024-06-05T03:55:55.132977
|
2005-4-19
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079783/",
"journal": "PLoS Biol. 2005 May 19; 3(5):e158",
"authors": [
{
"first": "Kyoungha",
"last": "Han"
},
{
"first": "Gene",
"last": "Yeo"
},
{
"first": "Ping",
"last": "An"
},
{
"first": "Christopher B",
"last": "Burge"
},
{
"first": "Paula J",
"last": "Grabowski"
}
]
}
|
PMC1079784
|
Sleep is absolutely essential for well-being. Just ask one of the 40 million Americans with sleep disorders who suffer crippling fatigue, impaired judgment, irritability, moodiness, and myriad health problems. Still, its precise function remains unclear. An intriguing role for REM sleep---the stage most closely associated with dreaming---was suggested almost 40 years ago when sleep researchers Howard Roffwarg and William Dement discovered that babies spend far more time in REM sleep than adults---prompting their hypothesis that infant REM sleep plays a role in central nervous system development.
A central element of their hypothesis revolves around the nature of infant sleep and whether the neural mechanisms of infant sleep differ significantly from those of adult sleep. Infant rats, like the offspring of other "altricial" species (born naked, helpless, and blind), spend most of their time in what\'s now called active sleep, indicated by intermittent muscle twitching and low muscle tone (atonia)---behaviors characteristic of adult REM sleep. At issue is whether infant mechanisms are primitive, undifferentiated, and distinct from adult mechanisms or whether they contain elementary components that are integrated into the developing sleep system.
In a new study, Karl Karlsson, Mark Blumberg, and their colleagues tackle the technical difficulties involved in studying the tiny neonatal brain to investigate the neural activity associated with infant sleep states. The active sleep of week-old rats, they show, bears a striking resemblance to the conventional definitions of adult sleep. What\'s more, the neural mechanisms underlying the infant sleep state contain the primary components of adult sleep.
In previous studies, Karlsson and Blumberg discovered a brainstem region in the ventromedial medulla, which they called the medullary inhibitory area (MIA), that appears functionally equivalent to the region that generates REM atonia in adults. They also found that the MIA doesn\'t generate infant sleep on its own but depends on a network that spans both lower brainstem and midbrain regions. In this study, the authors set out to identify the neural structures that project to the MIA and better characterize the network.[](#pbio-0030168-g001){ref-type="fig"}
::: {#pbio-0030168-g001 .fig}
::: {.caption}
###### Though naked, helpless, and blind, this week-old rat (pictured with a quarter) already has the fundamental neural components of adult sleep (Photo: Mark Blumberg)
:::
:::
Karlsson et al. first established that there are neurons that connect to the MIA from areas in the medulla and pons. Then, by recording from neurons in these areas, they found neurons that are active only during sleep or wakefulness and that appear to control muscle tone and twitching. Neurons active mostly during atonia---indicating sleep---concentrated in the subcoeruleus (SubLC) region of the pons; those active mostly during wakefulness clustered in an area within the dorsolateral pontine tegmentum (DLPT) in the midbrain. The authors went on to link different sets of neurons with specific behaviors and brain regions. A group of neurons within the DLPT, for example, showed distinct bursts of activity just before muscle twitching. And a subset of SubLC neurons fired at much higher rates when atonia was accompanied by bouts of tail and neck muscle twitching.
Introducing lesions in the SubLC and another pontine nucleus, called the pontis oralis, caused significant changes in muscle tone and twitching. Lesions in the two pontine nuclei reduced periods of atonia but not the number of muscle twitches---in effect decoupling the key components of REM sleep, twitching and atonia. Lesions in the DLPT had the opposite effect: increased atonia and significantly less muscle twitching.
Altogether, the authors argue, these results show that sleep development elaborates on elementary components already in place soon after birth. If the neural mechanisms of infant and adult sleep were entirely different, then sleep might serve different purposes in infancy and adulthood. But the striking parallels outlined in this study suggest a developmental continuity between the two states. They also chart a course for future study that might even test Roffwarg\'s view that the neonatal brainstem primes the central nervous system for the sensory challenges that lie ahead---and could even be the stuff that dreams are made of.
|
PubMed Central
|
2024-06-05T03:55:55.140611
|
2005-4-19
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079784/",
"journal": "PLoS Biol. 2005 May 19; 3(5):e168",
"authors": []
}
|
PMC1079785
|
Our cells make many more kinds of proteins than can be accounted for by the relatively modest number of genes in our genome. The key to this protein-coding bounty is alternative splicing, in which one or more transcribed exons---nucleotide sequences that code for a specific segment of the protein---are excluded from the final messenger RNA before it is translated into protein. While the majority of human genes are alternatively spliced, little is known about specific RNA sequences that dictate exclusion of these exons. In a new study, Paula Grabowski and colleagues show that three short sequences, two within the excluded exon and one in an adjacent intron, or non-coding nucleotide sequence, trigger exclusion in at least one gene, and probably a large handful of others as well.
Grabowski and colleagues studied this process in a class of proteins essential for brain function called glutamate receptors. As the name implies the glutamate receptors bind to glutamate, the principal excitatory neurotransmitter in the brain. NMDA glutamate receptors, which play a role in memory formation and neuronal development, are composed of multiple subunits. Within the NR1 subunit, exclusion or inclusion of the CI cassette exon has dramatic functional consequences. The CI exon appears in the forebrain but is virtually absent in the hindbrain. How this differential splicing is regulated is poorly understood.
The authors noted an atypical but highly conserved GGGG motif in the intron just downstream from the splice site that ends the CI exon. When they introduced point mutations in this motif, the exon was included up to four times as often. The rate of exclusion, or silencing, could be dramatically increased by the addition of another GGGG tetrad farther inside the intron.
Systematic mutation within the exon identified a pair of UAGG motifs that also promoted exon silencing, an effect that could be enhanced even further by introducing a third, artificial, UAGG. The pair of UAGG tetrads appears to work in combination with the GGGG tetrad, since without the former sequences, the latter had little power to silence CI expression. Silencing is mediated by binding of UAGG to the ribonucleoprotein hnRNP A1, which also apparently interacts with the GGGG within the intron.
The authors next did a series of genomic database searches, to identify these motifs in other genes. They reasoned that if the triad was a common means of exon silencing, it should be overrepresented among genes known to undergo alternative splicing. In more than 90,000 exons in human and mouse genomes, they found 16 with the motif pattern, of which three (19%) were known skipped exons. In contrast, among those without the pattern, the proportion of skipped exons was only 5%. They also found that the GGGG motif by itself was overrepresented among skipped exons, indicating it probably plays a significant role in exon exclusion even without its UAGG partners.
These results alone cannot explain why one cell type includes an exon while another excludes it, since the primary transcript in different cell types is the same. Instead, these differences are likely explained by tissue-specific differences in levels of splicing factors or binding proteins. With such small absolute gene numbers, it is clear that the specific trio identified by Grabowski and colleagues is only one of many likely to regulate exon inclusion. In the search for others, this study indicates the value of bioinformatics strategies that employ not only specific sequences, but also spatial configurations.
|
PubMed Central
|
2024-06-05T03:55:55.141331
|
2005-4-19
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079785/",
"journal": "PLoS Biol. 2005 May 19; 3(5):e173",
"authors": []
}
|
PMC1079786
|
Cyanobacteria have a long and checkered past. When their ancestors first appeared some 3 billion years ago, earth\'s atmosphere likely contained mostly carbon dioxide, along with hydrogen sulfide, ammonia, nitrogen, and water vapor. Thought to be the first photosynthesizers, cyanobacteria forebears used water from their ocean habitat, carbon dioxide, and sunlight to make sugar, and produced oxygen as waste---the kiss of death for most ancient microorganisms, which eventually died from oxygen poisoning.
Modern cyanobacteria continue to exert disproportionate influence for their size. The Prochlorococcus group of cyanobacteria---which measure in at less than a micron in diameter, allowing 500-plus individuals to fit comfortably on the head of a pin---account for a significant fraction of global photosynthesis by virtue of their ubiquitous presence in nutrient-depleted ocean waters. Even tinier agents---the viruses that infect these bacteria, called cyanophages---appear capable of wielding equally surprising influence on global cycles by affecting the population dynamics and evolutionary path of Prochlorococcus.
To better understand the nature of virus--host interactions at sea, Sallie Chisholm and colleagues investigated the genetic makeup of three cyanophages. The marine phages resemble two terrestrial phages---called T4 and T7---that infect Escherichia coli but carry genes that appear specially adapted to infecting photosynthetic bacteria in nutrient-poor oceans.
Of over 430 completed phage genomes, only one (P60) infects cyanobacteria. Since marine phages likely face different selection pressures than their terrestrial equivalents, the authors explain, genome analysis can shed light on the agents of selection, besides providing a survey of marine phage types. Chisholm and colleagues chose to sequence three marine phages---one podovirus (P-SSP7) and two myoviruses (P-SSM2 and P-SSM4)---based on their morphology and host range, and characterized their genomes. The P-SSP7 virus has genes that closely match many of T7\'s so-called core genes---signature genes required for that virus\'s mode of infection, which involves killing its host. P-SSP7 also has the same genome structure as other T7-like phages, though it appears capable of coexisting with its host (based on the presence of an integrating enzyme) while T7 kills as it infects. Chisholm and colleagues go on to characterize the two myoviruses and find that both viruses share most of the core genes found in T4-like phages. And like T4 phages, both myoviruses lack the integrating enzymes, suggesting they share T4 phages\' homicidal approach to infection.[](#pbio-0030184-g001){ref-type="fig"}
::: {#pbio-0030184-g001 .fig}
::: {.caption}
###### Despite a striking resemblance to an E. coli virus, this marine virus appears to have evolved genes adapted to infecting photosynthetic bacteria inhabiting low-nutrient oceans (Scale bar indicates 100 nm) (Photo: Peter Weigele)
:::
:::
Beyond the core phage genes, Chisholm and colleagues also present a survey of genes likely derived from cyanobacteria that "could play defining functional roles" in marine phage--host interactions. All three cyanophages contain photosynthesis-related genes, some of which, the authors propose, may mean the virus helps the host maintain photosynthesis during infection. The podovirus also has a candidate gene involved in DNA synthesis, which the authors speculate might help the virus reproduce in nutrient-poor environments, and all three cyanophages carry genes involved in metabolizing carbon. The absence of such genes in terrestrial phages, the authors argue, lends support to the notion that marine phages have evolved different adaptive mechanisms in response to the ocean environment.
Given the intimate relation between virus and host, the effects of gene swapping between virus and host is likely to be a two-way street. Just as cyanophages may help shape the fate of their hosts, it\'s likely that cyanobacterial genes influence phage ecology and perhaps even its range. The cyanophages characterized here take after two phages that were central to many fundamental breakthroughs in molecular biology, including the discovery that genes are made of DNA. It remains to be seen how the marine versions of these legendary laboratory viruses contribute to our understanding of phage infections in one of the most abundant, ecologically diverse primary producers in the open seas. See also the related Primer "The Third Age of Phage" (DOI: [10.1371/journal.pbio.0030182](10.1371/journal.pbio.0030182)).
|
PubMed Central
|
2024-06-05T03:55:55.141904
|
2005-4-19
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079786/",
"journal": "PLoS Biol. 2005 May 19; 3(5):e184",
"authors": []
}
|
PMC1079787
|
Introduction {#s1}
============
The addition of a tract of poly(A) to the 3′ ends of most eukaryotic messenger RNA precursors (pre-mRNAs) is an important step in their processing. Poly(A) tails have multiple functions: they facilitate the export of mRNAs from the nucleus, increase the efficiency of translation initiation, and stabilize the mRNAs in the cytoplasm. 3′ End processing is carried out by a multiprotein machinery and is coupled to transcription by RNA polymerase II and to the other RNA-processing reactions \[[@pbio-0030189-b01]--[@pbio-0030189-b06]\]. Polyadenylation is catalyzed by poly(A) polymerases (PAPs) that belong to a family of related nucleotidyltransferases \[[@pbio-0030189-b07]--[@pbio-0030189-b09]\]. Although they can be isolated as single enzymatically active polypeptides, the conventional nuclear PAPs acquire their substrate specificity by association with additional proteins \[[@pbio-0030189-b02],[@pbio-0030189-b10]\]. All PAPs require an RNA primer with a free 3′ hydroxyl end to which adenosine monophosphate residues are added from adenosine triphosphate (ATP). The canonical PAPs are U-shaped molecules with a tripartite domain organization \[[@pbio-0030189-b11]--[@pbio-0030189-b13]\]. The N-terminal domain carries the active site with three highly conserved aspartate residues \[[@pbio-0030189-b09]\]. The central domain orients the incoming ATP and possibly interacts with the 3′ end of the RNA primer \[[@pbio-0030189-b12]\]. The C-terminal domain contains an RNA-binding domain, nuclear localization signals, and multiple phosphorylation sites involved in regulating enzyme activity.
In 2002, three groups reported the discovery of a new class of poly(A) polymerases. The new polymerases were Cid1p and Cid13p in the fission yeast Schizosaccharomyces pombe \[[@pbio-0030189-b14],[@pbio-0030189-b15]\] and GLD-2 and GLD-3 in Caenorhabditis elegans \[[@pbio-0030189-b16]\]. Recombinant Cid1p and Cid13p from *S. pombe,* members of the Trf4/5 family of proteins first described in budding yeast \[[@pbio-0030189-b17]\], were shown to have RNA-dependent PAP activity in vitro \[[@pbio-0030189-b14],[@pbio-0030189-b15]\]. The proteins are located in the cytoplasm and have been proposed to stabilize and activate specific mRNAs by extending their poly(A) tails \[[@pbio-0030189-b15]\]. The same studies also reported that HA-tagged Trf4p from S. cerevisiae had PAP activity in vitro; however, the activity was not characterized in any detail. Earlier reports had indicated that recombinant Trf4p has DNA-dependent DNA polymerase activity \[[@pbio-0030189-b18],[@pbio-0030189-b19]\].
The GLD-2 and GLD-3 proteins control different steps of germline development in *C. elegans,* including the mitosis/meiosis decision \[[@pbio-0030189-b20]\]. The GLD-2 protein contains a domain shared by Trf4p and Trf5p that resembles the catalytic core of nuclear PAPs. GLD-3 belongs to the bicaudal-C family of RNA-binding proteins and specifically interacts with GLD-2 \[[@pbio-0030189-b16]\]. Both proteins are located in the cytoplasm of germline and early embryonic cells. In vitro translated recombinant GLD-2 was found to have low levels of PAP activity that could be stimulated by GLD-3 \[[@pbio-0030189-b16],[@pbio-0030189-b21]\]. GLD-2 and the other members of the Trf4/5 family lack an RNA-binding domain found in conventional PAPs \[[@pbio-0030189-b21],[@pbio-0030189-b22]\]. Thus, GLD-2 and GLD-3 represent a new type of bipartite PAP, where one subunit contributes the catalytic activity and the other the RNA-binding function.
Polyadenylation in eukaryotes has long been considered to be a unique feature of mRNAs. However, there is increasing evidence that polyadenylation also plays a role in the 3′ end processing of noncoding RNA precursors, such as pre-snoRNAs \[[@pbio-0030189-b23]\] and pre-rRNAs \[[@pbio-0030189-b24]\]. Furthermore, recent in vivo experiments have implicated Trf4p as a component of an RNA surveillance pathway that involves the polyadenylation of hypomodified initiator tRNA^Met^ (tRNA~i~ ^Met^) and its subsequent degradation by the nuclear exosome \[[@pbio-0030189-b25]\]. Mutations in *TRF4,* and in *RRP44,* a gene coding for a subunit of the exosome, were identified as spontaneous suppressors of temperature-sensitive and drug-resistant phenotypes caused by mutations of *TRM6* (*trm6--504). TRM6* encodes a subunit of the enzyme that methylates adenine 58 (m^1^A58) of tRNA~i~ ^Met^. It was shown that molecules of tRNA~i~ ^Met^ lacking the methyl group at A58 are prone to degradation and that they could be stabilized by the deletion of Rrp6p \[[@pbio-0030189-b25]\], a 3′--5′ exonuclease unique to the nuclear form of the exosome (reviewed in \[[@pbio-0030189-b26]\]). The role of Trf4p in this tRNA surveillance pathway was substantiated by overexpressing Trf4p in *trm6--504* cells, which led to a further reduction of the levels of hypomethylated tRNA~i~ ^Met^. Because the overexpression of Trf4p in a *trm6--504/*Δ*rrp6* mutant resulted in the accumulation of the polyadenylated form of undermethylated tRNA~i~ ^Met^, it was proposed that Trf4p was catalyzing this polyadenylation and that the polyadenylation mediated the degradation. However, the authors could not exclude the possibility that Pap1p was responsible for the reaction. The polyadenylation-mediated disposal of aberrant yeast tRNA~i~ ^Met^ is strikingly similar to the degradation mechanism of aberrant tRNAs and of fragmented mRNAs in E. coli \[[@pbio-0030189-b27]--[@pbio-0030189-b29]\]. Presumably, poly(A) tails added to any type of RNA in bacteria serve as tags for binding the degradosome (reviewed in \[[@pbio-0030189-b30]\]).
Here, we report the affinity purification of tagged Trf4p from *S. cerevisiae.* We show that the purified complex has poly(A) polymerase activity provided by the Trf4p subunit and that the activity depends on the zinc knuckle protein Air1p and/or Air2p \[[@pbio-0030189-b31]\]. Comparison of native tRNA~i~ ^Met^ with its in vitro transcribed unmodified counterpart demonstrates that the latter RNA is preferentially polyadenylated by the Trf4 complex. The same substrate specificity is maintained by Trf4 complexes reconstituted from recombinantly expressed components. Additional results from experiments with unmodified and native tRNA^Ala^ and a mutant tRNA^Ala^ believed to fold aberrantly, suggest that the Trf4 complex discriminates between correctly and incorrectly folded tRNAs. We also show that the degradation of unmodified tRNA~i~ ^Met^ by nuclear exosome fractions is dependent on the polyadenylation activity of the Trf4 complex and is stimulated by the putative RNA helicase Mtr4p, which also copurifies with the Trf4 complex. Efficient digestion requires simultaneous polyadenylation by the Trf4 complex, suggesting that the poly(A) tails serve to activate the exosome.
Results {#s2}
=======
Trf4p is a Poly(A) Polymerase {#s2a}
-----------------------------
Sequence alignments have indicated that Trf4p and Trf5p share several characteristic features with canonical poly(A) polymerases: a catalytic domain at the N-terminus with three conserved aspartate residues, a conserved strand loop motif as found in other template-independent polyribonucleotide polymerases \[[@pbio-0030189-b22],[@pbio-0030189-b32]\], and a central domain ([Figure 1](#pbio-0030189-g001){ref-type="fig"}). In contrast to conventional nuclear poly(A) polymerases and similar to the newly identified cytoplasmic PAPs, Trf4p and Trf5p lack a recognizable RNA-binding domain.
::: {#pbio-0030189-g001 .fig}
Figure 1
::: {.caption}
###### Schematic Alignment of the Domain Organization of Trf4p and Trf5p with Other Members of the Pol-β-like Nucleotidyltransferase Family
The proteins are represented as lines, with conserved regions shown as boxes (CAT, catalytic domain; CD, central domain; RBD, RNA-binding domain; sizes in amino acids are indicated to the right). The green boxes in the sequence alignment mark regions with 50% similarity, light blue boxes indicate 100% similarity, and dark blue and orange boxes are regions with 100% identity. The three conserved catalytic aspartates at positions 236, 238, and 294 are in orange and marked by triangles. Regions with four additional amino acids present in Cid1 and GLD-2 are marked with a orange "4."
:::
:::
In order to test whether Trf4p is a bona fide PAP, we affinity-purified Trf4p and assayed the resulting fractions for polyadenylation activity. We used proteinA/\[His\]~6~-tagged Trf4p expressed from a plasmid in a Δ*TRF4* background strain as well as from a TAP-tagged version (Trf4-TAP; \[[@pbio-0030189-b33]\]) of the *TRF4* gene inserted into its original chromosomal locus. Affinity-purified Trf4p-containing fractions from either strain were able to extend a synthetic oligo(A)~15~ RNA substrate. [Figure 2](#pbio-0030189-g002){ref-type="fig"}A shows a polyadenylation assay with Trf4-TAP fractions (lanes 5--7); recombinant yeast Pap1p served as positive control (lanes 2--4). The activity of the Trf4 complex was completely abolished when two of the three highly conserved aspartate residues at positions 236 and 238 in the predicted catalytic domain of Trf4p were mutated to alanines (DADA-TAP; [Figure 2](#pbio-0030189-g002){ref-type="fig"}A, lanes 8--10). The loss of activity was due to the mutations and not to the instability of the mutant protein because both proteins were expressed at similar levels (data not shown). The reaction catalyzed by Trf4-TAP required ATP, and no activity was detected in the presence of GTP, CTP, or UTP ([Figure 2](#pbio-0030189-g002){ref-type="fig"}B, lanes 6--9). In a quantitative assay with oligo(A)~15~ as primer (\[[@pbio-0030189-b34]\]; [Materials and Methods](#s4){ref-type="sec"}), the specific activity of the Trf4 complex was 5.3 × 10^5^ pmol/mg/min. This was similar to the specific activity (5.9 × 10^5^ pmol/mg/min) of recombinant Pap1p measured in parallel. Moreover, incorporation of adenosine monophosphate was strictly dependent on a single-stranded RNA primer (data not shown). These results identified Trf4p as the catalytic component responsible for the polyadenylation activity associated with the purified Trf4 complex. Recombinant Trf4p has been reported to have template-dependent DNA polymerase activity in vitro \[[@pbio-0030189-b18],[@pbio-0030189-b19]\]. We therefore tested the Trf4 complex for DNA polymerase activity using different primer--template combinations and Klenow DNA polymerase as control. In contrast to the previous reports, we could not detect any activity (data not shown).
::: {#pbio-0030189-g002 .fig}
Figure 2
::: {.caption}
###### Trf4p Is the Catalytic Subunit of a New Poly(A) Polymerase
\(A) The Trf4 complex has poly(A) polymerase activity. The 5′-end-labeled oligo(A)~15~ was incubated 30 min with 5, 10, or 20 ng of affinity-purified fractions of the wild-type TAP-tagged Trf4p (Trf4-TAP) or mutant Trf4p with the aspartic acid residues 236 and 238 changed to alanines (DADA-TAP). Protein was omitted in lane 1. Recombinant yeast poly(A) polymerase (Pap1), 1, 2, and 4 ng, was used as a positive control. The migration position of oligo(A)~15~ is indicated by an arrow.
\(B) The Trf4p activity is specific for the addition of adenosine monophosphate. Polyadenylation assays with 20 ng of Trf4-TAP in the presence of different ribonucleoside triphosphates. Recombinant yeast Pap1p, 5 ng, was used as a control. All samples were separated on 15% denaturing gels.
:::
:::
Trf4p Belongs to an Oligomeric Protein Complex {#s2b}
----------------------------------------------
To test whether Trf4p is part of a specific protein complex, affinity-purified Trf4p eluates were analyzed by tandem mass spectrometry (MS-MS). We identified five proteins ([Figure 3](#pbio-0030189-g003){ref-type="fig"}A): Trf4p; Mtr4p, a putative RNA helicase associated with the nuclear exosome \[[@pbio-0030189-b35]\]; Hul4p, a putative E3 ubiquitin ligase \[[@pbio-0030189-b36]\]; and Air1p and Air2p, two zinc knuckle proteins of the CCHC type \[[@pbio-0030189-b31]\]. As an independent test for the significance of the protein interactions, a yeast two-hybrid analysis was carried out with Trf4p as bait. The screen identified fourteen target sequences: five were fragments of Air1p, four were clones of Air2p, and five were clones of the splicing factor Prp16p (\[[@pbio-0030189-b37]\]; [Table S1](#st001){ref-type="supplementary-material"}). The minimal regions required for the interaction with Trf4p were between amino acid residues 127--149 in Air1p, residues 80--171 in Air2p, and residues 315--503 in Prp16p ([Table S1](#st001){ref-type="supplementary-material"}). The minimal Trf4p-interacting regions in Air1p and Air2p overlap ([Figure 3](#pbio-0030189-g003){ref-type="fig"}B, green and black lines above the alignment), suggesting that both proteins bind to Trf4p in a similar fashion. This region is highly conserved and covers one out of four zinc knuckle motifs in Air1p and three out of four present in Air2p, respectively ([Figure 3](#pbio-0030189-g003){ref-type="fig"}B).
::: {#pbio-0030189-g003 .fig}
Figure 3
::: {.caption}
###### Characterization of the Trf4 Complex
\(A) Affinity purification of \[His\]~6~-tagged Trf4 complex identifies five associated polypeptides. Affinity-purified fractions were separated by SDS-PAGE on a 12% gel and stained with Colloidal Coomassie Blue. The protein bands indicated by arrows were identified by MS-MS sequencing. The round holes in the gel lane are a result of punching samples for MS-MS analysis.
\(B) Alignment of conserved regions of Air1p and Air2p. Boxes indicate five predicted CCHC-type zinc knuckle motifs. Lines above the sequences mark regions of interaction between Air1p (green) or Air2p (black) with Trf4p, as inferred from a yeast two-hybrid screen.
\(C) Polyadenylation activity associated with TAP-tagged versions of proteins identified as components of the Trf4 complex (Trf4, Mtr4, Air1, and Air2) and of Trf5. The 5′-end-labeled oligo(A~15~) was incubated with 5 or 10 ng of the affinity-purified extracts and analyzed by electrophoresis on a 15% gel. The position of the input RNA is indicated by an arrow. Protein was omitted in lane I. Recombinant yeast poly(A) polymerase (Pap1), 5 or 15 ng, was used as a positive control.
\(D) Trf4p and Mtr4p copurified with TAP-tagged proteins identified in the Trf4 complex. Western blot analysis of purified eluates with anti-Trf4p and anti-Mtr4p antibody and antibody against the calmodulin-binding region in the TAP-tag (α-TAP). The same amounts of protein complexes as used in the assay (C) were applied to the gel. The higher molecular weight of the Trf4p-TAP protein is due to the TAP-tag. All antibodies were used at a 1:2,000 dilution.
:::
:::
To confirm the polypeptide composition of the Trf4p-containing complex, we affinity-purified complexes from yeast strains carrying Air1p-TAP, Air2p-TAP, and Mtr4p-TAP. All the "reverse tagged" complexes showed polyadenylation activity ([Figure 3](#pbio-0030189-g003){ref-type="fig"}C). Western blot analysis confirmed the presence of Trf4p in all complexes, indicating that it is their catalytic moiety ([Figure 3](#pbio-0030189-g003){ref-type="fig"}D). The Air1-TAP complex contained lower amounts of Trf4p than the other complexes ([Figure 3](#pbio-0030189-g003){ref-type="fig"}D). This was reflected by a lower polyadenylation activity of Air1-TAP ([Figure 3](#pbio-0030189-g003){ref-type="fig"}C). Because Trf5p was not found in the Trf4p-TAP complex, we also purified a TAP-tagged version of Trf5p. However, the Trf5p complexes did not show any significant polyadenylation activity ([Figure 3](#pbio-0030189-g003){ref-type="fig"}C). Instead, we repeatedly observed the incorporation of a single adenosine residue. Compared to the other complexes, Trf5-TAP was expressed less efficiently ([Figure 3](#pbio-0030189-g003){ref-type="fig"}D). It is not clear whether the weak activity was catalyzed by Trf5p or resulted from traces of contaminating Trf4p or Pap1p undetectable by Western blotting ([Figure 3](#pbio-0030189-g003){ref-type="fig"}D; results not shown). The role of Hul4p in the Trf4 complex is currently not known.
Substrate Specificity of the Trf4 Complex {#s2c}
-----------------------------------------
Prompted by the observation that Trf4p appeared to be responsible for the polyadenylation of hypomodified tRNA~i~ ^Met^ in vivo, we decided to test unmodified tRNA~i~ ^Met^ and native tRNA~i~ ^Met^ as substrates for the Trf4 complex ([Figure 4](#pbio-0030189-g004){ref-type="fig"}). To this end we used fully modified tRNA~i~ ^Met^ purified from yeast cells ([Figure 4](#pbio-0030189-g004){ref-type="fig"}B) and its unmodified counterpart prepared by in vitro transcription. As shown in [Figure 4](#pbio-0030189-g004){ref-type="fig"}A, the Trf4 complex could selectively polyadenylate the unmodified tRNA, whereas no polyadenylation activity was observed with the native modified tRNA even after long periods of incubation. Within 30 min, poly(A) tails of about 30 nucleotides had been added to unmodified tRNA. The tails continued to grow linearly with time, reaching a length of about 60 to 70 nucleotides after 2 h. To test whether the 3′ hydroxyl end of the native tRNA was accessible, we showed that both tRNA substrates could be polyadenylated with recombinant yeast Pap1p or E. coli PAP ([Figure S1](#sg001){ref-type="supplementary-material"}).
::: {#pbio-0030189-g004 .fig}
Figure 4
::: {.caption}
###### The Trf4 Complex Preferentially Polyadenylates Unmodified tRNA~i~ ^Met^ and the Unmodified A34GΔU13 Mutant of tRNA^Ala^
\(A) Polyadenylation assay with Trf4p-TAP and unmodified and native tRNA~i~ ^Met^ as substrates. The 5′-end-labeled tRNAs were incubated with 50 ng of Trf4 complex for times indicated and resolved by gel electrophoresis. The migration position of the input tRNA is indicated by an arrow.
\(B) Sequence of yeast tRNA~i~ ^Met^ in cloverleaf form with all the known nucleotide modifications. Tertiary interactions are indicated in solid and dashed red lines. The network of hydrogen bonds involving A20, G57, m^1^A58, A59, and A60, necessary and unique for tertiary interactions in eukaryotic initiator tRNA, is outlined by solid red lines. A\*, 5′-phosphoribosyl-2′-adenosine; D, dihydrouridine; m^1^A, 1-methyladenosine; m^1^G, 1-methylguanosine; m^2^G, *N^2^*-methylguanosine; m^2^ ~2~G, *N^2^,N^2^*-dimethylguanosine; m^7^G, 7-methylguanosine; t^6^A, *N^6^*-threonylcarbamoyladenosine.
\(C) Polyadenylation activity of Trf4p-TAP on unmodified wild-type, A34GΔU13 mutant tRNA^Ala^, and native tRNA^Ala^. The 5′-end-labeled tRNAs were incubated with 20 or 50 ng of Trf4 complex for 30 min at 30 °C. Lane I in each panel shows input tRNA. tRNAs were separated on 10% denaturing gels.
\(D) Yeast tRNA^Ala^ in its cloverleaf form. All known nucleotide modifications are indicated. In the mutant tRNA^Ala^ used in this study, the uracil at position 13 was deleted, and the A34 in the anticodon loop was changed to guanine (marked in red). ψ, pseudouridine; m^1^I^37^, 1-methylinosine generated by deamination of adenosine at position 37. Other modifications are labeled as in (B).
:::
:::
The methyl group on nucleotide A58 of tRNA~i~ ^Met^ is required to ensure that the T-loop adopts its correct structure. Thus, the absence of the methyl group at residue A58 of tRNA~i~ ^Met^ was predicted to interfere with the correct folding of the RNA \[[@pbio-0030189-b25],[@pbio-0030189-b38]\]. To test whether the Trf4 complex specifically recognizes any tRNA with impaired tertiary structure, we searched for a tRNA that does not need to be modified in order to fold correctly in vitro and whose conformation could be disrupted by mutation. It has been shown previously that the conversion of adenosine 37 to inosine of in vitro transcribed unmodified yeast tRNA^Ala^ by the editing enzyme Tad1p was abolished by deletion of U13 \[[@pbio-0030189-b39]\]. This deletion most likely disrupted the last base pair (U13--G22) of the D-stem, and possibly also other crucial tertiary contacts in the neighborhood, such as the triple interaction with G46 (reviewed in \[[@pbio-0030189-b40]\]; [Figure 4](#pbio-0030189-g004){ref-type="fig"}D). It was therefore concluded that in vitro transcribed tRNA^Ala^ could adopt an editing-competent tertiary structure in the absence of modifications. We thus employed unmodified wild-type tRNA^Ala^, the ΔU13 mutant, and native tRNA^Ala^ as substrates for the Trf4 complex. Most of the mutated tRNA^Ala^ was polyadenylated, whereas only little activity was observed with the unmodified wild-type tRNA^Ala^ and none with native tRNA^Ala^ ([Figure 4](#pbio-0030189-g004){ref-type="fig"}C). Taken together, these results suggest that correctly folded tRNAs are poor substrates for the Trf4 complex, whereas aberrantly folded RNAs are efficiently polyadenylated.
Trf4p Is a Bipartite Poly(A) Polymerase That Can Be Reconstituted from Recombinant Components {#s2d}
---------------------------------------------------------------------------------------------
In order to test which of the proteins identified in the Trf4 complex are required for PAP activity, we aimed to reconstitute the activity from recombinant proteins. Because the recombinant Trf4 protein was insoluble when expressed in *E. coli,* a baculovirus expression system was used to isolate Trf4p from insect cells (Trf4-bac). The affinity-purified Trf4-bac alone showed no poladenylation activity on unmodified tRNA~i~ ^Met^ ([Figure 5](#pbio-0030189-g005){ref-type="fig"}A, lanes 5--7). However, polyadenylation was obtained by combining Trf4-bac with either recombinant Air1p or Air2p or both ([Figure 5](#pbio-0030189-g005){ref-type="fig"}A, lanes 8--16). In these assays, 20 ng of Trf4-bac was mixed with 0.5, 3, or 15 ng of recombinant Air1p (Air1) and/or recombinant Air2p (Air2). The simultaneous addition of both Air1p and Air2p proteins led to the formation of longer poly(A) tails than in reactions containing a single Air protein. This was most likely due to the higher protein concentration in the presence of both proteins. The resulting activities were similar to that of Trf4p-TAP isolated from yeast ([Figure 5](#pbio-0030189-g005){ref-type="fig"}A, lanes 2--4). Moreover, the activity of the reconstituted complex was specific for unmodified tRNA~i~ ^Met^ because no significant activity was seen with native tRNA as a substrate ([Figure 5](#pbio-0030189-g005){ref-type="fig"}B). In a control experiment, the recombinant mutant Trf4p (DADA-bac) containing the same amino acid changes as the double aspartate mutant described above had no polyadenylation activity on either tRNA~i~ ^Met^, even in the presence of recombinant Air1p and Air2p ([Figure 5](#pbio-0030189-g005){ref-type="fig"}C and [5](#pbio-0030189-g005){ref-type="fig"}D, lanes 2--7). Also, no activity was observed with a control eluate from a Ni^2+^-NTA column to which extracts from uninfected cells had been applied ([Figure 5](#pbio-0030189-g005){ref-type="fig"}C and [5](#pbio-0030189-g005){ref-type="fig"}D, lanes 8--13). These results demonstrated that the Trf4-PAP reconstituted from recombinant proteins retains the ability to selectively polyadenylate unmodified tRNA~i~ ^Met^.
::: {#pbio-0030189-g005 .fig}
Figure 5
::: {.caption}
###### In Vitro Reconstitution of Trf4 Complex from Recombinant Proteins
Polyadenylation assays were performed with radiolabeled unmodified (A and C) and native (B and D) tRNA~i~ ^Met^ substrate incubated with 25, 50, or 100 ng of Trf4-TAP complex (Trf4-TAP; A and B), or with 5, 10, or 20 ng of recombinant Trf4 protein expressed in the baculovirus system (Trf4-bac; A and B), or with 5, 10, or 20 ng of mutant Trf4-bac (DADA-bac; C and D), or with equal amounts and dilutions of control eluates (CTRL-bac; proteins from cell lysates that unspecifically bound to the Ni^2+^-NTA matrix; C and D). In reconstitution experiments 20 ng of Trf4-bac, or DADA-bac, or control eluates were mixed with 0.5, 3, or 15 ng of recombinant Air1p and/or recombinant Air2p in the combinations indicated. The proteins were pre-incubated for 30 min on ice to allow for binding. Reactions were incubated for 50 min at 30 °C. Control reactions contained no protein (lane 1 in each gel).
:::
:::
Trf4p Stimulates the Activity of the Nuclear Exosome {#s2e}
----------------------------------------------------
Next we asked whether the Trf4 complex could directly stimulate the degradation of unmodified tRNA~i~ ^Met^ by the nuclear exosome in vitro. The fraction of nuclear exosome was prepared by affinity purification of the TAP-tagged Rrp6 protein ([Figure S2](#sg002){ref-type="supplementary-material"}), an exonuclease exclusively associated with the nuclear and not the cytoplasmic form of the exosome \[[@pbio-0030189-b26]\]. First, we tested the effect of partially purified Rrp6p-TAP on unmodified tRNA~i~ ^Met^ ([Figure S2](#sg002){ref-type="supplementary-material"}C). The Rrp6p-TAP complex showed weak exonuclease activity on this RNA, as indicated by the generation of digestion intermediates and mononucleotide degradation products ([Figure S2](#sg002){ref-type="supplementary-material"}C, lanes 2--5). This low activity depended on the presence of ATP in the assay (compare lanes 2--5 with 6--9). Most gel lanes showed a labeled band of about 15 nucleotides. We assumed that these molecules resulted from spontaneous cleavage of a small portion of the RNA.
The effect of the Trf4 complex on the activity of the exosome was examined in coupled exosome/polyadenylation assays. Unmodified tRNA~i~ ^Met^ was pre-incubated with Rrp6-TAP for 30 min. By the end of this time period, a very low amount of degradation products had appeared ([Figure 6](#pbio-0030189-g006){ref-type="fig"}A, lane 2). Addition of Trf4-TAP resulted in a transient appearance of molecules with short poly(A) tails after 10 min ([Figure 6](#pbio-0030189-g006){ref-type="fig"}A, lane 3). The extended molecules decreased after this time and had essentially disappeared after 60 min. The decrease of full-length and elongated tRNAs was accompanied by the formation of short degradation products, the majority of which were dinucleotides ([Figure 6](#pbio-0030189-g006){ref-type="fig"}A, lanes 3--6). Because the RNA substrate was labeled at its 5′ end, the accumulation of dinucleotides indicated that the degradation activity occurred in the 3′ to 5′ direction. This is typical for the nucleases of the exosome \[[@pbio-0030189-b41]\]. The direction of digestion was further confirmed with 3′ end labeled polyadenylated tRNA~i~ ^Met^ as substrate. Incubation with Rrp6-TAP resulted in the rapid production of labeled mononucleotides and the concomitant disappearance of the precursors without any detectable intermediates ([Figure S3](#sg003){ref-type="supplementary-material"}, lanes 2--6). In a control experiment where only buffer was added to Rrp6-TAP fractions, no polyadenylation and no significant degradation of tRNA~i~ ^Met^ were observed ([Figure 6](#pbio-0030189-g006){ref-type="fig"}A, lanes 11--14). The possibility that the low amount of degradation observed in the buffer control could be caused by contaminating Trf4-TAP or by copurifying Mtr4p was tested by Western blotting. We could detect no Trf4p and only traces of Mtr4p ([Figure S2](#sg002){ref-type="supplementary-material"}B). Stimulation of the exosome by the Trf4-TAP complex required the latter\'s polyadenylation activity because no short degradation products accumulated after the addition of DADA-TAP complexes to Rrp6-TAP eluates ([Figure 6](#pbio-0030189-g006){ref-type="fig"}A, lanes 7--10). Very similar results were obtained by first polyadenylating tRNA~i~ ^Met^ with Trf4-TAP and adding Rrp6-TAP in a second step. Almost all the RNA was processed to low molecular weight degradation products within 90 min ([Figure 6](#pbio-0030189-g006){ref-type="fig"}B, lanes 3--6). The experiment showed that adding longer poly(A) tails to the tRNA prior to starting the coupled reaction did not change the kinetics of the subsequent degradation. In a control experiment, assay buffer was added instead of the exosome fraction. In this case Trf4-TAP continued to elongate the poly(A) tails, which reached an average length of 60--70 nucleotides after 90 min and essentially no degradation was detected ([Figure 6](#pbio-0030189-g006){ref-type="fig"}B, lanes 7--10). The kinetics of the coupled reactions suggested a processive mode of digestion because only low amounts of partially degraded RNA molecules were generated during the time course.
::: {#pbio-0030189-g006 .fig}
Figure 6
::: {.caption}
###### The Polyadenylation Activity of the Trf4 Complex Stimulates the Degradation of Unmodified tRNA~i~ ^Met^ by the Nuclear Exosome
\(A) The PAP activity of Trf4 complex is required to stimulate the exosome activity. In a coupled exosome/polyadenylation assay, 5′-end-labeled unmodified tRNA~i~ ^Met^ was incubated with 50 ng of affinity-purified Rrp6-TAP eluate for 30 min as described in [Materials and Methods](#s4){ref-type="sec"} (lane 2), followed by addition of 50 ng of wild-type (Trf4-TAP), mutant complex (DADA-TAP), or buffer A (buffer). Reactions were stopped after 10 (lanes 3, 7, and 11), 30 (lanes 4, 8, and 12), 60 (lanes 5, 9, and 13), or 90 min (lanes 6, 10, and 14) and separated on a 15% gel. Arrows indicate the position of the input tRNA. Protein was omitted in lane 1 of each gel. The migration positions of the degradation products (dp) are indicated by a bracket.
\(B) Coupled polyadenylation/exosome assay. The 5′-end-labeled unmodified tRNA~i~ ^Met^ was pre-adenylated with 50 ng of affinity-purified Trf4-TAP complex for 30 min (lane 2). Then 50 ng of exosome complex (Rrp6-TAP) or buffer A (buffer) was added, and the reactions were continued as in (A).
\(C) Depletion of Mtr4p results in incomplete degradation. Coupled-assay, 5′-end-labeled unmodified tRNA~i~ ^Met^ was pre-incubated for 30 min with Trf4p-TAP lacking Mtr4p (Trf4-TAP w/o Mtr4), followed by the addition of 50 ng of Rrp6-TAP complex or buffer A, and the incubation was continued as in (A).
:::
:::
We then examined the influence of the putative RNA helicase Mtr4p that copurified with the Trf4 complex on the efficiency of digestion by the exosome. We had found previously that the interaction between Trf4p-TAP and Mtr4p could be disrupted by washing the affinity columns with 1 M NaCl prior to elution ([Figure S4](#sg004){ref-type="supplementary-material"}, lanes 2 and 3). Most likely, the Mtr4p-depleted Trf4-TAP still contained the Air proteins because the fractions retained full PAP activity ([Figure 6](#pbio-0030189-g006){ref-type="fig"}C, lanes 7--10). In addition, the interaction of the Air proteins with Trf4p was stable upon high-salt treatment of TAP-tagged Air1 or Air2 complexes (data not shown). We used salt-washed eluates of Trf4p lacking Mtr4p to pre-adenylate the tRNA substrate and for subsequent coupled polyadenylation/exosome assays ([Figure 6](#pbio-0030189-g006){ref-type="fig"}C). Addition of the Rrp6 complexes resulted in the complete removal of the poly(A) tails with time, but the body of the tRNA was not degraded and no significant amounts of small digestion products appeared (compare lanes 3--6 of [Figure 6](#pbio-0030189-g006){ref-type="fig"}B and [6](#pbio-0030189-g006){ref-type="fig"}C). As in the previous experiment, the poly(A) tails continued to grow in the absence of the exosome ([Figure 6](#pbio-0030189-g006){ref-type="fig"}C, lanes 7--10). These results indicate that Mtr4p is required for the degradation of the tRNA body and support the idea that Mtr4p might function as an RNA helicase to remove the highly ordered tRNA structure.
To test whether the presence of a poly(A) tail is sufficient to mediate tRNA degradation we carried out uncoupled exosome assays with pre-adenylated tRNA~i~ ^Met^. Incubation of such poly(A)-tRNA with Rrp6-TAP resulted in removal of the poly(A) tails only ([Figure 7](#pbio-0030189-g007){ref-type="fig"}, lanes 2--5). Complete degradation required the combined action of Rrp6-TAP and Trf4-TAP ([Figure 7](#pbio-0030189-g007){ref-type="fig"}, lanes 6--9). The combination of Rrp6-TAP and mutant Trf4 complex (DADA-TAP) generated only small amounts of dinucleotides ([Figure 7](#pbio-0030189-g007){ref-type="fig"}, lanes 10--13). This likely reflected a mild stimulation of the exosome by Mtr4p present in DADA-TAP, as detected by Western blotting (data not shown). This experiment showed that the addition of a poly(A) tail is not sufficient for complete degradation of tRNA by Rrp6 complex in the absence of Trf4-PAP activity. As in the previous experiments, Trf4-TAP alone had no nuclease activity above background and elongated the pre-adenylated tRNA during the time course ([Figure 7](#pbio-0030189-g007){ref-type="fig"}, lanes 14--17).
::: {#pbio-0030189-g007 .fig}
Figure 7
::: {.caption}
###### Efficient Degradation of tRNA~i~ ^Met^ May Involve Multiple Rounds of Deadenylation and Readenylation
Uncoupled exosome assay on pre-adenylated tRNA. 5′-end-labeled and in vitro polyadenylated unmodified tRNA~i~ ^Met^ was incubated with 50 ng of affinity-purified Rrp6-TAP eluate alone or in combination with 50 ng of wild-type (Trf4-TAP) or mutant (DADA) complex or 50 ng of Trf4-TAP alone. Reactions were stopped after 30 (lanes 2, 6, 10, and 14), 60 (lanes 3, 7, 11, and 15), 90 (lanes 4, 8, 12, and 16), or 120 min (lanes 5, 9, 13, and 17) and separated on a 15% gel. The arrow indicates the position of the nonadenylated tRNA. Protein was omitted in lane 1. The migration positions of the polyadenylated tRNA (poly\[A\]) and the degradation products (dp) are indicated by brackets.
:::
:::
Discussion {#s3}
==========
Trf4p Is the Catalytic Subunit of a New Heteromeric Yeast Poly(A) Polymerase {#s3a}
----------------------------------------------------------------------------
Athough a number of new eukaryotic poly(A) polymerases have been discovered in recent years, the functions of most of them are still unknown. Here we describe the properties of Trf4p, the second poly(A) polymerase of S. cerevisiae. Unlike the other newly described PAPs, which are cytoplasmic, Trf4p is located in the nucleus \[[@pbio-0030189-b42]\]. However, in contrast to the complex multiprotein machinery processing pre-mRNAs associated with canonical yeast Pap1p, Trf4p is part of a small complex in vivo and is capable of polyadenylating tRNAs, products of RNA polymerase III. Lastly, the poly(A) tails made by the Trf4 complex primarily serve to initiate RNA degradation.
The minimal active Trf4-PAP consists of the catalytic subunit Trf4p and either Air1p or Air2p. This minimal complex interacts with Mtr4p, which functionally and physically connects the Trf4-PAP to the nuclear exosome. Although we have identified the Trf4p as the catalytic subunit of the Trf4 complex, the protein is inactive by itself. Polyadenylation activity could be reconstituted by combining recombinant Trf4p and Air1p or Air2p. The Air proteins contain five tandemly arranged zinc knuckles of the CCHC type. Zinc knuckle motifs have been implicated in RNA binding based on mutational and structural studies \[[@pbio-0030189-b43],[@pbio-0030189-b44]\]. We therefore suggest that Air1p and Air2p are RNA-binding subunits of the Trf4 complex. This would explain why the addition of recombinant Air1p or Air2p confers activity to the Trf4p subunit, which lacks a recognizable RNA-binding domain. Such a bipartite composition resembles the heterodimeric cytoplasmic PAP of *C. elegans,* in which the GLD-2 protein carries the catalytic activity and GLD-3 is believed to contribute the RNA-binding function \[[@pbio-0030189-b16]\]. It will be interesting to study whether other noncanonical PAPs are organized in a similar architecture.
The characteristic sequence features of Trf5p and the fact that the *TRF4* and the *TRF5* genes can complement each other and are synthetically lethal when mutated \[[@pbio-0030189-b45]\] suggest that Trf5p is also a poly(A) polymerase. However, except for an extremely inefficient extension of an oligo(A)~15~ primer, no other RNA substrate has been identified until now. Because Trf5p did not copurify with Trf4-TAP, it is likely that it is assembled into a separate complex.
The Trf4 Complex May Recognize Structural Features of Misfolded tRNAs {#s3b}
---------------------------------------------------------------------
We have shown that the Trf4 complex exclusively polyadenylates unmodified and not native tRNA~i~ ^Met^. This raised the question of how the Trf4 complex could distinguish between the two otherwise identical RNAs. Residue m^1^A58 has been proposed to be particularly important for specifying a tertiary substructure unique to initiator tRNA \[[@pbio-0030189-b46]\]. The methylation on a Watson--Crick site forces m^1^A58 to form a symmetric Hoogsteen pair with A54, which promotes the extrusion of A59 and A60 of the loop by stacking with the last base pair of the T-stem (see [Figure 4](#pbio-0030189-g004){ref-type="fig"}B). This conformation of the T-loop induces the specific interactions between the D- and T-loops that are necessary to stabilize the tRNA\'s tertiary structure \[[@pbio-0030189-b46]\]. This may explain why tRNA~i~ ^Met^ is particularly sensitive to the absence of the methyl group on residue A58 compared to other tRNAs containing m^1^A58 \[[@pbio-0030189-b25],[@pbio-0030189-b46]\]. The lack of this methyl group presumably results in a change of the overall shape of the tRNA and a less compact conformation of the T- and D-hairpins. Exposed stem-loops of the hypomethylated tRNA could serve as recognition elements for the Trf4 complex. This model predicts that the Trf4 complex monitors structural features and not the modification state of the tRNAs. Aberrant folding of the tRNA is recognized and leads to the selective activation of the polyadenylation reaction. This idea is further supported by our results obtained with alanine tRNAs. These tRNAs are thought to fold correctly without the need for modifications \[[@pbio-0030189-b39]\]. Accordingly, both unmodified and native wild-type tRNA^Ala^ were poor substrates for the Trf4 complex. In contrast, the ΔU13 mutant was polyadenylated very efficiently, presumably because of its altered tertiary structure.
While finding this model attractive, we realize that experimental proof will require much further work. Probing the structural features of the different forms of tRNAs by chemical and enzymatic methods may help to elucidate the mechanism of RNA recognition. Ultimate proof should be obtained by determining the structure of RNA--protein complexes. The reconstituted recombinant Trf4 complexes showed the same preference for unmodified tRNA substrates as found with Trf4p-TAP from yeast. It is surprising that the substrate preference for unmodified tRNA~i~ ^Met^ could be provided either by Air1p or by Air2p. We assume that Trf4p can form two different complexes containing Air1p or Air2p as alternative subunits. The prediction that the Air proteins are the RNA-binding subunits of Trf4-PAP implies that they not only contribute the RNA-binding functions but also the differential substrate recognition. Future analysis of the RNA-binding properties and the kinetic parameters of the Trf4-catalyzed reaction should help to confirm this prediction and elucidate the mechanism of substrate recognition.
The Trf4 Complex Stimulates the Nuclear Exosome In Vitro {#s3c}
--------------------------------------------------------
We have shown that the polyadenylation activity of the Trf4 complex stimulates the activity of the exosome on unmodified tRNA~i~ ^Met^ in vitro. In agreement with previous work \[[@pbio-0030189-b41]\], the nuclear exosome complex purified from yeast had very low activity on its own. It has been proposed that the activation of the exosome requires additional factors (reviewed in \[[@pbio-0030189-b26]\]). Because Trf4 complexes contain Mtr4p, it is reasonable to assume that this protein mediates the stimulation by attracting the exosome to the tRNA-bound Trf4 complex. As shown here, Mtr4p is needed for the exosome to digest through the structured regions of the tRNA substrates. However, Mtr4p alone (data not shown) or provided by the complex of mutant DADA-TAP was not sufficient to promote tRNA degradation*.* Instead, we demonstrated that the activation of the exosome completely depended on both the polyadenylation activity of the Trf4 complex as well as on the presence of Mtr4p. Moreover, the last experiment implied that complete degradation of the RNA needs the cooperative activity of both the Trf4 and the Rrp6 complexes. This interplay may involve continuous cycles of poly(A) addition and removal before the exosome moves into the structured parts of the tRNA substrates.
The mechanism of the degradation of unmodified yeast tRNA in vitro described above and the model proposed by Kadaba and coworkers \[[@pbio-0030189-b25]\] is similar to the quality control mechanism disposing of aberrant tRNAs and the pathways of mRNA turnover in E. coli \[[@pbio-0030189-b28],[@pbio-0030189-b30],[@pbio-0030189-b47]\]. In eubacteria, tRNA surveillance also depends on polyadenylation and on subsequent degradation by 3′--5′ exonucleases, components of the degradosome. Furthermore, many of the eukaryotic exosome components are related to the enzymes of the bacterial degradosome (reviewed in \[[@pbio-0030189-b26],[@pbio-0030189-b48],[@pbio-0030189-b49]\]). We speculate that the prokaryotic paradigm of PAP-mediated RNA turnover represents the ancient role of polyadenylation and that the stabilization function of poly(A) tails evolved as a consequence of the compartmentalization during the evolution of eukaryotic cells. The fact that Trf4-PAP-mediated polyadenylation resembles the system operating in prokaryotes and that homologs of Trf4p can be identified in all eukaryotic lineages (S. V., unpublished data) implies that the two polyadenylation pathways operate simultaneously in extant eukaryotes.
Moreover, it has been shown that under certain conditions exosome-processed precursors to many mRNAs and stable RNAs accumulate as polyadenylated molecules \[[@pbio-0030189-b23],[@pbio-0030189-b24],[@pbio-0030189-b50]\]. It has been suggested that this polyadenylation requires the canonical nuclear Pap1p; even so, the reactions occur independent of the pre-mRNA cleavage and polyadenylation machinery. It will be interesting to examine whether the Trf4 complex is involved in the polyadenylation-mediated processing of some of these RNAs or whether its substrate repertoire is restricted to aberrant tRNA molecules.
Materials and Methods {#s4}
=====================
{#s4a}
### Plasmids {#s4a1}
The following plasmids were used in this study: WK337 (TRF4-pBluescript KS+), WK339 (wt TRF4-pNOPPATA), WK340 (DADA-TRF4-pNOPPATA), AIR1-pET22a(+), and AIR2-pET22a(+). PAP1p was expressed from plasmid pGM10-YPAP-tag1 and E. coli PAP from a plasmid obtained from Dr. Mario Mörl, Max-Planck-Institute for Evolutionary Anthropology, Leipzig, Germany.
### Construction of plasmids {#s4a2}
Yeast genomic DNA was used to PCR-amplify the coding regions of *TRF4, AIR1, AIR2, MTR4,* and *HUL4.* The full-length ORF of *TRF4* (YOL115W) was amplified by PCR with primers encoding 6× His in frame with the C-terminus of Trf4p. The sequence of the primers used can be obtained upon request. The PCR product was subcloned into the NdeI and SalI restriction sites of the p-Bluescript II KS+ plasmid (Stratagene, La Jolla, California, United States). The TRF4-DADA mutant was generated by site-directed mutagenesis of wild-type *TRF4* in p-Bluescript II KS+. Correct cloning and mutagenesis was confirmed by sequencing. Both inserts were subsequently subcloned to a yeast expression vector (pNoppata) encoding an N-terminal protein A tag and a TEV cleavage site and digested with NdeI and SalI restriction enzymes, resulting in the plasmids WK339 (wild-type *TRF4*) and WK340 (mutant *TRF4*).
The recombinant Air1 and Air2 proteins for the in vitro reconstitution of the Trf4 complex were inserted into the NdeI and XhoI sites of the pET22a(+) expression vector (Novagen, Madison, Wisconsin, United States), expressed in BL21(DE3) and purified on Ni^2+^-NTA columns. All constructs were verified by sequencing.
### Plasmids for yeast two-hybrid analysis. {#s4a3}
The full-length coding region (ORF) of Trf4p (YOL115W) was amplified from genomic DNA of S. cerevisiae with the primers Trf4.13 (5′-aaaaccgcggccatgggggcaaagagtgtaaca-3′) and Trf4.14 (5′-aaattaattaaattaaagggtataaggattatatcc-3′). The PCR products were digested with PacI, filled in with Klenow DNA polymerase and digested with SacII. The purified DNA fragments were inserted into the corresponding restriction sites on the pB27 plasmid (Hybrigenics, Paris, France). Correct cloning was confirmed by sequencing.
### Yeast strains {#s4a4}
Manipulations and growth of Saccharomyces cerevisiae were performed by standard procedures. Genotypes of strains used are as follows: JW72 (*Mat a; BY4741; his3*Δ*1; leu2*Δ*0; met15D0; ura3*Δ*0; YOL115w::kanMX4;* \[pWK336; *TRF4*-pNoppata\]); and JW73 (*Mata; BY4741 his3*Δ*1; leu2*Δ*0; met15*Δ*0; ura3*Δ*0; YOL115w::kanMX4;* \[pWK340; Trf4-DADA-pNOPPATA\]). Strains with C-terminal TAP-fusion of *TRF4* (YOL115W), *AIR1* (YIL079C), *AIR2* (YDL175C), *TRF5* (YNL299W), *MTR4* (YJLO5OW), *RRP6* (YOR001W), and *PAP1* (YKR002W) isogenic to S288C ( ATCC 201388:*MATa his3*Δ*1 leu2*Δ*0 met15*Δ*0 ura3*Δ*0;* Open Biosystems, Huntsville, Alabama, United States) were purchased from BioCat (Heidelberg, Germany). The correct integrations of the TAP-tag were verified by PCR according to the manufacturer\'s instructions. To construct JW72 and JW73, plasmids pWK339 and pWK340 were transformed into the *trf4*Δ strain (clone ID 6265, isogenic to BY4741, Research Genetics, Invitrogen, Carlsbad, California, United States).
### Yeast two-hybrid analysis {#s4a5}
The Trf4p clone used for the yeast two-hybrid screen is described above. The two-hybrid tests were carried out under the auspices of the EEC (grant RNOMICS, QLG2-CT-2001--01554) by Hybrigenics as described \[[@pbio-0030189-b51]\].
### Affinity purification of protein complexes {#s4a6}
Yeast extracts were prepared from 2-l cultures grown to an OD~600~ of 2.0--3.0 as described \[[@pbio-0030189-b52]\]. The purifications on IgG Sepharose were done as described \[[@pbio-0030189-b53]\]. Samples were eluted and stored in buffer A (50 mM Tris (pH 7.9), 120 mM KCl, 0.02% NP-40, 0.5 mM EDTA, 1 mM DTT, 10% glycerol, and protease inhibitors \[0.5 μg/ml Leupeptin, 0.8 μg/ml Pepstatin A, and 0.6 mM PMSF\]). For further purification on Ni^2+^-NTA agarose, peak fractions were pooled and mixed 1:1 with buffer B (50 mM Tris (pH 7.9), 10% glycerol, 150 mM KCl, 0.02% NP-40, 2 mM β-mercaptoethanol, protease inhibitors as in buffer A) and bound to 1/10 of total volume of Ni^2+^-NTA agarose. Samples were washed five times by agitation with one volume of binding buffer containing 10 mM imidazole. Proteins were released with elution buffer containing 250 mM imidazole and dialyzed against buffer B containing 1 mM dithiothreitol instead of β-mercaptoethanol.
### Identification of polypeptides in the Trf4 complex {#s4a7}
Matrix-assisted laser desorption ionization time of flight (MALDI-TOF) mass spectrometry was conducted. Colloidal-Coomassie-Blue-stained bands of interest were in-gel digested with trypsin as described \[[@pbio-0030189-b54]\]. After overnight digestion, about 1 μl was mixed with 1 μl of saturated alpha-cyano cinnamic acid in 50% acetonitrile/0.1% trifluoroacetic acid in water and applied to the MALDI-target. The samples were analyzed with a Bruker Daltonics (Bremen, Germany) Ultraflex TOF/TOF mass spectrometer. An acceleration voltage of 25 kV was used. Calibration was internal to the samples with des-Arg-bradykinin and ACTH(18--38) (both peptides purchased from Sigma, St. Louis, Missouri, United States).
### Nanoelectrospray ionization MS-MS. {#s4a8}
For this approach the peptides obtained after tryptic digestion were desalted and further concentrated on a pulled capillary containing approximately 100 nl of POROS R2 reverse phase material (Applied Biosystems, Framingham, Massachusetts, United States). The peptides were eluted with about 1 μl of 60% acetonitrile in 5% formic acid directly into the nanoelectrospray capillary needle. Mass spectra were acquired on a QSTAR Pulsar I quadrupole TOF tandem mass spectrometer (Applied Biosystems/MDS-Sciex, Toronto, Canada) equipped with a nanoelectrospray ion source (Proxeon, Odense, Denmark) as described \[[@pbio-0030189-b55]\]. Fragmentation by MS-MS yields a stretch of amino acid sequence together with its location in the peptide (sequence tag). With this sequence tag information appropriate databases (e.g., MASCOT) were searched.
### Recombinant protein expression and purification {#s4a9}
E. coli BL21 or M15 cells transformed with the respective plasmids were grown in 2xYT medium at 25 °C to an OD~600~ of 1.0. After induction with 0.4 mM IPTG, incubation was continued for 10 h. Proteins carrying a C-terminal \[His\]~6~ tag were purified on Ni^2+^-NTA agarose (Sigma) according to the manufacturer\'s instructions.
### Expression of proteins with the baculovirus system {#s4a10}
The open reading frame of Trf4p was cloned into the StuI and XhoI sites of the pFASTBAC-vector HTb that contains a \[His\]~6~ affinity tag at both ends. Proteins were expressed with the BAC-To-BAC expression system (Invitrogen) according to the manufacturer\'s protocol. A total of 1 × 10^6^ Sf9 cells were seeded per 35-mm well of a six-well plate in 2 ml of Sf-900 II SFM (Gibco, San Diego, California, United States) containing penicillin/streptomycin and transfected with approximately 5 μl of bacmid DNA. After 48 h at 27 °C, the virus was used to infect 2 × 10^6^ cells that had been seeded on a 10-ml plate. After 4--5 d, 2 ml of the newly produced virus was added to a 30-ml suspension culture with 1 × 10^6^ Sf9 cells/ml. The virus was collected 3--5 d later and used to infect 100-ml cultures. After 4--5 d the cells were harvested and washed with 1× PBS. The pellet was resuspended in approximately two packed volumes of lysis buffer (200 mM NaCl, 50 mM Tris, 10% glycerol 0.02 % NP40, 0.5 mM PMSF, 0.7 μg/ml Pepstatin, and 0.4 μg/ml Leupeptin) and incubated on ice while shaking for 30 min. The lysate was centrifuged for 30 min at 10,000 rpm at 4 °C. The supernatant containing recombinant proteins carrying a \[His\]~6~ affinity tag was then incubated with an appropriate amount of Ni^2+^-NTA resin on a rotor arm in the cold room for approximately 1 h. Proteins attached to the Ni^2+^-NTA beads were either batch-purified or applied to a column and washed with lysis buffer. Elution was done with lysis buffer containing 250 mM imidazole, and proteins were analyzed by SDS-PAGE followed by Western blot analysis.
Protein concentrations were determined with a commercial protein assay kit (Bio-Rad, Hercules, California, United States), and protein amounts in SDS-PAGE were estimated by comparison to BSA standards.
### Antibodies {#s4a11}
A C-terminal portion of Trf4p (aa 356--584) was cloned into the pQE-9 vector and expressed in the E. coli strain M15pREP4 (Qiagen, Valencia, California, United States). The resulting C-terminal Trf4p fragment contained a \[His\]~6~ tag fusion on its C-terminus. The protein was expressed according to the manufacturer (Qiagen) and affinity-purified on Ni^2+^-NTA agarose (Sigma) under denaturing conditions as described. Approximately 100 μg of purified protein was used for three injections into a rabbit (Eurogentec, Seraing, Belgium). Mtr4 antibodies were kindly provided by Dr. Patrick Linder (University of Geneva). The Anti-TAP antibody against the C-terminal region of the TAP-tag were from Open Biosystems.
### Purification of native tRNAs {#s4a12}
About 20% pure fractions of tRNA~i~ ^Met^ and 80% enriched fractions of tRNA^Ala^ were prepared from S. cerevisiae by counter-current distribution \[[@pbio-0030189-b56]\]. tRNA~i~ ^Met^ was further purified to near homogeneity by column chromatography on Sepharose 4B and BD-cellulose under conditions similar to those described previously \[[@pbio-0030189-b57]\], followed by 5′ end labeling and gel purification. tRNA^Ala^ was gel purified after 5′ end labeling.
### In vitro transcription and RNA labeling {#s4a13}
A plasmid ptRNA^Met^ for the in vitro transcription of yeast tRNA~i~ ^Met^ \[[@pbio-0030189-b58]\] was obtained from Dr. Bruno Senger (IBMC Strasbourg). The plasmid was linearized with BstNI and transcribed with T7 RNA polymerase. The plasmid for the transcription of sCYC1 \[[@pbio-0030189-b59]\] was linearized with EcoRI and transcribed with SP6 RNA polymerase. The RNAs for mutant and wild-type tRNA^Ala^ were in vitro transcribed as described \[[@pbio-0030189-b39]\]. All transcripts were purified on 8.3 M urea/12% polyacrylamide gels. For 5′ end labeling, RNAs were treated with alkaline phosphatase and labeled with γ-^32^P-ATP (Amersham Pharmacia, Piscataway, New Jersey, United States) and T4 polynucleotide kinase. Labeled RNAs were purified as described \[[@pbio-0030189-b60]\].
### Renaturation of tRNAs. {#s4a14}
RNA pellets were dissolved in RNase-free water, incubated for 1 min at 65 °C, followed by incubation for 10 min on ice after the addition of an equal volume of twice-concentrated assay buffer containing 10 mM MgCl~2~ (see below).
### Polyadenylation assays and reactions {#s4a15}
Polyadenylation assays were carried out in 10- to 25-μl reaction mixtures containing 5--50 ng of affinity-purified protein, 50 fmol of 5′-end-labeled RNA, 0.5 mM ATP, 5 mM MgCl~2~, 25 mM Tris-HCl (pH 7.9), 20 mM KCl, 10% glycerol, 0.01 mM EDTA, 0.1 mg/ml BSA, 1 mM DTT, 0.02% Nonidet P-40, and 5U of RNA guard (Promega, Fitchburg, Wisconsin, United States). Reactions were incubated at 30 °C for the times indicated and stopped by the addition of 25 mM EDTA. The RNA was precipitated by adding 0.1 volumes of 3M ammonium acetate and three volumes of ethanol. Pellets were resuspended in 6 μl of formamide loading buffer and separated on denaturing polyacrylamide gels. Radioactivity was scanned with a PhosphorImager and results analyzed with ImageQuant software (Molecular Dynamics, Sunnyvale, California, United States).
The specific activities of poly(A) polymerases were measured in 10-μl reaction mixtures containing 20 mM Tris-Cl (pH 8.4), 40 mM KCl, 5 mM MgCl~2~, 2 mM DTT, 0.01% NP-40, and 0.1 mg/ml BSA with 0.5 mM ATP, 0.1 μCi \[α-^33^P\]ATP (3,000 Ci/mmol), and 5 μM (A)~15~ or unmodified tRNA~i~ ^Met^.Recombinant yeast Pap1p \[[@pbio-0030189-b34]\] or Trf4p-TAP complex, 2, 4, or 8 ng, containing 1, 2, or 4 ng of Trf4 protein was incubated for 20 min at 30 °C. The reactions were stopped by spotting onto a 1.5-cm^2^ DE-81 paper. The filters were washed 3× 10 min in 0.3 M ammonium formate/10 mM Na-pyrophosphate, and the incorporated radioactivity was measured in a scintillation counter.
Polyadenylation of tRNA for uncoupled exosome assays was done in 50-μl reaction volumes containing 1 μg of recombinant Pap1p in the presence 100 ng of 5′-end-labeled tRNA, 50 μM ATP, 5 mM MgCl~2~, 25 mM Tris-HCl (pH 7.9), 20 mM KCl, 10% glycerol, 0.01 mM EDTA, 0.1 mg/ml BSA, 1 mM DTT, 0.02% Nonidet P-40, 5U of RNA guard (Promega), and 100 ng of heparin. Reactions were incubated at 30 °C for 30 min and stopped by the addition of 25 mM EDTA. The RNA was precipitated by adding 0.1 volumes of 3 M ammonium acetate and three volumes of ethanol. Pellets were resuspended in 6 μl of formamide loading buffer and separated on denaturing polyacrylamide gels. The polyadenylated population of tRNAs was purified as described \[[@pbio-0030189-b60]\].
### Exosome assays {#s4a16}
Exosome assays were carried out under the same conditions used for polyadenylation assays except that the reactions contained 1 mM ATP and 2 mM DTT. To prevent loss of small degradation products during RNA precipitation, the reactions were stopped by the addition of an equal volume of gel loading buffer containing 80% formamide and separated by electrophoresis without prior precipitation.
### Database searches and sequence analysis {#s4a17}
BLAST searches were conducted in the genome databases (<http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi>) or other nonredundant or EST databases at the Swiss EMBnet node \[[@pbio-0030189-b61]\] (<http://www.ch.embnet.org/software/BottomBLASTadvanced.html>) or at the NCBI (<http://www.ncbi.nlm.nih.gov/BLAST/>).
Supporting Information {#s5}
======================
Figure S1
::: {.caption}
###### Native tRNA~i~ ^Met^ Can Be Polyadenylated by the Activity of Recombinant Yeast Pap1p and Eubacterial PAP
Polyadenylation activity of Pap1p, Trf4p-TAP, and E. coli PAP (Eco-PAP) on native and unmodified tRNA~i~ ^Met^. The 5′-end-labeled tRNAs were incubated with 50 ng of Trf4 complex, 50 ng of Pap1p, or 50 ng of E. coli PAP for the times indicated and separated on a 10% gel. The migration position of the input tRNA is indicated by an arrow. Lane I on each gel contained input tRNA alone.
(871 KB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Figure S2
::: {.caption}
###### Composition and Activity of the Rrp6-TAP Complex
\(A) Polypeptide composition of the affinity-purified TAP-tagged Rrp6p (Rrp6-TAP). The purification was done as described in [Materials and Methods](#s4){ref-type="sec"}, and a sample was separated by 12% SDS-PAGE and stained with silver. The predicted position of Rrp6p is indicated by an arrow.
\(B) Rrp6p-TAP eluates contain residual amounts of Mtr4p. Western blot analysis of Rrp6-TAP with antibodies directed against Mtr4p, Trf4p, and the C-terminus of the TAP-tag. Affinity-purified fractions of Mtr4p (Mtr4-TAP) were used as a control.
\(C) The Rrp6p-TAP complex alone does not efficiently degrade unmodified tRNA~i~ ^Met^. The 5′-end-labeled in vitro transcribed tRNA~i~ ^Met^ was incubated with 100 ng of Rrp6p-TAP complex in the presence of 1 mM ATP (lanes 2--5) or without ATP (lanes 6--9). Reactions were stopped after 10, 30, 60, or 120 min by the addition of one volume of formamide loading buffer and analyzed by electrophoresis on a 15% gel. The position of the input RNA is indicated by an arrow.
(1.2 MB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Figure S3
::: {.caption}
###### The Rrp6p-TAP Complex Has 3′ to 5′ Exonuclease Activity
In vitro transcribed tRNA~i~ ^Met^ was polyadenylated with recombinant yeast Pap1p; the polyadenylated products were eluted from polyacrylamide gel and labeled at their 3′ ends with α-^32^P cordycepin triphosphate (3′ dATP, NEN). The poly(A)-tRNA~i~ ^Met^ was then incubated with 100 ng of Rrp6p-TAP complex (lanes 2--6) or with buffer A (lanes 7--11) for 2, 15, 30, 45, or 60 min, stopped by the addition of one volume of formamide loading buffer, and analyzed by electrophoresis on a 15% gel.
(750 KB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Figure S4
::: {.caption}
###### Mtr4p Can Be Removed from TAP-Tagged Trf4 Complexes by High Salt
Western blot analysis of 100 ng of Trf4-TAP and Mtr4-TAP complexes purified on IgG Sepharose at low salt (L) and high salt (H) conditions. Lane W on each gel shows samples eluted by 1 M NaCl from the affinity columns prior to TEV protease treatment.
(337 KB PDF).
:::
::: {.caption}
######
Click here for additional data file.
:::
Table S1
::: {.caption}
###### Proteins Interacting with Full-Length Trf4p in a Yeast Two-Hybrid Screen
"Start" and "stop" indicate regions of proteins expressed in particular clones. Numbers in brackets indicate that the clone extended beyond the open reading frame. Global predicted biological score (PBS) is a confidence score assigned to each interaction by Hybrigenics, which predicts the occurrence of an interaction in vivo \[[@pbio-0030189-b62]\].
(27 KB DOC).
:::
::: {.caption}
######
Click here for additional data file.
:::
Accession Numbers {#s5a6}
-----------------
The Structural Genomics Consortium (<http://www.sgc.ox.ac.uk/>) accession numbers for genes and gene products discussed in this paper are Air1p (SGDID:S000001341), Air2p (SGDID:S000002334), Hul4p (SGDID:S000003797), Mtr4p (SGDID:S000003586), Pap1p (SGDID:S000001710), Prp16p (SGDID:S000001794), *RRP44* (SGDID:S000005381), Rrp6p (SGDID:S000005527), Tad1p (SGDID:S000003212), Trf4p (SGDID:S000005475), Trf5p (SGDID:S000005243), and *TRM6* (SGDID:S000005006). The Swiss-Prot/EMBL (<http://www.ebi.ac.uk/swissprot/>) accession numbers for genes and gene products discussed in this paper are Cid1 (*S. pombe;* O13833), Cid13 (*S. pombe;* Q9UT49), GLD-2 (isoform a, *C. elegans;* O17087), GLD-3 (*C. elegans;* Q95ZK6), PAP (*E. coli;* P13685), Pap1p (*S. cerevisiae;* P29468), PAPOLA (bovine; P25500), Trf4 (*S. cerevisiae;* P53632), and Trf5 (*S. cerevisiae;* P48561).
We are very grateful to Eric Westhof for invaluable advice and many stimulating discussions and to David Tollervey for generously communicating unpublished results. We thank Jim Anderson, Bernhard Dichtl, and Mihaela Zavolan for helpful suggestions and for improving the manuscript. We also thank Patrick Linder and Bruno Senger for generous gifts of antisera and plasmids and Verena Widmer for excellent technical help. This work was supported by the University of Basel, the Swiss National Science Fund, a European Commission grant (QLG2-CT-2001--01554) to support the RNOMICS Project ([www.eurnomics.org](www.eurnomics.org)) via the Bundesamt für Bildung und Wissenschaft, Bern (grant 01.0123), and the Louis-Jeantet-Foundation for Medicine.
**Competing interests.** The authors have declared that no competing interests exist.
**Author contributions.** SV, JW, GM, SD, and WK conceived and designed the experiments. SV, JW, GM, DB, SD, and AF performed the experiments. SV, JW, GM, DB, SD, AF, HL, and WK analyzed the data. SV, JW, GM, DB, SD, AF, HL, and GK contributed reagents/materials/analysis tools. SV and WK wrote the paper.
Citation: Vanˇácˇová S, Wolf J, Martin G, Blank D, Dettwiler S, et al. (2005) A new yeast poly(A) polymerase complex involved in RNA quality control. PLoS Biol 3(6): e189.
ATP
: adenosine triphosphate
MS-MS
: tandem mass spectrometry
PAP
: poly(A) polymerase
pre-mRNA
: messenger RNA precursor
TAP
: tandem affinity purification
tRNA~i~^Met^
: initiator tRNA^Met^
|
PubMed Central
|
2024-06-05T03:55:55.142541
|
2005-4-19
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079787/",
"journal": "PLoS Biol. 2005 Jun 19; 3(6):e189",
"authors": [
{
"first": "Sˇteˇpánka",
"last": "Vanˇácˇová"
},
{
"first": "Jeannette",
"last": "Wolf"
},
{
"first": "Georges",
"last": "Martin"
},
{
"first": "Diana",
"last": "Blank"
},
{
"first": "Sabine",
"last": "Dettwiler"
},
{
"first": "Arno",
"last": "Friedlein"
},
{
"first": "Hanno",
"last": "Langen"
},
{
"first": "Gérard",
"last": "Keith"
},
{
"first": "Walter",
"last": "Keller"
}
]
}
|
PMC1079788
|
The path from DNA to RNA to protein sounds straightforward enough, but few processes in the life of a cell could be called simple. Gene expression is a complex, highly regulated affair that involves the activity of discrete teams of molecular manipulators at key steps in the protein production pathway. As soon as transcription begins, RNA processing machinery sets to work on messenger RNA (mRNA) precursors in the nucleus, proofreading the RNA copy, stabilizing the elongating transcripts, and making sure mRNAs reach the translation machinery in the cytoplasm.
A key step in RNA processing involves the addition of a string of adenine nucleotides to the 3′ end of the growing transcript, a modification called polyadenylation. (Each RNA chain has what\'s called a 5′ end and a 3′ end, which relates to the chemical polarity of the nucleotides.) In eukaryotes like yeast and humans, polyadenylation helps to stabilize the mRNA transcript and to ensure its export from the nucleus. Once in the cytoplasm, the mRNAs\' poly(A) tails interact with components of the translation apparatus and facilitate protein synthesis. Polyadenylation is mediated by a class of enzymes called poly(A) polymerases (PAPs), which typically act in concert with other proteins in the cell\'s nucleus.
Just three years ago, a new class of PAPs was discovered that belong to the Trf4/5 family of yeast proteins and are found in the cytoplasm, rather than the nucleus, of the cell. Unlike the nuclear PAPs, in which one protein contains both catalytic activity and an RNA-binding domain, the new class relies on two or more protein subunits to carry out these tasks. It\'s been suggested that these enzymes stabilize specific mRNAs in the cytoplasm by extending their poly(A) tails. It\'s long been known that adding poly(A) tails to RNAs in prokaryotic bacteria promotes the degradation of defective RNAs, but the existence of a similar mechanism in yeast has just recently come to light. Previous genetic experiments on live yeast cells by James Anderson and collaborators suggested that when the protein Trf4p polyadenylated a particular type of abnormal transfer RNA (tRNA)---the intermediary that translates the nucleotide code into the amino acid code---the tRNA was destroyed. In a new study, Walter Keller and colleagues investigate the biochemistry, composition, and function of Trf4p in the brewer\'s yeast Saccharomyces cerevisiae, and find evidence that Trf4p-mediated polyadenylation plays a role in RNA quality control in eukaryotes.[](#pbio-0030191-g001){ref-type="fig"}
::: {#pbio-0030191-g001 .fig}
::: {.caption}
###### Model of the tRNA surveillance mechanism via polyadenylation of misfolded RNA by Trf4 complex leading to degradation by the exosome
:::
:::
After showing that mutating two amino acid residues in the predicted catalytic center of Trf4p eliminates its activity, Keller and coworkers determined that the enzyme forms a PAP complex with three other proteins (including two putative RNA-binding proteins, Air1p and Air2p, and a putative RNA-unwinding enzyme called Mtr4p that most likely functions to unwind structured regions in the RNAs). They also showed that the Trf4p complex selectively targets and successfully polyadenylated only tRNA molecules that were either lacking the chemical modifications required for normal folding or that were misfolded by mutation. Polyadenylation appears to tag the aberrant RNA as damaged goods, signaling the cell\'s nuclear molecule-degradation complex, the exosome, to initiate destruction.
Keller and colleagues propose that the Trf4p complex recognizes structural defects in RNA, prompting the Trf4p subunit to add poly(A) tails to the RNA, which initiates RNA degradation. In this model, Trf4p, along with either Air1p or Air2p, interacts with the RNA enzyme Mtr4p, which physically connects the tRNA-Trf4-PAP complex to the exosome. These results suggest that Trf4-PAP monitors the quality of tRNAs by detecting misfolded RNAs and engineering their destruction before they can gum up the works of protein assembly.
That the polyadenylation pathway for discarding defective tRNA appears in both bacteria and yeast suggests that this quality-control mechanism represents the ancient role for polyadenylation, the authors propose; the stabilization function of adding poly(A) tails may have arisen as eukaryotes evolved a nucleus and other organelles. Whether this notion or the model described here proves correct remains to be seen, and the authors outline a number of avenues for further study. Determining the structure of RNA--protein complexes, for example, and their binding properties and interactions, the authors argue, should elucidate the mechanism by which this RNA surveillance complex operates and what features of its RNA substrates it recognizes.
|
PubMed Central
|
2024-06-05T03:55:55.147297
|
2005-4-19
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079788/",
"journal": "PLoS Biol. 2005 Jun 19; 3(6):e191",
"authors": []
}
|
PMC1079789
|
Background
==========
As health care costs rise around the world, and the quest for value for money increases, the role of health services research becomes more and more crucial \[[@B1]-[@B3]\]. Health services research is a multidisciplinary field of scientific inquiry into questions about the appropriateness, equity, effectiveness and efficiency of different means of improving the health status of individuals and populations. It is broad in its approach, and considers interventions across the spectrum from health promotion and illness prevention through treatment to rehabilitation, recovery and/or palliation, some of which may involve sectors other than health \[[@B4],[@B5]\].
With such an ambitious remit, there is a clear need for a critical mass of health services researchers, equipped with the appropriate skills to conduct high quality research and transfer the findings of this research to policy-makers, planners and practitioners \[[@B6]\]. Internationally, there have been calls to increase the capacity of the health services research community, with demands for investment in training and support, and adequate funding for the research endeavour itself \[[@B6],[@B7]\].
Australia and New Zealand have followed the lead of countries like the United States and the United Kingdom in this movement. Health services research took longer to \'get off the ground\' as a coherent discipline here, but similar demands for high quality interventions delivered at the most reasonable cost have led to a greater reliance on evidence from health services research \[[@B8]\]. With the advent of the Health Services Research Association of Australia and New Zealand (HSRAANZ) in 1999, there is now a growing community of health services researchers in these countries \[[@B9],[@B10]\] -- in 2004, the HSRAANZ had 112 individual members and 14 corporate members. However, as is the case elsewhere, health services research has remained largely neglected by key granting bodies, such as the National Health and Medical Research Council (NHMRC, Australia) and the Health Research Council (HRC, New Zealand). With only a few exceptions, these bodies have, at best, subsumed health services research under public health research and, at worst, ignored it altogether. So, for example, although Australian researchers have welcomed new NHMRC program grants for health services research, they have noted that these grants have only been introduced after decades of successive reviews have proposed such initiatives, and that they fall short of the recommendations for broader, strategic changes. New Zealand is one step further behind still, with no such grants available.
As a consequence, most of the work of health services research in Australia and New Zealand is conducted at the behest of health departments, and is rarely basic (adding to the conceptual, theoretical or methodological base of the discipline) or investigator-driven. Again, there are exceptions (e.g., funding of the Primary Health Care Research and Evaluation initiative in Australia), but they are limited in number. Funding tends to be project-based, resulting in many health services researchers being employed on tenuous contracts, which culminate in commissioned reports and do not allow for writing peer-reviewed journal articles. There is limited funding for formal training opportunities, which means that most health services researchers come to the discipline by luck, rather than by design. Consequently the capacity of the health services community is still lacking \[[@B8]-[@B11]\].
To date, this fairly bleak picture has been largely based on anecdotal evidence, rather than academic inquiry. With the exception of some attempts to quantify the inputs to and outputs of health services research in these countries \[[@B9]-[@B11]\], there has been no scientific attempt to assess the capacity of the health services research community in Australia and New Zealand. Indeed, little work of this kind has been done internationally.
The current paper aims to systematically profile the health services research community in Australia and New Zealand, with a view to assessing its capacity. It reports on a survey, conducted to describe the backgrounds and current arrangements of these health services researchers, and to explore their opinions regarding what needs to change if health services research is to maximise its impact. This descriptive work is crucial in an era where health services research is still trying to make its mark.
Methods
=======
Potential respondents included resident Australians and New Zealanders who were identified from the current individual membership list of the HSRAANZ and the list of delegates of the HSRAANZ\'s Third Health Services Research and Policy Conference, held in Melbourne from 16--19 September 2003. This yielded a sampling frame of 282 -- 37 (13%) were HSRAANZ members only, 179 (63%) were conference attendees only, and 66 (23%) featured on both lists.
On 2 August 2004, emails were sent to all 282 potential respondents, inviting them to complete a confidential, web-based survey. Reminder emails were sent on 16 August, and follow-up phone calls were made on 24 August.
The survey elicited information on respondents\' professional role, employment setting, level of appointment, work environment, security of tenure, sources of funding, and research outputs, as well as demographic details. It also invited them to comment on issues they perceived as important in relation to the ability of health services research to have an impact on policy and practice.
Quantitative data were analysed using SPSS and are presented as frequencies and percentages.
Qualitative data were examined to identify major content and themes, and then individual responses were classified according to these themes. The intention was to present \'typical\' responses, while at the same time indicating the range of views presented within a given theme.
Results
=======
Response rate and sub-sample of interest
----------------------------------------
In total, 282 invitations to complete the survey were emailed, and 191 individuals submitted responses (68%). All respondents were asked, \'What is your role in health services research, and what percentage of your time do you spend in that role?\' One hundred and sixty five (86%) indicated that they conducted or managed health services research, at least part of the time. The remainder commissioned or utilised health services research, but were not actually involved in carrying it out, and are therefore excluded from the following analyses.
Demographic details
-------------------
Of the 165 respondents of interest, 136 respondents (82%) were from Australia, 27 (16%) were from New Zealand, and two (1%) had recently begun working overseas. One hundred and three (62%) were female; 62 (38%) were male. Sixteen (10%) were aged less than 35; 48 (29%) between 35 and 44; 68 (41%) between 45 and 54; 33 (20%) 55 or more.
Academic qualifications
-----------------------
Respondents were asked to indicate their academic qualifications. Of the 162 who provided this information, 56 (35%) indicated that their original undergraduate degree was in Arts, 43 (27%) in Science (including Applied Science, Behavioural Science and Social Science), 28 (17%) in Medicine, 9 (6%) in Economics, 6 (4%) in Business/Commerce/Management, and 3 (2%) some other discipline. Seventeen (11%) did not specify their original degree.
The majority had postgraduate qualifications: 68 (42%) had attained a Masters degree (most commonly a Master of Public Health or Master of Arts) as their highest qualification; 61 (38%) had been awarded Doctorates.
Respondents\' qualifications were examined more closely for evidence of clinical backgrounds (e.g., nursing diplomas etc). At a minimum, 56 (35%) had clinical backgrounds of some sort (e.g., medicine, nursing, social work, psychology, physiotherapy, nutrition, health promotion, exercise and pharmacy). This figure is likely to represent an underestimate, since there may have been other respondents with general degrees (e.g., Bachelors of Applied Science) that led to clinical work.
Professional roles and involvement in health services research
--------------------------------------------------------------
Table [1](#T1){ref-type="table"} provides a picture of the professional roles of respondents. One hundred and twenty one (64%) saw \'health services research\' as within their professional role, but other professional roles were also relatively common: \'health policy analysis/planning\' was indicated by 75 respondents (39%); and \'other\' by 66 (35%). The latter included teaching/education, administration/management of services other than health services, research and evaluation in other allied areas (e.g., social policy analysis), community service provision/management, service development and various types of consultancy. Less common were \'health services management/administration\' and \'health care provision\', selected by 33 respondents (17%) and 23 respondents (12%), respectively. Typically, respondents worked part time rather than full time in these roles, often juggling more than one role. The mean number of years spent in each role by respondents was around 10, with the exception of \'health care provision\' where the average was much higher at nearly 21 years.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Professional roles by time fraction and years in role
:::
**Health services research** **Frequency** **Percentage**
----------------------------------------------- --------------- --------------- ------------------------
**Health policy analysis/planning** **Frequency** **Percentage**
Full time 24 36.4
Part time 42 63.6
Total 66 100.0
**Mean** **Standard deviation**
Years in role 10.4 6.8
**Health services management/administration** **Frequency** **Percentage**
Full time 8 27.6
Part time 21 72.4
Total 29 100.0
**Mean** **Standard deviation**
Years in role 10.4 5.9
**Health care provision** **Frequency** **Percentage**
Full time 2 10.5
Part time 17 89.5
Total 19 100.0
**Mean** **Standard deviation**
Years in role 20.9 8.3
**Other professional role** **Frequency** **Percentage**
Full time 28 47.5
Part time 31 52.5
Total 59 100.0
**Mean** **Standard deviation**
Years in role 10.7 8.5
:::
Employment setting and arrangements
-----------------------------------
Table [2](#T2){ref-type="table"} provides a breakdown of respondents\' employment settings. Ninety six respondents (58%) were employed in university settings. Government health departments employed 30 respondents (18%), and health services employed 26 (16%). Employment in private companies and self-employment were less common. With the exception of those who were self-employed, respondents tended to be employed on a full-time basis, and the majority described their level of appointment as mid-career or senior.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Employment setting by time fraction and level of appointment
:::
**University** **Frequency** **Percentage**
---------------------------------- ------------ --------------- ----------------
**Government health department** **Frequency** **Percentage**
Full time 25 83.3
Part time 5 16.7
Total 30 100.0
Junior 0 0.0
Mid-career 16 53.3
Senior 14 46.7
Total 30 100.0
**Health service** **Frequency** **Percentage**
Full time 15 57.7
Part time 11 42.3
Total 26 100.0
Junior 2 8.3
Mid-career 11 45.8
Senior 11 45.8
Total 24 100.0
**Private company** **Frequency** **Percentage**
Full time 7 100.0
Part time 0 0.0
Total 7 100.0
Junior 0 0.0
Mid-career 3 50.0
Senior 3 50.0
Total 6 100.0
**Self-employed** **Frequency** **Percentage**
Full time 3 23.1
Part time 10 76.9
Total 13 100.0
Junior 0 0.0
Mid-career 1 9.1
Senior 10 90.9
Total 11 100.0
:::
Security of tenure
------------------
Respondents were asked about their security of tenure. Of the 157 who responded to this question, 50 (32%) had employment contracts of greater than five years. However, a similar proportion had far more precarious tenure: 23 (15%) had no contract at all, 7 (5%) had less than a one year contract and 18 (12%) had a one year contract. The remaining 59 (38%) had contracts of between two and five years.
Individual funding sources
--------------------------
Respondents were asked to indicate their general source(s) of current funding. Eighty one (49%) had core funding from their employing institution, 72 (44%) had project funding from key granting agencies, 51 (30%) had funding from government consultancies, and 29 (18%) had funding from other sources, including other government funding, funding from non-government organisations, private funding, self-funding and scholarships.
Significant funding bodies
--------------------------
Respondents were asked to list the funding bodies they regarded as the most significant in terms of health services research funding in their country, and 148 (90%) did so (see Table [3](#T3){ref-type="table"}). Overwhelmingly, they indicated that competitive funding bodies (e.g., the NHMRC and the Australian Research Council in Australia and the Health Research Council in New Zealand) were the most significant: 90 (61%) listed a competitive funding body as their first choice; 45 (30%) as their second choice; and 29 (20%) as their third choice. Similarly, respondents gave considerable weight to government departments as funders (usually the Commonwealth Department of Health and Ageing or State/Territory Health Departments in Australia or the Ministry of Health in New Zealand): 52 respondents (35%) listed a government department as the most significant; 64 (43%) as the second most significant; and 26 (18%) as the third most significant. Other funding bodies deemed to be significant were universities and institutes, health foundations, non-government organisations, philanthropic trusts, private companies, drug companies, insurers and professional bodies, but each was only listed by a small number of respondents.
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Funding bodies regarded as most significant in terms of health services research funding
:::
**1^st^choice** **2^nd^choice** **3^rd^choice**
---------------------------------- ----------------- ----------------- ----------------- ------ ---- ------
**Competitive funding bodies** 90 60.8 45 30.4 29 19.6
**Government departments** 52 35.1 64 43.2 26 17.6
**Universities and institutes** 2 1.4 4 2.7 5 3.4
**Health foundations** 1 0.7 6 4.1 7 4.7
**Local services** 1 0.7 1 0.7 1 0.7
**Professional bodies** 1 0.7 1 0.7 0 0.0
**Non-government organisations** 1 0.7 0 0.0 4 2.7
**Other** 0 0.0 2 1.4 2 1.4
**Philanthropic trusts** 0 0.0 2 1.4 2 1.4
**Drug companies** 0 0.0 1 0.7 1 0.7
**Private companies** 0 0.0 1 0.7 1 0.7
**Insurers** 0 0.0 0 0.0 2 1.4
:::
Work environment
----------------
Respondents were asked to consider five statements about their respective work environments. Table [4](#T4){ref-type="table"} shows that the majority agreed with the statements \'I work in a setting where peer support for health services research is available\' (81%), \'At work, I am co-located with other health services researchers\' (65%) and \'I work in a setting where I can easily seek and receive advice about health services research\' (81%). There was less agreement with the statements \'I work with a critical mass of health services researchers\' and \'The institution/centre/unit where I work has a clear health services research focus\', with 41% and 55% indicating that these statements described their work situations, respectively.
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Work environment
:::
**I work in a setting where peer support for health services research is available** **Frequency** **Percentage**
--------------------------------------------------------------------------------------------------- -------- --------------- ----------------
**At work, I am co-located with other health services researchers** **Frequency** **Percentage**
Yes 101 64.7
No 53 34.0
Unsure 2 1.3
Total 156 100.0
**I work in a setting where I can easily seek and receive advice about health services research** **Frequency** **Percentage**
Yes 128 80.5
No 22 13.8
Unsure 9 5.7
Total 159 100.0
**I work with a critical mass of health services researchers** **Frequency** **Percentage**
Yes 64 40.8
No 82 52.2
Unsure 11 7.0
Total 157 100.0
**The institution/centre/unit where I work has a clear health services research focus** **Frequency** **Percentage**
Yes 90 55.2
No 62 38.0
Unsure 11 6.7
Total 163 100.0
:::
Health services research outputs
--------------------------------
The research outputs of the sub-sample are shown in Table [5](#T5){ref-type="table"}. One hundred and eight respondents (65%) had produced journal articles, 10 (6%) had published books, 37 (22%) had contributed to book chapters, 80 (48%) had written commissioned reports, and 68 (41%) had showcased their work in conference proceedings. In the case of journal articles, commissioned reports and conference proceedings, the majority had produced at least three and sometimes four or more outputs of this kind; for books and book chapters, the norm was one.
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
Health services research outputs
:::
**Peer-reviewed publications in academic journals** **Frequency** **Percentage**
----------------------------------------------------- ----------- --------------- ----------------
**Books** **Frequency** **Percentage**
1 9 90.0
2 1 10.0
3 0 0.0
4 or more 0 0.0
Total 10 100.0
**Book chapters** **Frequency** **Percentage**
1 22 59.5
2 9 24.3
3 2 5.4
4 or more 4 10.8
Total 37 100.0
**Commissioned reports** **Frequency** **Percentage**
1 20 25.0
2 22 27.5
3 13 16.3
4 or more 25 31.3
Total 80 100.0
**Published conference proceedings** **Frequency** **Percentage**
1 16 23.5
2 17 25.0
3 12 17.6
4 or more 23 33.8
Total 68 100.0
:::
Respondents were also given the opportunity to indicate other outputs in free text. Additional outputs included conference presentations, website publications, non-peer-reviewed journal articles and letters, other reports (e.g., guidelines, manuals, policy documents, methodology papers, reports to study participants, internal reports, research bulletins) and theses.
Issues impacting upon health services research
----------------------------------------------
Respondents were asked to respond (in free text) to the question, \'What do you think are the most important issues for health services research if it is to have an impact on health policy and planning in Australia and New Zealand?\' Several common themes emerged, including building the capacity of the health services research community, funding, translating research into policy/practice, collaborating with others, and undertaking research in specific content areas. Each of these themes is explored in more detail below.
### (1) Building the capacity of the health services research community
Many respondents viewed building the capacity of the health services research community as a critical issue, commenting on different aspects that needed to be addressed in order for this to occur.
Several felt that the profile of health services research needed to be raised, in order for it to receive greater recognition among funders and decision-makers. Typical comments emphasised: \'increas \[ing\] recognition of health services research as a distinct and important research stream\', \'increasing the profile as a priority area for research\' and \'establish \[ing\] a more visible, cohesive and research-oriented presence as a disciplinary and professional grouping\'. These respondents offered a variety of means of achieving this end, most notably \' \[seeking\] clarity about the role of health services research\' by \' \[setting\] a coherent national research agenda\' with agreed priorities, and \'improving the quality of research performed\', both in terms of its \'theoretical foundations\' and \'methodological rigour\'.
Another common theme was the perceived need to improve the \'critical mass of the health services research community\', by such means as improving \'continuity and longevity of research groups\', developing \'linked network \[s\] of health services research centres throughout Australasia\', and by offering opportunities for \'peer support (e.g., development of the HSRAANZ)\'.
Training and career development opportunities were also seen as issues. Many respondents highlighted a need for \'strengthening research training\' through \'funding to support capacity building at all levels from \[postgraduate\] educational programs to furthering professional development of mid to senior researchers\' and \'fellowships provided to facilitate the development of health services research as a viable career\'. Beyond training, respondents stressed the need for \'nurturing researchers\', calling for \'funding and policies to support proper career pathways for health services researchers\', \'a proper career structure and transparent promotion arrangements\'. Many respondents indicated that a pre-requisite for strengthening career trajectories for health services researchers was improving security of tenure, commenting that \'people need some job security\' and \'insecurity of tenure of staff employed on short-term contracts means that experienced people leave for more secure employment elsewhere\'.
### (2) Funding
Over and above funding for training and career development, outlined above, respondents stressed a broader need for \'more funding for longer periods of time\' to support \'rigorous, well-planned research\'. They observed a need for both \'infrastructure funding\' and project funding, particularly for \'larger, in-depth studies \... \[as opposed to\] \... a lot of smaller projects that duplicate each other\'. They noted that health services research has been paid lip-service, being given \'priority \... on paper but not in practice\'.
Problems were noted with securing funding from key granting bodies, where there was a perception that these agencies viewed other areas of health as more significant (e.g., \'it\'s as important as biomedical research but not recognised as such\'). Problems were also noted with seeking funding from government departments, which, according to respondents, often view their role as funding health care, and not the research that underpins it (e.g., \'it needs to be funded as an intrinsic component of health service delivery\').
### (3) Knowledge transfer
Many respondents felt that improving knowledge transfer was crucial, making comments like \'there is presently a lack of translation of research results into practice and policy\' and \'research translation issues are a priority\'.
They suggested several ways in which the current situation could be improved. A number stressed the need to increase the relevance of research to decision-makers, as exemplified by comments like \'the research needs to be relevant to the needs of service providers and policy makers\', \' \[it\] needs to be applied and relevant to contemporary health policy issues\' and \' \[we must\] focus on providing evidence which can help answer key policy questions\'.
Many commented on the need for effective dissemination methods, noting that strategies should include \'timely\' and \'appropriately written\' reports that are \'useful to service providers and policy-makers\'. Some also observed that \'being innovative about dissemination of \... evidence\' could be helpful, citing alternative strategies such as \'get \[ting\] into the public media more, get \[ting\] into policy debates, be \[ing\] seen and heard\'.
Also common were comments about the need for improved relationships between health services researchers and decision-makers, with calls for \'increased collaboration \[between\] researchers and end-users\' and \'increased partnerships between researchers and policy representatives to increase the relevance of research to policy issues\'. There was recognition that this would require strategies to involve decision-makers on individual projects (e.g., \'be inclusive of policy-makers on working groups and steering committees\'), but that this alone would have \'little impact\'. \'Developing strong ongoing relationships between health service researchers and policy-makers around programs of work\' and \'frequent and timely dialogue with decision-makers\' were seen as important for longer-term, stronger relationships. Several respondents commented that improved relationships between health services researchers and decision-makers would lead to an improved capacity on the part of the former to understand decision-making processes and an increased ability on the part of the latter to commission, understand and interpret health services research.
### (4) Collaboration with others
Many respondents were positive about the \'multidisciplinary\' nature of health services research, but noted that health services researchers need to collaborate with others in order to maximise their contribution. As well as forging stronger relationships with policy-makers and planners (described above), respondents noted that there was a need to develop collaborations with \' \[service\] providers\' and \'partner agencies\', to conduct \'joint research with those at the coal face\'.
### (5) Specific content areas for research
A number of respondents saw opportunities to strengthen health services research and contribute to policy and planning decisions by focusing on specific content areas. Commonly identified areas included: health care financing; health system reform; health workforce; health insurance; primary care; hospital care; outcomes, effectiveness and cost-effectiveness; quality of care; and health inequalities.
Discussion
==========
Study limitations
-----------------
The study had several limitations which must be taken into account in interpreting the above findings. Firstly, and most importantly, the study\'s sampling frame (defined by membership of the HSRAANZ and/or attendance at the Third Health Services Research Conference) may have excluded some health services researchers, introducing some systematic biases. For example, junior and \'commercial\' health services researchers may be less likely to join professional associations and attend conferences, which may have artificially inflated the representation of mid-career and senior researchers, and given undue weight to the views of \'academic\' health services researchers, respectively. Similarly, New Zealanders may have been under-represented (given the cost of attending a conference in Melbourne), which would have exaggerated the influence of the Australian respondents. However, analyses of the quantitative data indicated that there were no significant differences between respondents on the basis of country of residence. It is acknowledged that the study findings are drawn from a sub-sample of the health services research community, albeit a significant sub-sample, and therefore caution needs to be exercised in extrapolating the findings to the entire health services research community. Defining the sampling frame to ensure broader representation would have been extremely difficult (e.g., identifying a list of \'commercial\' researchers), and beyond the financial constraints of the current study.
Secondly, there is a question about the extent to which respondents were representative of the overall sample. The survey\'s response rate was relatively high, but it may have been that those who did not respond were individuals who had fewest concerns about the state of health services research in Australia and New Zealand. It was not possible to compare the profiles or respondents and non-respondents, since no demographic data were available on the latter.
Thirdly, the survey faced the same problems as all self-report surveys, including issues such as recall bias and social desirability of responses (e.g., the question on respondents\' research outputs required them to estimate the total number of each kind of output in the previous year and some may have had difficulty remembering the exact number, and therefore \'rounded up\', particularly if they felt that this brought them closer to some imagined norm).
Finally, the survey did not collect any ethnicity data. This was beyond the scope of the survey, but would have been desirable since some of the situations described in the survey may vary for particular population groups. For example, in New Zealand, the introduction of Māori and Pacific Island Career Development Awards has led to a growth in junior health services researchers from indigenous backgrounds who see the awards as a way of gaining qualifications, skills and expertise while working in an area of benefit to their people.
Interpretation of findings
--------------------------
These limitations aside, the study has largely confirmed the conjecture of key commentators on the capacity of the health services research community in Australia and New Zealand. This community is characterised by highly qualified professionals who have come to health services research via a range of academic and professional routes (including clinical backgrounds), the majority of whom are women aged between 35 and 54 who have mid- to senior- level appointments. They are primarily employed in universities and, to a lesser extent, government departments and health services. Although most are employed in full time positions, many are only able to devote part of their time to health services research, often juggling this with other professional roles. They rely heavily on external funding, as only half have core funding from their employing institution and around one third have employment contracts of one year or less. Many view issues around building the capacity of the health services research community and addressing funding deficits as crucial if health services research is to be translated into policy and practice. Despite the difficulties they face, most are positive about the support and advice available from peers in their work settings, and many are actively contributing to knowledge through academic and other written outputs.
The insecurity of tenure, competing demands on time, paucity of training opportunities, and the poor definition of career pathways described by survey respondents are significant issues. Indeed, health services researchers are taking a stand, lobbying for funding and support commensurate with the importance of their work \[[@B10]\]. There may be a relatively brief window of opportunity within which change must occur, since the current study provides some evidence that the discipline is top-heavy with mid- and senior-level researchers, and that there are not enough junior researchers coming through the ranks. The mid- and senior-level researchers receive strong peer support from a small but cohesive group of colleagues, and are presumably committed to providing sound evidence upon which health policy and planning decisions can be made. But other career options may begin to look attractive and attainable, particularly given their broad discipline base and their high level of qualifications, jeopardising the discipline\'s critical mass.
The current study provides evidence of a strongly-perceived need for funding commitment for health services research from competitive granting bodies and government departments. Competitive granting bodies are viewed as the most significant funding sources by health services researchers, despite the fact that, collectively, their policies and processes have favoured bio-medical, clinical and public health research over health services research. Government departments are also seen as important, but also have a relatively poor record of supporting health services research, largely viewing it as outside their bailiwick. There is a clear need for a strategic approach, where the health services research community (through the HSRAANZ) collaborates with competitive granting bodies and government departments to define and fund a research agenda that balances priority-driven and investigator-driven research and which provides support for training and career development.
Conclusion
==========
Unlike many other forms of health research, health services research has the potential to generate savings in the health system by providing evidence about the best way to deliver health care that achieves optimal outcomes at the most reasonable cost. For this reason alone, it would make sense for policy-makers and funders to take the concerns of the health services research community seriously, foster its development, and contribute to maximising its capacity through a sustainable approach to funding.
Declaration of competing interests
==================================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
JP, SG and SP conceptualised the study and took the lead on designing the survey, and MH, JC and JH helped to refine the sampling frame and survey design. JP and SD undertook the data management and analysis, with input from all other authors. JP took primary responsibility for drafting the original version of the paper, and all other authors contributed substantially to revised drafts. All authors read and approved the final manuscript.
Acknowledgements
================
Funding for survey development and administration was provided by the HSRAANZ. The authors would like to thank members of the HSRAANZ Executive for commenting on early drafts of the survey, and Simon Palmer, Gayle Main and Adam Clarke at Strategic Data Pty Ltd for managing the survey administration process.
|
PubMed Central
|
2024-06-05T03:55:55.147936
|
2005-3-8
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079789/",
"journal": "Aust New Zealand Health Policy. 2005 Mar 8; 2:4",
"authors": [
{
"first": "Jane",
"last": "Pirkis"
},
{
"first": "Sharon",
"last": "Goldfeld"
},
{
"first": "Stuart",
"last": "Peacock"
},
{
"first": "Sarity",
"last": "Dodson"
},
{
"first": "Marion",
"last": "Haas"
},
{
"first": "Jackie",
"last": "Cumming"
},
{
"first": "Jane",
"last": "Hall"
},
{
"first": "Amohia",
"last": "Boulton"
}
]
}
|
PMC1079790
|
Background
==========
The last decade has witnessed a revolution in payment arrangements for public hospitals in Australia. Victoria was the first state to introduce casemix using diagnosis related groups (DRGs) as the basis of paying for inpatient care in public hospitals in 1993. Within 3 years, all other states except New South Wales had followed suit \[[@B1]\]. The widespread adoption of DRGs to replace historical budgets resulted in measurable gains in productivity as public hospitals became accountable for delivery of specified outputs \[[@B2]\]. DRGs group together patients with similar expected resource utilization, shifting the incentive from purchasers to health providers to manage the components of care more efficiently.
Contemporaneously, the Commonwealth Government legislated in 1995 to enable contracting between insurers, hospitals and doctors, arising from concerns about declining private health insurance membership fuelled by consumer dissatisfaction at insurance co-payments \[[@B3],[@B4]\]. Commonwealth regulation of contracting arrangements under the *National Health Act 1953*requires that contracts between hospitals and insurers be described on the basis of casemix episodic payments using Australian National DRGs (AN-DRGs). However insurers and hospitals have flexibility as to the actual structure of the payment mechanism used in such contracts.
In introducing the 1995 contracting reforms, the Commonwealth Minister proposed that one of the central objectives was to transform private health insurers from \"passive bill payers\" into active purchasers of services for their members \[[@B5]\]. Health purchasing has been defined as \"buying the best value for money services to achieve the maximum health gain for those most in need\" \[[@B6]\]. However this definition is from a study of the experience of the British National Health Service and the reference to people \'most in need\' may be less relevant to the role of private sector purchasers. In a review of major health service purchasers in the United States, \'value purchasing\' was defined as \"an organized attempt by a private- or public-sector purchaser to ensure quality and to improve health outcomes, as well as negotiating prices, as an explicit part of its health care buying strategy\" \[[@B7]\]. The simpler version of this definition was that value purchasing involved \"getting the best care for the best price\".
Australian private health insurance provides access to doctor of choice and private hospitals, together with covering ancillary services such as physiotherapy and dental care. In turn, the public Medicare program provides free access to public hospitals and subsidized access to medical services and pharmaceuticals. The issue of whether private health insurance duplicates or supplements Medicare has been considered in several major reviews, including by the Industry Commission and the Organization for Economic Co-operation and Development \[[@B8],[@B9]\]. The lack of consensus on the desired role of private health insurance in the context of a universal public insurance program complicates the design of the appropriate regulatory framework for private health insurance.
Private health insurance is tightly regulated by the federal government under the *National Health Act 1973*, although the legislation itself lacks a set of clear policy objectives to provide a coherent rationale for regulation \[[@B9]\]. The purchasing role of health insurers is significantly shaped and potentially constrained by regulatory requirements including:
• Clinical autonomy -- health insurance contracts are required to maintain the medical practitioner\'s professional freedom within the scope of accepted clinical practice;
• Contracting flexibility -- health insurers are prohibited from refusing to contract with hospitals on the basis of the number of hospital beds, the range of hospital treatments or the hospital ownership arrangements;
• Benefit mandates -- all health insurance products must provide coverage for palliative care, psychiatric services and rehabilitation;
• Second tier default benefits -- health insurers are required to pay benefits equal to 85% of average contracted benefits to private hospitals with which they do not contract; and
• Reinsurance -- the costs of high service users and the elderly are redistributed and equalized across all health insurers.
While the first three of these requirements directly limit the scope of acceptable contracting by health insurers, the default benefits and reinsurance arrangements are generally perceived to be equally, if not more, significant in influencing purchasing behavior. Interestingly, the legislation is silent on \'positive\' criteria for purchasing and does not stipulate use of methods such as cost-effectiveness, or require insurers to consider factors such as quality, clinical outcomes, value-for-money or efficiency, in entering contracts with hospitals and medical professionals.
At the macro-regulation level, two other factors influence the purchasing environment of private health insurers. The 1999 introduction of a 30% tax rebate for private health insurance creates a clear interest by the Commonwealth Government in whether health insurers are purchasing services efficiently. However, recent boosts in membership following the introduction of lifetime community rating create a countervailing pressure to maintain the value proposition of private health insurance through unrestricted choice and access to private services.
The evolution of purchasing among Australian health insurers subsequent to the 1995 legislation has received no direct consideration in the academic literature. Since 1999 the Australian Competition and Consumer Commission (ACCC) has produced regular reports that examine contracting by health insurers to assess the extent of any anti-competitive practices by either health insurers or health providers which may have a detrimental effect on consumers \[[@B10]\]. While these reports constitute a rich resource of the views of stakeholders on contracting, they do not directly examine how health insurers are attempting to meet the purchasing challenge of \'getting the best care for the best price\'. Industry conferences, together with trade publications, provide some partial and tantalising glimpses of the evolving purchasing behavior of health insurers, but again do not provide a comprehensive picture.
Accordingly, a study was designed to address this gap in our knowledge of whether Australian health insurers are buying best value health care for their members. Semi-structured interviews were conducted with key decision-makers across all Australian private health insurers to elicit their views on the evolution of purchasing across the dimensions of quality and coverage of new technology (\'best care\') and payment models (\'best price\'). This paper reports the findings relating to payment models. A future paper will provide full details of the study methodology, together with reporting the \'best care\' findings on quality and coverage.
Results
=======
Health insurers were initially asked to describe their current payment models including the perceived benefits and risks of different models, and how their approach to payment policy had changed over time. This led to more expanded discussions on the challenges associated with purchasing private health services including the regulatory framework for health insurance, relationships with private hospitals and issues with contracting. The findings from these interviews are presented under three main themes.
First, the purchasing environment is examined from the perspective of insurers, including their views on options for contracting, supplier-induced demand, gap payments, the impact of regulation on contract negotiations, and their relationships with private hospitals. The second section describes the current status of payment models across the industry including issues and barriers. Finally, the third section identifies examples of innovation in purchasing by private health insurers with a view to highlighting potential future developments in payment models.
Purchasing environment
----------------------
### Selective contracting
Health insurers were acutely conscious that government policy changes to encourage higher levels of health insurance membership had a direct influence on public perceptions about the value of insurance, and hence impacted on their ability to exercise greater purchasing discipline. A critical dimension of pro-active purchasing of health services involves the ability to select providers on the basis of criteria such as quality and efficiency in service delivery. However, this implies some restriction in choice of service providers for members, a difficult message as noted by one insurer:
\"Funds tend to have a philosophy of wanting to offer a wide a choice as possible, and at this point in time, after the initiatives of lifetime health cover and all that entails, and the 30% rebate, this is not a good time to be selectively contracting, because it means that you may restrict access.\"
In addition to anticipated consumer resistance, some insurers believed that previous selective tendering processes by other insurers had generated \"a lot of angst\" and \"mistrust\" among private hospitals. One insurer noted that the concept of selective contracting was almost a \"taboo\" topic. Moreover, insurers argued that government regulation, requiring insurers to pay a second-tier default benefit to hospitals that did not win contracts (equal to 85% of contracted benefit payment), undermined their negotiation ability and resulted in \"propping up\" private hospitals.
\"So second tier and portability are going to make the competitive tension very difficult in the industry, very difficult\....It needs some sort of competitive tension to, if you wish to direct your business or have some sort of preferential arrangements.\"
Insurers also noted that the potential for some \"iconic hospitals\" to not gain contracts had major public and political risks as these disenfranchised hospitals used the media to argue against the merits of contracting decisions by insurers. Despite general reservations about the difficulty in refusing to contract with certain hospitals, two insurers gave examples of situations where they had confronted proprietors planning to open new facilities to inform them that they would not receive insurance benefits. One case was based on the insurer\'s view that there was an adequate supply of psychiatric services in a geographic region, while the other case related to concerns about establishing a cardiac catheterisation laboratory in a small hospital with no immediate access to supporting intensive care or coronary care units. It appeared in both cases that this pre-emptive positioning by insurers had halted further development, although private hospitals could have simply ignored this threat, established the services and proceeded to claim second tier default benefits.
### Supplier-induced demand
One of the reasons why insurers expressed concern about not being able to exercise greater purchasing power through selective contracting was because they believed that providers could often generate demand for unnecessary health care services. This view was expressed most succinctly by one insurer:
\"Access equals demand. You open it and it\'ll be full and it\'ll be new demand.\"
However another insurer challenged this view, citing the spare capacity in many private hospitals and examples of private hospitals that had experienced financial difficulty due to low occupancy. Most insurers expressed reservations about the lack of government planning or quality controls for new hospitals, singling out the growth in day procedure facilities that they suggested were \"jumping up everywhere\".
The lack of success by insurers in having medical services included in their contracts with private hospitals (the \'practitioner agreements\' provisions in the 1995 legislative amendments) has also stymied their ability to influence service provision levels. Insurers noted the potential for over-servicing, with one respondent arguing that loose arrangements between private hospitals and doctors, together with the lack of a \"strong teaching environment\" created the preconditions in which inappropriate or unnecessary care could be provided. Insurers noted that entrepreneurial claiming behavior by some health providers required constant vigilance in benefits management.
\"The incentives that exist to generate income are quite strong and as a result, you get some pretty poor behavior, sometimes fraudulent, but certainly creative as to how you would generate income as an individual\... So we have a whole industry now of what we call benefits limitation, benefits management.\"
### Consumer co-payments
Since 2000, government regulation has required that private health insurance for medical services related to a hospital admission be provided on either a \'no gap\' or \'known gap\' basis \[[@B11]\]. While there is no similar legislative requirement on hospital costs, interviews confirmed that the widespread industry practice was that hospital contracts were negotiated to minimize consumer co-payments. As with selective contracting, this was justified on the basis that co-payments would \"water down the value proposition\" of private health insurance. Only one major insurer had a policy of explicitly using hospital co-payments to \"promote competition\" and to \"send a message to consumers\" to reassess the medical necessity of their use of hospital services.
Generally, insurers noted that even members who had chosen to purchase cheaper insurance products, with front-end deductibles or specified hospital co-payments, often became upset when faced with having to meet these costs themselves on admission to hospital.
Perceptions of public aversion to insurance co-payments meant that most insurers were reluctant to establish a system of tiered hospitals, that might encourage members to use preferred hospitals with zero co-payments compared to other hospitals with higher co-payments. Insurers argued that their members would view this as discriminatory and that the complexity of such a payment system would make health insurance more confusing. While some insurers distinguished between \'participating\' and \'non-participating\' hospitals, this tended to be related to insurers largely contracting with hospitals in the service area of their members.
### Impact of regulation on purchasing ability
Insurers often referred to the complex web of Commonwealth Government regulation of their industry and its impact on their purchasing behavior.
\"Purchasing is certainly not easy with any sector. It\'s never quite black and white; there\'s a whole range of other issues than price. So it\'s not like you\'re going out and buying a simple can of baked beans. But it\'s certainly made a whole lot more difficult when you\'ve got legislation that actually prevents you from doing a good job of purchasing.\"
The most frequently cited examples of problems with regulation related to second tier default benefits, reinsurance, outreach services and prostheses. Of these, the regulation of outreach services provides a useful illustration of the difficulty in reconciling conflicting policy objectives. Government regulation has established a reinsurance pool that redistributes the costs of particular groups across all insurers, in order to support the principle of community rating and ensure that insurers with higher claiming members do not face a spiralling cycle of adverse selection. However the reinsurance arrangements only apply to services provided under hospital insurance products, not ancillary products, meaning that services not directly associated with a hospital admission are not eligible for inclusion in the reinsurance pool. To overcome the consequential disincentive to provide effective substitutes to hospital-based care, the Commonwealth Government enacted further regulation establishing a tightly prescribed framework under which outreach services could be provided.
However insurers were critical of the resulting approval process needed for outreach services, noting:
\"And it just seemed to be completely an aberration of going back to, God knows how many decades ago, where getting out of the hospital walls, you\'ve got to have this huge bureaucratic process just to get a home-based care program, which has been running maybe in the public system for years.\"
Another problem raised by insurers was that the outreach legislation only recognizes outreach programs that are provided by hospitals, whereas insurers may want to purchase these services directly from community-based providers.
### Relationships with the private hospital sector
Insurers often commented on the mutually dependent relationship they had with private hospitals, arguing that they were \"working for the common cause\". One insurer linked the financial viability of private hospitals and health insurers as follows:
\"Value for money in a health fund product is important, so if you have an argument with a hospital that says: you get more business if we have more customers, if you charge us too much more, then we can\'t keep our customers. So we can actually improve your profitability and keep you full, but only if we don\'t get gouged. So make profit off volume, by all means, but don\'t make profit off individual patients, because the number of patients off who you\'ll be able to make a profit will diminish rapidly over a short period of time. So we have to live symbiotically here.\"
The importance of building mature relationships with private hospitals was also raised. It was argued that this required insurers to have \"credibility\" and be viewed by private hospitals as \"member and patient focussed, not just money focussed\". Credibility included having the strength of being principled, that is, \"we actually do as we say we\'ll do\", but also being open to discussion with private hospitals. In turn, insurers noted that they were more prepared to consider funding new service models if they had \"faith\" in particular private hospitals and there was \"goodwill on both sides\".
One insurer stressed that contracting was only part of their relationship with private hospitals, which needed to be maintained over time irrespective of whether they were in or out of contract with individual hospitals. Another insurer lamented the focus on \"perpetual contract negotiation\" as detrimental to the ability to build strong relationships, arguing that it would be preferable to move to long term contracts. Most insurers spoke of annual contract negotiations with providers, linked to the inability to accurately estimate future cost growth.
Current status of payment models
--------------------------------
Payment models can be classified along a continuum according to the unit of reimbursement \[[@B12]\]. Figure [1](#F1){ref-type="fig"} illustrates that purchasers face greater financial risk with relatively unbundled payment models, such as fee-for-service or daily payments as providers have an incentive to increase the level of outputs. In contrast, financial risk is transferred away from purchasers to providers through payment models such as capitation models (where the unit of reimbursement is the patient) and case payments or episodic models. (see figure [1](#F1){ref-type="fig"})
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Payment models for hospital services.
:::
:::
Interviews revealed that there is currently considerable variation across private health insurers as to the payment models used as the basis of hospital contracts. Figure [1](#F1){ref-type="fig"} also illustrates the two \'typical\' payment models used by health insurers for most hospital services.
Most insurers still rely strongly on a \'traditional\' payment model that involves separate payments for each of the components of hospital care including accommodation, theatre, prostheses, pharmaceuticals and intensive care unit (ICU) costs. Insurers usually refer to this as a per diem payment model. It is based on classifying patients according to the type of medical service provided using Medicare Benefit Schedule (MBS) items, with payment step-downs to encourage shorter lengths of stay. When viewed against the payment model continuum, it can be seen that this traditional model is actually more similar to an item of service payment model than a true per diem model.
The second typical payment model is the \'bundled DRG case\' payment model where most hospital admission costs are bundled into a single payment. Currently only two major insurers (representing about one-quarter of the insurance market) are using an almost fully bundled DRG model (with bundling of theatre but separate payments for prostheses) for the majority of their acute episodes. Two other major insurers (representing about one-half of the market) had undertaken substantial developmental work and were planning significant expansion in their use of DRGs. However, one crucial difference in the use of DRGs between the private and public sectors is that medical costs are specifically excluded in the private sector, with DRGs only applying to non-medical services.
Within these two \'typical\' payment models, there is scope for numerous permutations. For example, some insurers relying largely on per diem payments still operate a more bundled case payment model for a limited number of conditions with relatively predictable costs, such as obstetrics, dialysis and endoscopy. One insurer referred to bundling in the cost of low cost prostheses in a predominantly per diem payment model. Of the two insurers using a bundled DRG case payment model, one bundled in ICU costs.
### Attitudes and barriers to adoption of DRGs
Insurers, irrespective of their payment model, commented on some of the advantages of using DRGs relative to the predominant per diem model including:
• \"DRGs are the only thing that is statistically coherent and recognizes all elements of care and you really can\'t say \'we\'re different\'.\"
• \"And there\'s considerable scope, we believe, something like a 30% scope for reduction of length of stay in the private sector.\"
• \"The hospital has an opportunity to do more, to innovate and to make things more cost-effective, in the way they can and they gain by that. And we gain by a greater predictability.\"
• \"I think it\'s crazy when you benchmark by DRG and then pay by per diem. I think there has to be consistency between your benchmarking and your payment model.\"
• \"So our aim was to try and get the hospitals onboard, to try and share the risk, that\'s our big basis behind our case payment, share the risk.\"
The relatively slow progress in uptake of DRGs relative to the public sector was attributed to a wide range of factors including limited data and poor understanding of the cost structures of private hospitals, the lack of familiarity and understanding of DRG based payment models, resistance from the medical profession, and concerns about potentially premature discharge.
Of these, the lack of good costing systems was most frequently mentioned, with insurers noting that \"the industry is very much historically based\". Accordingly, some insurers referred to using \"price weights\" based on historical prices, rather than cost weights. Most insurers were keen to get greater access to private hospital cost data. However a dissenting view was offered by one insurer who cautioned that \"the more you get involved in their cost structure, the more you become a funder\". This insurer was hesitant about moving from a purchasing role to a situation where insurers might be regarded as being responsible for the financial sustainability of contracted private hospitals. Insurers also suggested that some private hospitals did not have the capacity to assess properly how efficiency gains flowing from the use of DRGs could positively impact on their operating margins.
\"And what we\'ve been trying to do is get the facilities to look at their margins, rather than their revenue. Their pure revenue increases. And not many of them are at that level of sophistication. Because we believe and can show some of them that if they do things differently, they can actually get a greater margin, rather than just increases in pure revenue. But they don\'t think like that, many of them.\"
Insurers reported that providers sometimes equated the use of DRGs with insurers taking on an undesirable \'managed care\' role.
\"So even to bring in a casemix funding system which is already out there, and has already been in the public system for years, we got pushback from the AMA saying: \'that\'s managed care\', because it was imposing a length of stay through the inlier type separation. And those sorts of barriers are just so alive and well because of the nature of the power that the doctor has in the private sector.\"
The medical profession has long resisted entering contracts with health insurers \[[@B11]\] and this insurer\'s comments highlight the power of doctors to influence other payment arrangements between insurers and hospitals. From the perspective of insurers, the exclusion of medical costs from DRGs in the private sector is likely to reduce the potential efficiency gains and may partially explain the slower adoption of DRGs in the private sector.
Insurers who had introduced DRGs considered that they had done so in a cost neutral manner, without clawing back the ensuing efficiency savings made by private hospitals. This was motivated partially by concerns about avoiding the \'quicker and sicker\' phenomenon, often associated with the use of DRGs.
Several insurers noted that there was now considerable interest, and indeed pressure, from many private hospitals for the more uniform adoption of DRGs across private health insurers.
\"I think they want homogeneity. If you look from a provider\'s perspective\.... the thing that you hate the most is that each fund has a different methodology about how you submit your claims and how you get paid\... So their perspective is, if it standardizes along the lines of a casemix type structure, I think their view, and the Australian Private Hospitals\' Association view, is that that would be a good thing. If they can achieve some back office efficiencies, then that would help them in the management of their business and it would also assist them into getting into e-claiming. That\'s where they want to get into. It has huge efficiencies for them, and for us, but that needs, largely, a similar contracting, purchasing base.\"
Innovation in purchasing by health insurers
-------------------------------------------
Bundled DRG case payments are one mechanism by which purchasers can encourage providers to seek new approaches to delivering care more effectively, through transferring risk to providers. However, Figure [1](#F1){ref-type="fig"} shows that there is potential beyond case payments to encourage greater allocative efficiency through paying providers on the basis of capitation or episodic payments.
Insurers who have experience of using DRGs are more likely to become aware of some of the limitations of DRG case payments. These include ensuring that providers take responsibility for pre-admission and post-discharge services, or the fact that DRG case payments still contain inherent incentives for the production of unnecessary hospital admissions. In response, these insurers might move towards greater innovation in purchasing by trialling payment models that require providers to assume financial responsibility for an expanded scope of care.
In fact, interviews for this study found that most innovation in purchasing was concentrated in one of the two major insurers that had already implemented a bundled DRG case payment model. Table 1 provides an outline of four examples of innovative purchasing developed by this insurer. Two of these examples, episodic management units and the members extended care arrangements, involve the provider assuming greater responsibility for services over an extended period of time relative to the narrow DRG case definition. The capitation model involves the psychiatric services provider receiving a fixed payment for each insured member, irrespective of the actual services provided. (see [Additional file 1](#S1){ref-type="supplementary-material"})
While all these innovative purchasing models originated with one insurer, several other insurers had followed its lead in adopting the same capitation model for psychiatric services in one state. At interview, the second major insurer using a bundled DRG case payment model did not outline similar payment models focussed on transferring risk, but instead emphasized the importance of clear specification of required quality parameters in hospital contracts.
Insurers were also asked about their views on innovative payment models that explicitly link purchasing to quality. This might include pay for performance frameworks, or only purchasing services if hospitals undertook at least the minimum volume recommended to enhance safe patient outcomes.
### Volume-outcome purchasing
Several insurers had undertaken exploratory work on the potential introduction of \'volume-outcome\' purchasing. In particular, one insurer was cautiously beginning to apply this framework in purchasing cardiac, neurosurgery and obstetric services. As one insurer noted:
\"If you\'ve got good volume, you don\'t necessarily have good quality, but if you\'ve got a small volume, you\'re really asking for trouble\....We\'re moving to the view, we\'re saying that if you haven\'t got enough cases to meet industry norms of what\'s a critical minimum volume, then we\'re not going to pay you. We want you to get out of this business. Now that\'s stepping on toes, it\'s not very popular.\"
Insurers recognized that volume-outcome purchasing may be perceived as a threat to the professional autonomy of individual clinicians. One insurer noted that some medical colleges or professional associations appeared to be reluctant to specify recommended minimum volumes, while another suggested that \"the challenge is actually who will be the arbiter of just what that volume will be\". Another insurer suggested that volume-outcome purchasing might lead to a \'bidding war\' between private hospitals to poach high profile clinicians and establish centers of excellence, as follows:
\"So what you\'ve got is the hospital operators working very much with the doctors, trying to keep them happy. So all of a sudden, gee, I\'d actually like to have lobster on Friday in future. You laugh, but it actually does happen. Some doctors insist on what they\'re going to have provided to them for lunch while they\'re on their days there. And if they don\'t get it, no qualms about packing up and going to the next hospital. They\'ll take their patients with them.\"
Another more practical barrier to implementing volume-outcome purchasing is that data on the volume of procedures done at individual private hospitals is never publicly released by government, as this information is considered commercially sensitive. Insurers can only access such data if private hospitals voluntarily agree to provide it during contract negotiations. However, insurers noted that measurement of volume was complicated by several factors including the fact that specialists may operate at several hospitals and that some of the literature also stressed the importance of teams, not just individual clinicians, in achieving better outcomes. The impact of volume-outcome purchasing on centralizing services, with reduced geographic access in more rural locations, was also mentioned as a concern.
Finally, the potentially negative impacts on both reducing consumer choice and increasing member confusion were highlighted by this insurer.
\"For instance, we\'ve talked about what would it look like if we recommended a particular facility, you know, for maternity and another for orthopaedics. Very hard to explain to members. We think we could do that, but members still really see us and it\'s our responsibility is to see that we are credible as a health fund. And they essentially see us as a payer of claims, and interfering with their choice? There are ways you can structure your products and things differently, but again the product becomes much more difficult to communicate. And one of the great criticisms of the industry, and it is, is that it\'s complex. It is difficult to explain to members their products and the relationships with providers anyway.\"
### Pay for performance
There was growing interest in the use of pay for performance models. One insurer was about to commence contract negotiations using a pay for performance model that linked payment levels to the achievement of quality standards. Several other insurers expressed interest in trialling pay for performance models in the near future. A perceived benefit was the potential of pay for performance models to promote the early adoption of practice changes such as sentinel events reporting or computerised order-entry systems for pharmaceuticals.
\"Some of the things we might consider is, whether using it a bit like the Leapfrog Group, whether you can use a pay for performance approach in a sense to encourage early adoption of standards\....We may look at some rewards upfront, and perhaps scaling the rewards, as it is with Leapfrog, so it\'s a loading of say 4% in year 1 and 2% in year 2, and sorry, it\'s going to become standard practice by year 3. So if you get your act together earlier, you\'re actually going to benefit from that.\"
While insurers were generally positive about the value of using financial incentives \"to pursue good practice\", they acknowledged that, as with volume-outcome purchasing, there was likely to be professional resistance. Moreover, it was suggested that this might result in regulatory intervention, with the government setting boundaries on what was considered to be acceptable purchasing behaviour by health insurers.
Discussion
==========
Progress towards adoption of DRGs in the private sector
-------------------------------------------------------
Even prior to the 1995 legislation enabling insurers to contract with private hospitals, there had been considerable effort invested in investigating the potential adoption of DRGs. Further to the establishment of its Casemix Development Program, the Commonwealth Government commissioned a comprehensive 1990 report that identified options for introducing DRGs in both public and private hospitals \[[@B13]\]. However, in 1991 the Australian Private Hospitals Association urged in relation to introducing DRGs, that the private sector should \"slow down the indecent haste towards radical change to allow us to get it right the first time and avoid chaos\" \[[@B14]\]. In 1994 private insurers, private hospitals and Commonwealth Government officials jointly developed the so-called \'Gold Book\', essentially an agreed implementation guide for how casemix using AN-DRGs should be introduced into the private sector \[[@B15]\].
Using the Gold Book blueprint, BUPA Australia (or National Mutual Health Insurance as it was at that time) was the first private health insurer to implement an episodic case payment approach using AN-DRGs in Victoria in 1997 and South Australia in 1998. Interestingly, these two states were also the first states to adopt DRGs in the public sector. This may have facilitated uptake of case payments in private hospitals in these states, due to the familiarity of clinicians working in both public and private hospitals. Speaking at the eleventh annual casemix conference in 1999, a BUPA Australia representative indicated that the changeover process from a per diem to a DRG case payment had only been achieved by virtue of BUPA Australia mandating DRG payments in its selective tendering process for private hospital services \[[@B15]\]. Similarly, by 2000 MBF was able to publicly report that its competitive tendering process in Queensland had resulted in a major expansion in its use of DRGs for Queensland private hospitals \[[@B16]\].
This study has found that in 2004 these two companies are still the only insurers to use DRGs as the basis of purchasing most of their acute hospital services. Hence the question arises as to why the other insurers have not yet implemented DRGs, given both the 10--15 year history of developmental work and the apparent success of two leading insurers in paving the way in the private sector. The answer seems to be complex and multi-factorial, with the reasons for slow progress varying between insurers.
On the surface, the non-DRG insurers most commonly attributed the problem to concerns about the quality of private hospital cost data, claiming that the industry was historically based and that neither insurers nor private hospitals had access to adequate costing data. However, given that MBF and BUPA Australia were apparently able to overcome these problems between four to seven years ago, it is not obvious that this argument provides sufficient explanatory power. In fact, the leadership by two companies in implementing DRGs should have resulted in substantial improvement in private hospital costing data, noting that BUPA Australia and MBF have significant market shares between them in Victoria, South Australia, Queensland and New South Wales \[[@B17]\]. The advocacy by some private hospitals for greater uniformity in uptake of DRGs suggests that these private hospitals have confidence in their costing systems and their ability to manage in a DRG funding environment. For some of the non-DRG insurers, there seemed to be a \'chicken and egg\' problem with their lack of confidence in private hospital cost data related to their lack of internal experience in working with and analysing DRG data. In contrast, BUPA Australia \[[@B15],[@B18],[@B19]\] and MBF \[[@B16],[@B20]-[@B22]\] have regularly documented at annual casemix conferences their significant investment in casemix development.
In 1998 the Australian Health Insurance Association, argued in relation to private sector uptake of DRGs, that there was a \"culture of inertia\" with both health system providers and funders being \"relatively conservative by nature\" \[[@B23]\]. This explanation continued to ring true in 2004 for many (but not all) of the non-DRG insurers based on interviews for this study. The culture of these insurers appeared to be inherently conservative and risk averse, as reflected in their general approach to purchasing. For example, several insurers spoke of the challenges in getting support from their Boards for major changes to their purchasing frameworks. Some of these non-DRG insurers also seemed to be less confident of their ability to use their market power in driving payment model reforms, often referring to the countervailing power of both private hospitals and medical professionals.
In relation to market power, it is of interest to note that both BUPA Australia and MBF introduced DRGs in the context of competitive tendering processes. Selective contracting represents a major shift in the power dynamics between insurers and private hospitals, unleashing the power of insurers to be purchasers rather than \'passive bill-payers\'. One insurer was clearly uncomfortable with this use of market power, commenting that the BUPA Australia and MBF tenders had generated \"a lot of angst and a lot of dislike\". However, another insurer welcomed their role in pioneering change across the industry, noting that BUPA Australia was well-regarded for its commercial adeptness and MBF was viewed as a strong leader in purchasing for quality.
Role of current payment models in \'buying best value health care\'
-------------------------------------------------------------------
Based on the continuum of payment systems featured in Figure [1](#F1){ref-type="fig"}, the assumption behind this study is that greater use of DRGs would enable health insurers to be more effective in \'buying best value health care\'. From a technical efficiency perspective, the advantages of DRG case payments over charging fees for individual items or per diem rates are well accepted. For example, the OECD in its most recent review of high-performing health systems has argued that prospective, case-related payment systems offer significant benefits in inducing hospitals to seek productivity improvements \[[@B24]\].
However it is evident from this study that the payment models used by most Australian private health insurers represent relatively unsophisticated approaches to purchasing. In addition to most insurers not using DRG case payments, innovation in developing new purchasing models beyond casemix was largely concentrated in one major insurer, BUPA Australia. The culture of conservatism seemed to limit the willingness of some insurers to engage in learning through implementation of payment model reforms. In 1997 a BUPA Australia representative argued in relation to the development of new funding models for rehabilitation services that:
*\"The expectation is that we will learn as we proceed. We are unlikely to learn as much by standing still.\"*\[[@B18]\]
In 2004 this spirit of learning, including learning through making mistakes, did not appear to be a strongly defining characteristic of many private health insurers.
Another interpretation of the findings of this study is that health insurers view their purchasing role as having to balance competing objectives, only one of which is about improvements in technical efficiency. Insured consumers may consider that \'best value\' incorporates access to a wide range of hospital services at the time of their choosing, assurances about the quality and safety of health services, and no intervention by their insurer to limit health service choices made by the patient or their doctor. As reported earlier in this paper, health insurers frequently referred to the challenges of maintaining the value proposition of health insurance. More specifically, the use of consumer co-payments, selective contracting of private hospitals and volume-outcome purchasing were viewed by some insurers as potentially reducing the value of private health insurance.
Relationship between regulation and purchasing
----------------------------------------------
This study highlighted some of the frustrations experienced by insurers in moving to a more proactive purchasing role, encapsulated in the complaint about \"legislation that actually prevents you from doing a good job of purchasing\". Examples of regulation viewed as problematic by insurers included the requirement that insurers pay second-tier default benefits to non-contracted hospitals, cumbersome approval processes for purchasing outreach services and the impact of reinsurance in inhibiting substitution between hospital and community-based services.
However, it is not uncommon for industry generally to use regulation as a convenient scapegoat to justify public concerns about its performance. If regulation were really a significant barrier to improved purchasing behavior by private health insurers, it would be expected that it would impact equally on all health insurance companies. Instead, Australian health insurers exposed to an identical regulatory environment have developed very different approaches to purchasing health services for their members. The innovative purchasing models used by BUPA were developed despite the negative impacts through reinsurance arrangements for BUPA of some of these models.
While it is clearly desirable for regulation to promote effectiveness and efficiency in purchasing so that insured people get \'the best care for the best price\', the regulatory environment is not the only factor influencing purchasing behavior. In particular, regulation cannot trump institutional history and culture, nor can it create savvy, entrepreneurial insurers that are skilful in parlaying relationships with health service providers into best value health care for their members.
The relationship between regulation and effective purchasing is complex. Most commentators in the health arena recognize the need for a significant role for government in regulating health markets due to well accepted market failures such as information asymmetry, externalities and the potential for supplier induced demand \[[@B25]\]. It has previously been noted that the *National Health Act 1953*lacks clear and coherent policy objectives that explain the purpose of government regulation of this industry. The theoretical rationale for government intervention can include promoting consumer access to affordable and attractive products, improving accountability to consumers, encouraging competition and efficiency in the private health sector, and promoting the financial sustainability of the sector \[[@B26]\]. Insurers noted the tension in balancing some of these objectives, including the trade-off between broad choice of providers and services, low co-payments and affordable health insurance premiums.
While the first priority must be greater clarity about the policy objectives behind private health insurance regulation, the regulatory framework for purchasing could also be improved. The approval process for outreach services is an example of a detailed command-and-control approach to regulation that would no longer be regarded as best practice \[[@B27]\]. A preferred approach to private health insurance regulation generally would be greater adoption of performance-based regulation. This would clearly specify the desired outcomes but allow private health insurers the opportunity to be flexible and innovative in how they achieve these outcomes \[[@B28]\].
Conclusion
==========
Since the passage of the 1995 legislation allowing health insurers to take a more pro-active role in purchasing, health insurers have sought to balance the competing demands of private hospitals, medical professionals, consumers and government. While health insurers see the relationship with private hospitals as symbiotic, some have concerns about the potential for unnecessary medical servicing and the lack of transparency on private hospital costs. There are also mixed views about the merits of pursuing a more aggressive policy on selective contracting of private hospitals, notwithstanding that competitive tendering has allowed two major insurers to implement more sophisticated payment models. Progress towards greater adoption of case payments using DRGs has been much slower than anticipated. Industry conservatism and resistance to change, coupled with inadequate investment in hospital cost analysis, may have contributed to this situation. However, the foreshadowed move by two other major insurers to base their purchasing strategies on DRGs, together with the growing demand from private hospitals for uniformity in payment models, may finally swing the pendulum towards more effective purchasing of best value care.
Competing interests
===================
The author(s) declare that they have no competing interests.
Supplementary Material
======================
::: {.caption}
###### Additional File 1
Table 1, which shows 4 different payment models (e.g. Interim care, EMU) in use by one insurer.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
This study was made possible by the willingness of the health insurance fund executive managers to share their experience and views. Their generous participation, together with valuable comments provided by reviewers, is gratefully acknowledged, although the conclusions expressed are solely those of the author. No official endorsement by the Victorian Department of Human Services is intended.
|
PubMed Central
|
2024-06-05T03:55:55.152712
|
2005-3-31
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079790/",
"journal": "Aust New Zealand Health Policy. 2005 Mar 31; 2:6",
"authors": [
{
"first": "Sharon",
"last": "Willcox"
}
]
}
|
PMC1079791
|
Background
==========
The National Library of Medicine (NLM) created PubMed, which is one of the largest literature databases in biomedicine. The database contains nearly 15 million references from more than 4,800 biomedical journals. In general, authors of scientific articles are addressed by their last name and forename initial. For example, a search for \"Lee C\", a common Chinese name, would retrieve over 9000 hits (Fig [1](#F1){ref-type="fig"}).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**PubMed search for \'Lee C\'**. A simple search for \'Lee C\' results over 9000 articles from PubMed.
:::
:::
Discussion
==========
In general, names can be too common and not unique enough to be search criteria. Modern life advances this point. In former times only a few scientists, normally professors in universities and academic health sciences centers, published scientific results of their work under their names. Most scientists were men. Their names were consistent and did not change by marriage. Today, Ph.D. students, other researchers and women publish scientific work. Women may be using a married name, which may be a compound name for authorship. Thus a person may not only have one name but several names and publish under each name. In Germany, there can be nameconfusion for titles of nobility from former times or for composed names. Claus-Wilhelm von der Lieth (with \'Claus-Wilhelm\' as forename and \'von der Lieth\' as surname) is cited in several mutations, such as \'Vonderlieth CW\', \'Lieth von der CW\' or only \'Lieth CW\'.
Especially with a view to the future, such a Unique Scientist ID could help to address people in peer-to-peer (P2P) networks. One such network based article reference P2P network is Bibster \[[@B1]\]. Bibster is a Java-based system which assists researchers in managing, searching and sharing bibliographic metadata in a peer-to-peer network.
How can one easily find articles written by a particular person? A Unique Scientist ID could help to address persons. A Unique Scientist ID should contain all the versions of a scientist\'s name. As a starting point, perhaps PubMed could generate and manage such a scientist ID. Adding the Scientist ID should become as routine as adding an email address to the article\'s citation.
Conclusion
==========
A Unique Scientist ID would improve knowledge management in science. At the moment authors of publications are only addressed in most online databases by name, initials and address. Unfortunately in some of the publications, and then within the online databases, only one letter abbreviates the author\'s forename. A search for a special author is successful if the name is quite unique or a combination of authors is used for the search. In such a case, a common name with only one initial could retrieve pertinent citations, but include many *false drops*(retrieval matching searched criteria but indisputably irrelevant).
Authors\' contributions
=======================
Both authors contributed equally.
Competing interests
===================
The author(s) declare that they have no competing interests.
|
PubMed Central
|
2024-06-05T03:55:55.157190
|
2005-3-22
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079791/",
"journal": "Biomed Digit Libr. 2005 Mar 22; 2:1",
"authors": [
{
"first": "Andreas",
"last": "Bohne-Lang"
},
{
"first": "Elke",
"last": "Lang"
}
]
}
|
PMC1079792
|
Introduction
============
Within one month Google announced two projects that will have profound implications for the future of librarianship.
First up in November 2004 was *Google Scholar*\[[@B1]\]. Currently in beta, *Scholar*aims to provide access to scholarly materials via the crisp and familiar Google search box. The results pages display the number of other citations to the resource in the *Scholar*database, in a manner reminiscent of the \"cited by\" search feature within Thomson-ISI\'s *Web of Science*.
The unveiling of *Scholar*caused a flurry of excitement, and even the creation of the somewhat pretentious blog \"On *Google Scholar*\" \[[@B2]\]. At New York University Medical Center a doctor spontaneously brought up *Scholar*in conversation with me, and it also stimulated discussion at an international conference about grey literature in early December.
Soon the *Scholar*buzz was overshadowed by the December announcement that Google has entered into a partnership to digitize the materials of five leading research libraries: Harvard, Oxford, Michigan, Stanford, and the New York Public Library (NYPL). Terms of agreement vary between libraries. For example, Michigan and Stanford will provide access to the full range of their materials, while Harvard has authorized a pilot of 40,000 volumes. Depending on the copyright restrictions of the material in question, searchers will be able to browse all or part of it.
The principal rationale for this project is that it will democratize access to the intellectual resources of elite institutions. In addition, integrating library resources into Google will hopefully entice those students who might never consult a library catalog. To reach these students, Harvard plans to develop a seamless link between Google searches conducted at Harvard and Harvard\'s online catalog \[[@B3]\].
The library material represents a radical expansion of the *Google Print*program \[[@B4]\]. Searchers would not search for books specifically; instead, Google would highlight books within the results of a normal Google search. The library material will support the same e-commerce stream as the rest of *Google Print*. Contextual advertising would be integrated into the search results, and it is likely that searchers will be pointed to online book vendors.
Whereas the excitement about *Scholar*was concentrated in research circles, the *Google Print*projects received widespread public attention. The *New York Times*considered this to be the lead news item for December 14, and it was a major topic on the *National Public Radio*(NPR) show \"Talk of the Nation\" on December 15. The show\'s guests included Michael Keller, Stanford University Librarian, and Brewster Kahle, founder of the Internet Archive.
The Internet Archive launched a very similar digitization project as *Google Print*on the same day, which was buried in the flood of news about Google \[[@B5]\]. Kahle\'s efforts are worthy of wide promotion. His project has none of the nettlesome concerns facing *Google Print*, which I will describe later.
*Google Print*continues to generate significant discussion. One recent example is the March 2005 issue of *American Libraries*, which features a colloquium entitled, \"Google at the Gate\" \[[@B6]\]. As all librarians know, Google is the default search engine for millions of users. Because of this, it is essential that we critically examine both the benefits and shortcomings of *Google Scholar*and *Google Print*.
Discussion
==========
*Google Scholar*
----------------
The initial version of *Google Scholar*had numerous flaws. Peter Jasco excellently documented these shortcomings, while promising to \"write a hagiographic review about *Google Scholar*when it is done, and done well\" \[[@B7]\]. Some of the shortcomings were that Google was unacceptably vague about its sources, and could not eliminate duplicates from search results.
Librarians must monitor the evolution of *Google Scholar*, and educate our patrons about its limits. Google\'s target audience for *Scholar*is \"those in academia whose work has made Google itself a reality,\" and therefore Google aims to make *Scholar*\"as useful to this community as possible\" \[[@B8]\]. With our assistance, researchers would be able to offer suggestions about functionality that they might not consider on their own.
The educational effort about *Scholar*has already begun. The Georgia State University Libraries have developed a straightforward web page, which includes a search box for *Scholar*, the library\'s e-journal list, and the library catalog \[[@B9]\]. It is easy to foresee this page blossoming into a class about using *Scholar*, one goal of which might be to increase patron appreciation for the challenge of providing access to electronic scholarship. As patrons use *Scholar*and discover the barriers to obtaining research articles, they could be more receptive to the argument for open access publishing.
The \"cited by\" feature of *Scholar*presents another educational opportunity. An essential caveat that should be incorporated into this instruction is that *Scholar*\'s cited by algorithms are not yet fully reliable \[[@B10]\]. For example, I ran a *Scholar*search for \"bioinformatics,\" which returned \"about 62,600\" results within .06 seconds. I selected the third citation, which was cited by 88 other resources within *Scholar*. After 5 clicks I was down to one article, standing at the root of one chain of thematic connection.
My strategy on each screen was to click on articles that had less cited by citations than the article I selected on the screen before. Of course, I could have approached this in different ways, or tackled a different problem. The point is that *Scholar*provides such teaching opportunities, in an interface with which many people are already familiar.
A final educational opportunity I propose is a comparison of *Scholar*to products such as Elsevier\'s *Scirus*, which is a search engine exclusively focused on scientific research \[[@B11]\]. Mr. Reinhardt Wentz gave me this idea with a November 2004 posting to MEDLIB-L \[[@B12]\]. Unlike *Scholar*, *Scirus*differentiated a search for \"bioinformatics\" into journal and web page results, and also offered suggested search terms for refining these results. It also provides a \"similar results\" capability, which is analogous to PubMed\'s \"related articles\" feature. But it does not provide a \"cited by\" capability. A well-designed class would facilitate interesting discussion about the merits of these two approaches for identifying scholarly materials.
In addition to the educational work ahead, *Scholar*presents opportunities for librarian advocacy. In order to build comprehensive biomedical digital libraries, for example, it is essential that *Scholar*provides access to visual as well as textual material. Prime candidates for inclusion are the multimedia resources currently indexed in the Health Education Assets Library, or HEAL \[[@B13]\]. The \"bioinformatics\" search did not yield any results in HEAL, so I broadened it to \"informatics.\" This had 24 results. These multimedia resources would enliven the results of a *Scholar*search, and also be placed into a broader research context than HEAL can provide on its own.
*Google Print*
--------------
*Scholar\'s*lack of maturity is not surprising, because it has existed for less than 6 months. But I am hopeful that it will improve and open up new avenues for library instruction. I am less sanguine about the implications of the *Google Print*project.
At first blush I was swept up by the positive publicity surrounding the project, because it is inspiring to contemplate the democratization of knowledge that has previously been sequestered inside some of the world\'s leading research libraries.
After I read Rory Litwin\'s essay, \"On Google\'s Monetization of Libraries,\" I was forced to tamper my enthusiasm \[[@B14]\]. Litwin argues that the e-commerce foundations of *Google Print*are antithetical to the principles of librarianship. Until now a library\'s resources have served as their own advertisement, but now they will become a vehicle for selling something else. And, of course, only a select group will be able to afford the items available. In its implementation, the much-heralded idea of democratization of knowledge will actually reinforce existing class distinctions.
Another concern about *Google Print*, as Litwin points out, is that it flattens the distinctions between materials that are used for different purposes. A chief reason universities select resources is because of their enduring value for scholarship; a chief reason Amazon stocks books is to make money. *Google Print*collapses this difference. My search for \"gardening\" might link to a priceless treatise by Linnaeus just above a link to Martha Stewart\'s annual review. In either case, I\'ll be able to order planting soil from Home Depot.
Conclusion
==========
My concerns with *Google Scholar*are structural, while those with *Google Print*are philosophical. Because the *Google Print*project is so enormous, it will be many years until it becomes a reality. In that time librarians should strongly advocate for ways to minimize the problems noted above. One simple solution would be to segregate the library books within search results from the other books. Assuming agreement with this suggestion, a more controversial idea would be to forbid advertising on the library results pages. If this is unacceptable, it might be possible to define acceptable categories of advertising for these pages. One model for this could be the advertising that appears on NPR and the Public Broadcasting Service.
My fear is that Google will reject such ideas, on the grounds that the library community knew what it was getting into. And Google would be right. In the admirable desire to improve access to their collections, some of our best libraries may have struck a Faustian bargain.
NOTE: All of the described searches occurred on December 23, 2004.
List of Abbreviations Used
==========================
NYPL = New York Public Library
HEAL = Health Education Assets Library
NPR = National Public Radio
Competing Interests
===================
The author(s) declare that they have no competing interests.
|
PubMed Central
|
2024-06-05T03:55:55.157779
|
2005-3-22
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079792/",
"journal": "Biomed Digit Libr. 2005 Mar 22; 2:2",
"authors": [
{
"first": "Marcus A",
"last": "Banks"
}
]
}
|
PMC1079793
|
Background
==========
Over the last three decades, methicillin resistant *Staphylococcus aureus*(MRSA) had caused major problems in hospitals throughout the world\[[@B1]\]. The first outbreak caused by MRSA occurred in European hospitals in the early 1960\'s\[[@B2]\]. During the late 1970\'s, strains of *S. aureus*resistant to multiple antibiotics including methicillin and gentamycin were increasingly responsible for many outbreaks in the U.S.A and Great Britain\[[@B3]\], and by 1980\'s MRSA was considered a major clinical and epidemiological pathogen in human hospitals\[[@B4]\]. Since then strains of MRSA and coagulase-negative *Staphylococci*had spread worldwide. Recent reports indicated that MRSA strains account for 10 to 40% of *S. aureus*isolated from some European hospitals\[[@B5]-[@B7]\]
In many parts of the globe, particularly the developed countries, fluoroquinolones (pefloxacin, ciprofloxacin and ofloxacin) are recommended for serious infections associated with *Staphylococci*, although, occasional resistance among MRSA has been documented\[[@B2]\], Furthermore, in spite of recent reports of vancomycin resistant strains MRSA in some parts of the globe, vancomycin still remain the drug of choice for most MRSA-associated diseases\[[@B9]\].
The use of medicinal plants all over the world predates the introduction of antibiotics and other modern drugs into Africa continent. Herbal medicine has been widely used and formed an integral part of primary health care in China\[[@B10]\], Ethiopia\[[@B11]\], Argentina\[[@B12]\] and Papau New Guinea\[[@B13]\]. Traditional medical practitioners in Southwest, Nigeria, use a variety of herbal preparations to treat different kinds of microbial diseases including MRSA-associated diseases.
In recent times, the number of traditional healers claiming the efficacies of six medicinal plants viz: *Terminalia avicennioides*, *Bridella ferruginea*, *Ageratum conyzoides Ocimum gratissimum*, *Acalypha wilkesiana*and *Phylantus discoideus*for the cure of patients with *S. aureus*-associated diseases such as, eczema, carbuncles and osteomyelitis is on the increase.
*Terminalia. avicennioides*(*Combretaceae*) is a yellowish brown, hard and durable wood. The roots which are used as chewing sticks have been claimed to cure dental caries and skin infections\[[@B14]\]. Previous studies, showed that the bark extract of *T. avicennioides*exhibited both vibrocidal and typhoidal activities\[[@B15],[@B16]\]. *A. conyzoides*(*Compositae*) is an annual herb abundant in preclusive forests and farmland in southern part of Nigeria. Previous study showed that methanolic leaf extracts corrected fibrinogaemia in poultry chicks. Also both methanol and water extracts of the leaves exhibited anti-bacterial effect\[[@B17]\]. *B. ferruginea*(*Euphorbiaceae*) is used for treatment of insomnia. The bark in combination with other herbs is used to cure pile in western part of Nigeria\[[@B18]\].
Like other *Ocimum spp*, *Ocimum gratissimum (Lamiaceae*) traditionally called \"Efirin-Aja\" has been reported to have medicinal properties. The leaf extracts are popularly used for the treatment of diarrhea while the cold leaf infusions are used for the relief of stomach upset and haemorrhoids. The thymol-riched leaf has been reported to have antimicrobial properties\[[@B19]\]. *Acalypha wilkesiana (Euphorbiaceae*) is popularly used for the treatment of malaria, dermatological and gastrointestinal disorders\[[@B20]\]. The bark extract of *P. discoideus*(*Euphorbiaceae*) is used locally to cure stomachache and lumbago.
The reputed efficacies of these plants have been experienced and passed on from one generation to the other. Apparently, lack of scientific proof of efficacies claimed by traditional medical practitioners in Nigeria called for this study. The present study investigates effects of these medicinal plants against Methicillin Resistant *S. aureus*.
Methods
=======
Fresh plant materials were collected from users of these plants in Imota, Ikorodu, Ojo and Kosofe in Lagos State, Nigeria. Their botanical identities were determined and authenticated by Mr. O. K. Oluwa in Botany department, Lagos State University, Ojo, Lagos. Samples were deposited in the department herbarium.
Extraction
----------
The extraction method used in this study was a modification of Akinside & Olukoya,\[[@B15]\] and Akinyemi *et al.*\[[@B16]\]. In line with the traditional methods of preparation, shredded plant materials were put in sterile bottles containing either distilled water or 40% ethanol.
Water extract
-------------
Leaves of *B. ferruginea*(P) and *A. conyzoides*, (Q), and stem barks of *T. avicennioides (R) P. discoideus*(S) *O. gratissimum*(T) and *A. wilkesiana*(U) were oven-dried at temperature of 60°C for 6 days. They were subsequently grounded into fine powder in 25 ml of sterile distilled water, maintained at 60°C for 3 hr. The resulting suspensions were filtered and evaporated to dryness at 60°C in vacuo to produce 0.031, 0.041, 0.038, 0.042, 0.044 and 0.035 g of P, Q, R, S, T, and U respectively. They were further labeled as aqueous extracts and designated as PW, QW, RW SW, TW, and UW respectively.
Ethanol extract
---------------
Six grams of each of the powdered plant materials were put in a soxhlet extractor containing 25 ml of 40% ethanol. The resulting extracts were subsequently weighed to produce 0.032, 0.036, 0.038, 0.040, 0.030 and 0.034 g of P, Q, R, S, T, and U respectively. They were labeled as ethanol extracts and designed PE, QE, RE, SE, TW and UW respectively.
Bacterial culture
-----------------
The MRSA strains used in this study were clinical isolates from urethral swab, seminal fluid, urine, high virginal swab, blood, skin swab and sputum of patients presenting with symptoms of *S. aureus*-associated diseases. The isolates were identified by standard method\[[@B21]\]. Methicillin resistant *S. aureus*was determined by the reference broth microdilution methods using the established NCCLS\[[@B22]\] break points. The standard strain used was *S. aureus*
NCIB 8588. The organisms were maintained on agar slope at 4°C and sub-cultured for 24 hr before use.
Bacterial susceptibility testing
--------------------------------
A standardized inoculum (1--2 × 10^7^cfu/ml 0.5 McFarland standards) was introduced onto the surface of sterile agar plates, and a sterile glass spreader was used for even distribution of the inoculum. A sterile paper disc previously soaked in a known concentration of extract (20 mcg/ml per disc) was carefully placed at the centre of the labeled seeded plate. The same procedure was used for all the MRSA strains used. The plates were incubated aerobically at 37°C and examined for zones of inhibition after 24 hr. The inhibition zones were measured with a ruler and compared with the control disc23 (disc containing only physiological saline).
Determination of minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC)
----------------------------------------------------------------------------------------------------
MIC of the extracts was determined by dilution of P, Q, R, S, T &.U to various concentrations of 0.0--32, 0.0--91, 0.0--38, 0.0--42, 0.0--44 and 0.0--35 mcg/ml respectively. Equal volume of each extract and nutrient broth were mixed in a test tube. Specifically 0.1 ml of standardized inoculum (1--2 × 10^7^cfu/ml) was added to each tube. The tubes were incubated aerobically at 37°C for 18--24 hr. Two control tubes were maintained for each test batch. These included antibiotic control (tube containing extract and the growth medium without inoculum) and organism control (the tube containing the growth medium, physiological saline and the inoculum). The lowest concentration (highest dilution) of the extract that produced no visible bacterial growth (no turbidity) when compared with the control tubes was regarded as MIC. However, the MBC was determined by subculturing the test dilution on to a fresh drug-free solid medium and incubated further for 18--24 hr. The highest dilution that yielded no single bacterial colony on a solid medium was taken as MBC.
Phytochemical screening methods
-------------------------------
All plant parts were extracted on the day of collection. The screening procedures were adapted from Wall *et al*.,\[[@B24]\] and Sofowora\[[@B25]\]. An extraction of each plant part was prepared by macerating a known weight of the fresh plant material in a blender with redistilled methylated spirit. Each extract was suction-filtered and the process repeated until all soluble compounds had been extracted, as judged by loss of colour of the filtrate. Extract from each plant part was evaporated to dryness in vacuo at about 45°C and further dried to a constant weight at the same temperature in a hot-air oven. A portion of the residue was used to test for plant constituents.
The test for tannins was carried out by subjecting 3 g of each plant extract in 6 ml of distilled water, filtered and ferric chloride reagents added to the filtrate. For cardiac glycosides, legal test and the Killer-Kiliani test\[[@B26]\] were adopted (0.5 g of extract was added to 2 ml acetic anhydrate plus H~2~SO~4~). The test for alkaloids was carried out by subjecting 0.5 g aqueous extract in 5 ml 1% HCl, boiled, filtered and Mayer\'s reagent added \[[@B26],[@B27]\]. Cyanogenic glycosides were identified by subjecting 0.5 g extract in 10 ml sterile water, and were filtered. Sodium picrate paper was added to the filtrate and heated to boil. The extract was also tested for carbohydrates using resorcinol solution\[[@B24]\]. The extract was subjected to frothing test for the identification of saponin. Haemolysis test was further performed on the froted extracts in water to remove false positive results\[[@B25]\]. Fehling\'s solution was added to the extract and heated to detect reducing sugar. The extract was also tested for free glycoside bound anthraquinones\[[@B24],[@B25]\]. Five grams of extract was added to 10 ml benzene, filtered and ammonia solution added. The presence of flavonoids was determined using 1% aluminum chloride solution in methanol concentrated HCl, magnesium turnins, and potassium hydroxide solution\[[@B28]\].
Results
=======
The profile of six medicinal plants used in this study is shown in Table [1](#T1){ref-type="table"}. The results of antibacterial activity of the crude extracts of these plants revealed that only four of the six plants: *T. avicennioides, P. discoideus, O. gratissimum*and *A. wilkesiana*showed good antibacterial activity against MRSA used.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Profile of the six medicinal plants used
:::
**Botanical Name** **Family** **Local name** **Plant part used** **Voucher Number**
----------------------------------------- --------------- ---------------- --------------------- --------------------
*Bridella ferruginea*Engl. Euphorbiaceae Ira Odun Leaf LSH
*Ageratum conyzoides*Linn. Compositae Imi-Isu Leaf LSH
*Terminalia avicennioides*Guill & Perr. Combretaceae Idi Bark LSH
*Phylantus discoideus*muel. Muel-Arg. Euphorbiaceae Asin Bark LSH
*Ocimum gratissimum*Linn. Lamiaceae Efirin Aja Leaf LSH
*Acalypha wilkesiana*Muell-Arg. Euphorbiaceae Eweeala Leaf LSH
:::
Both water and ethanol extracts of these plants were effective on MRSA strains. On the other hand, the crude extracts (water & ethanol) of *B. ferruginea*and *A. conyzoides*were weakly effective against the cocci as judged by the zones of inhibition (Table [2](#T2){ref-type="table"}). The MIC and MBC values obtained for the extracts against the MRSA varied from one plant extract to the other. For instance, the MIC values of 20.8 and 18.2 mcg/ml were obtained for water and ethanol extracts of *T. avicennioides*respectively, while the corresponding MBC values are 33.0 and 30.4 mcg/ml. The MIC and MBC values of 43.0 and 63.2 mcg/ml were recorded for ethanol extract and 71.0 and 84.2 mcg/ml for water extract of *A. conyzoides*respectively. Also, MIC & MBC values of 55.4 & 63.6 mcg/ml and 30.6 & 55.0 mcg/ml were obtained for water and ethanol extracts of *B. ferruginea*respectively. These plant extracts were bacteriostatic at lower concentrations and bactericidal at higher concentrations as revealed by MIC and MBC values (Table [3](#T3){ref-type="table"}).
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Antibacterial activity of the crude plant extracts on MRSA
:::
**Plants used** **MRSA (n = 23)** **\**S aureus***
-------------------- ------------------- ------------------ ------ ------
*B. ferruginea* 2.6 4.7 2.4 4.6
*A. conyzoides* 2.0 3.3 2.1 3.4
*T. avicennioides* 12.2 17.6 12.0 18.0
*P. discoideus* 11.5 16.0 11.2 16.1
*O. gratissimum* 10.2 10.3 9.0 9.5
*A. wilkesiana* 8.3 9.5 8.2 9.2
Methicillin \- \- \- \-
Oxacillin \- \- \- \-
**Activity key**: Figures indicate average zone of inhibition (in mm), (-) = no inhibition, \**S. aureus*(NCIB 8588) = standard strain used. Methicillin & Oxacillin = (commercial antibiotics), n = number of MRSA tested.
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Minimum inhibitory and bactericidal concentrations of plant extracts on MRSA.
:::
**Plant tested** **Water extract** **Ethanol extract**
-------------------- ------------------- --------------------- ------ ------
*B. ferruginea* 55.4 63.6 30.6 55.0
*A. conyzoides* 71.0 84.2 43.0 63.2
*T. avicennioides* 20.8 33.0 18.2 30.4
*P. discoideus* 23.0 34.2 20.5 33.0
*O. gratissimum* 25.0 37.0 22.3 35.2
*A. wilkesiana* 24.5 37.0 24. 37.0
**Key:**mcg = microgram
:::
The result of photochemical screening showed that all the six tested plants exhibited positive reactions to alkaloids, tannins and saponins. However, anthraquinone was only found in the four active plants in this study. None contained cardiac glycosides and cyanogenic glycosides. Reducing and non-reducing carbohydrates were found in *P discoideus*, *O. gratissimum*and *T. avicennioides*(Table [4](#T4){ref-type="table"}).
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Phytochemical analysis of six crude plant extracts
:::
***Screened plants.***
----------------------------------------- --------------------------- --------------------------- ---------------------------- --------------------------- ---------------------------- ---------------------------
**Components** ***B. ferruginea*(leaf)** ***A. conyzoides*(leaf)** ***T. avicenioides*bark)** ***P. discoideus*(bark)** ***O. gratissimum*(leaf)** ***A. wilkesiana*(leaf)**
Alkaloids ++ ++ ++ +++ ++ ++
Tannins ++ ++ +++ +++ +++ ++
Saponins ++ ++ +++ +++ ++ ++
Anthraquinone \+ \+ +++ \+
Cardiac glycosides \- \- \- \- \-
Cyanogenic glycosides \- \- \- \- \- \-
Flavonoides \+ \+ \+ \+ \+ \+
Reducing and non-reducing carbohydrates \- \- +++ +++ +++ \-
Parenthesis = Plant part
+++ = Appreciable amount
++ = Moderate amount
\+ = Trace amount
\- = Completely absence
:::
Discussion
==========
Medicinal plants constitute an effective source of both traditional and modern medicines, herbal medicine has been shown to have genuine utility and about 80% of rural population depends on it as primary health care. Over the years, the World Health Organization advocated that countries should interact with traditional medicine with a view to identifying and exploiting aspects that provide safe and effective remedies for ailments of both microbial and non-microbial origins\[[@B29]\].
The results of the study indicated that four out of six medicinal plants commonly used by traditional medical practitioners to cure skin and upper respiratory tract infections such as pneumonia, carbuncle, purple, impetigo and tonsillitis were active against hospital strains of MRSA. The crude extracts of *B. ferruginea*and *A. conyzoides*were weakly active against MRSA strains with ethanol extract of both plants exerting stronger antibacterial activity than water extracts. Previous studies by Ogbeche *et al.*\[[@B17]\] and Oluyemi\[[@B18]\] indicated that the crude extracts of these plants were effective against *S. aureus*. The present study was slightly conformed to their findings but the only area of concern is that while their studies only dealt with the effect of crude extracts on *S. aureus*, our study focused on the effect of crude extract on the MRSA, and determination of both MIC & MBC values of the extracts.
For example, the MIC values of 30.6 and 43.0 mcg/ml obtained for ethanol extracts of *B. ferruginea*and *A. conyzoides*in this study were lowered than their corresponding water extracts of 55.4 and 71.0 mcg/ml. Similarly, MBC values of 63.6 and 84.2 mcg/ml were recorded for ethanol extract of *B. ferruginea*and *A. conyzoides*respectively (Table [3](#T3){ref-type="table"}). These values were too high to be considered active against the pathogen. It is worthy of note that traditional medical practitioners used these plant extracts solely without combining with other plant extracts for the treatment of MRSA-associated skin and respiratory diseases. This finding may disagree with the traditional therapeutic indications claimed on these plants we therefore suggest the immediate stoppage of their traditional use against MRSA-associated diseases in Lagos, Nigeria. However, the *in-vitro*inactivity of these plants on MRSA may not necessarily translate to their *in-vivo*inactivity but the extracts may probably be playing immuno-modulatory roles in the body system. Bever\[[@B30]\] had documented immunodulation of chemical compounds from medicinal plants many of which have been proved to be inactive or weakly active *in-vitro*against pathogens. In this study, we are unable to determine immuno-modulating action of these plants due to lack of facilities.
Our investigation further showed that both water and ethanol extracts of *T. avicennioides*, *P. discoideus*, *O. gratissimum*and *A. wilkessiana*were active against *S. aureus*and MRSA. The MIC value of the four active plant extracts obtained in this study were lower than the MBC values suggesting that the plant extracts were bacteriostatic at lower concentration and bactericidal at higher concentration. The ethanol extract of the four plants exerted greater antibacterial activity than corresponding water extract (Tables [2](#T2){ref-type="table"} and [3](#T3){ref-type="table"}) at the same concentrations. These observations may be attributed to two reasons; firstly, the nature of biological active components (saponins, tannins, alkaloids and anthraquinone) which could be enhanced in the presence of ethanol. It has been documented that tannins, saponins and alkaloids are plants metabolites well known for antimicrobial activity\[[@B31]\]. Secondly, the stronger extraction capacity of ethanol could have produced greater number of active constituents responsible for antibacterial activity.
Traditionally, leaves of *O. gratissimum*and *A. wilkesiana*and barks of *T. avicenniodes*and *P. discoideus*are soaked in ethanol or water (in case of patients forbidding alcoholic intake due to religious belief) for days, large quantities of these extracts, which lack specific concentration are usually administered to patients. Our results therefore tend to support the traditional claim that these four medicinal plants are preferably extracted in ethanol. Strong vibriocidal activity of water and ethanol extracts of *T. avicennioides*and typhoidal activity of aqueous extracts of *O. gratissimum*had been reported in Lagos, Nigeria \[[@B15],[@B16]\].
The result of phytochemical screening of six plants indicated the presence of tannins, alkaloids and saponins. Interestingly, only the four plant extracts that were active against MRSA in this study contained at least trace amount of anthraquinone (Table [4](#T4){ref-type="table"}). It is therefore most probable that the presence of anthraquinone contributed to anti-MRSA activity observed. We were unable to carry out bioautography of the extracts due to lack of facilities.
Conclusion
==========
Our results therefore offer a scientific basis for the traditional use of both water and ethanol extracts of *A. wilkesiana*, *O. gratissimum*, *T. avicennioides*and *P. discoideus*separately against MR*S*A-associated skin and respiratory diseases. But *in vivo*studies on these medicinal plants are necessary and should seek to determine toxicity of the active constituents, their side effects, serum-attainable levels, pharmacokinetic properties and diffusion in different body sites. The antimicrobial activities could be enhanced if the active components are purified and adequate dosage determined for proper administration. This may go a long way in curbing administration of inappropriate concentration; a common practice among many traditional medical practitioners in Nigeria. This study represents the first preliminary report on anti-methicillin resistant *Staphylococcus aureus*activity of the crude extracts of these medicinal plants in Lagos, Nigeria.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
KOA: Conceived the study, designed, coordinated, screened the MRSA strains used and drafted the manuscript.
OO: Carried out the extraction, susceptibility testing and determined MIC & MBC values of the extract.
CEO: Participated in both extraction and susceptibility testing.
CCI: Carried out susceptibility testing and determined MIC and MBC of the extracts.
KAF: Participated in phytochemical screening and susceptibility testing.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6882/5/6/prepub>
Acknowledgements
================
We hereby acknowledge the financial support for this study by The Ford Foundation through the Centre for Development and Democratic Studies of Lagos State University (LASU/CDDS). We are also grateful to the Management of the various hospitals used as study centers and the Nigerian Institute of Medical Research, Yaba, Lagos for support and technical assistance. We are equally gratefull to Mr E O Omonigbehin for his technical assistance and Mrs O. M. Lamidi for the secretariat work.
|
PubMed Central
|
2024-06-05T03:55:55.159612
|
2005-3-11
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079793/",
"journal": "BMC Complement Altern Med. 2005 Mar 11; 5:6",
"authors": [
{
"first": "Kabir O",
"last": "Akinyemi"
},
{
"first": "Olukayode",
"last": "Oladapo"
},
{
"first": "Chidi E",
"last": "Okwara"
},
{
"first": "Christopher C",
"last": "Ibe"
},
{
"first": "Kehinde A",
"last": "Fasure"
}
]
}
|
PMC1079794
|
After publication of this article \[[@B1]\], it was learned that there was an inadvertent omission of the following statement acknowledging the National Center for Complementary and Alternative Medicine funding, which supported Eric Manheimer\'s work on this article:
Eric Manheimer was funded by Grant Number R24 AT001293 from the National Center for Complementary and Alternative Medicine (NCCAM). The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the NCCAM, or the National Institutes of Health.
We regret any inconvenience that the omission of this acknowledgment statement might have caused.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6882/5/7/prepub>
|
PubMed Central
|
2024-06-05T03:55:55.161654
|
2005-3-21
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079794/",
"journal": "BMC Complement Altern Med. 2005 Mar 21; 5:7",
"authors": [
{
"first": "Joel J",
"last": "Gagnier"
},
{
"first": "Sigrun",
"last": "Chrubasik"
},
{
"first": "Eric",
"last": "Manheimer"
}
]
}
|
PMC1079795
|
Background
==========
Questions about the infection control practices of anaesthesiologists are as old as our specialty and raised as early as 1873 by Skinner \[[@B1]\]. To control infectious complications associated with regional anaesthesia, current recommendations are based on national organizations. Although several risk factors have been discussed, risk factors associated with bacterial colonization or even infection that could guide such recommendations have not been investigated systematically so far or clinical trials had too few patients to draw meaningful conclusions. Among the risk factors that have been suspected to abet catheter infection are age, pre-existing diseases (e.g. diabetes mellitus, drug abuse, alcoholism), sepsis, and medical treatment compromising the immune response \[[@B2]-[@B4]\], site of catheter insertion \[[@B2],[@B3],[@B5]\], technically difficult catheter insertion with development of an asymptomatic haematoma that may later become the focus of bacterial colonization \[[@B6]\], filter changing manoeuvres or disconnecting the system \[[@B7]\] and duration of catheter use\[[@B5]\]. Prophylactic antibiotics, use of local anaesthetic solution with bacteriostatic effect and antimicrobial filters are thought to decrease the risk of infection \[[@B8],[@B9]\].
Thus, the purpose of this observational study was to prospectively determine the incidence of catheter bacterial colonization and infectious complications in postoperative patients having peripheral nerve or epidural catheters at different sites, and to identify factors associated with bacterial colonization of peripheral or epidural nerve catheters.
Methods
=======
This prospective study was approved by the local ethics committee and informed consent was obtained from each patient. Consecutive patients scheduled for elective surgery (orthopaedic, cardiac, visceral and urologic surgery) receiving various peripheral or epidural catheters were enrolled in this study over a period of 5 months. All catheters were placed preoperatively in the operating room or in the pre-anaesthetic holding area. No patients for chronic pain therapy were considered.
Catheter insertion
------------------
The procedure for catheter insertion was standardized and carried out with a standardized aseptic technique, according to the guidelines of the German Robert-Koch-Institution. In short these included wearing a surgical hood, face mask, sterile gloves after hand disinfection, a sterile coat, and using a large sterile drape covering the insertion site. The skin was disinfected for at least one minute by wiping or by spraying (at the anaesthetist\'s discretion) with Cutisept^®^(contains in 100 g: 2-Propanol 63 g, benzalkoniumchlorid 0,025 g, cleaned water and dyestuff). This disinfectant is suitable for all sites and recommended by the DGHM (Deutsche Gesellschaft für Hygiene und Mikrobiologie = German Society for Hygiene and Microbiology).
Bacterial filters provided with the sets were attached to all catheters in a sterile manner. The catheter insertion sites were covered with a sterile transparent dressing that permits the escape of moisture from beneath the dressing (Tegaderm^®^, consisting of polyurethan). In case of blood sequestration on the insertion site, sterile gauze was placed under the dressing. No antimicrobial prophylaxis was administered specifically for the nerve catheter insertion, but nearly all patients received a single-shot perioperative antibiotic prophylaxis after catheter placement before surgery. In orthopaedic and cardiac surgery, cefuroxim 1.5 g, and in visceral and urologic surgery a fix combination of 2 g ampicillin + 1 g sulbactam was administered intravenously.
Perioperative catheter management
---------------------------------
An initial bolus dose of a local anaesthetic was injected preoperatively. Patients with a peripheral regional catheter received a mixture of 20 ml prilocaine 1% and 20 ml ropivacaine 0.75%, and patients with an epidural catheter had 10 ml of ropivacaine 0.5--0.75% after an initial test dose of 2--3 ml bupivacaine 0.5%. Then a continuous infusion of ropivacaine 0.2% (5--15 ml/h for peripheral regional anaesthesia and 4--10 ml/h for epidural anaesthesia) was started in the postanaesthesia holding area and continued on the ward.
The catheter management postoperatively was standardized and carried out by the acute pain service by one of the authors (A.M.M.). The catheters were kept in place as long as clinically indicated, depending on a daily evaluation of the intensity of pain (aiming at a pain level of 3 cm or less on a 10 cm visual analogue scale) and the evaluation of the insertion site. For these purposes the patients were visited twice a day and the dressing was inspected and palpated. The dressings were changed only if necessary. This was defined as follows: first the site of catheter insertion was contaminated with blood, second there was a wet chamber under the dressing, or third the dressing was about to peel away. The algorithm of care used after unintentional dressing removal as well as for intended replacements was disinfection of the skin by spraying on the insertion site with Cutisept^®^, cleaning the insertion site with sterile compresses and let dry for at least one minute, then fixing a new dressing. If a catheter was obviously disconnected for a short time (less than 30 minutes), it was cleaned and disinfected about 10 cm distant from the catheter end, cut with sterile scissors and reconnected using a new sterile connector and a new bacterial filter. If the time period since disconnection was unclear, the catheter was removed. Otherwise no filter change was performed, even if the catheter was in place for a longer time.
Measurements of body temperature and a neurological examination were performed at least once a day as long as the catheter was in situ and again two days after its removal.
Catheter removal
----------------
The catheters were removed under aseptic conditions. To prevent bacterial contamination of catheter tips, the skin was disinfected with Cutisept^®^for one minute. Only when the skin had dried completely the catheter was removed to avoid direct contact of the catheter tip with the disinfectant agent. The distal catheter tip was cut with sterile scissors, placed in a sterile transport medium and transferred immediately to the microbiology laboratory.
Bacteriological methodology
---------------------------
Semi-quantitative culture techniques were used as described by Maki et al. \[[@B10]\]. The catheter segment was rolled several times across the surface of an agar plate and incubated overnight at 35°C under aerobic conditions. Then, the same catheter segment was immersed in 5 ml thioglycolate broth. After overnight incubation at 35°C aliquots of the broth were transferred to a 5% sheep blood agar plate and a MacConcey agar plate (Becton Dickinson, Heidelberg, Germany) and again incubated at 35°C for 24 h. Colony forming units (CFU) were counted and identified by standard microbiological methods. The presence of more than 15 CFU of a single organism per catheter was considered colonization, and if accompanied by signs of local inflammation (redness, swelling, and pain with pressure or tapping on the insertion site) it was defined as local infection.
Collection and processing of the data
-------------------------------------
To allow comprehensive analysis of potential factors associated with bacterial colonization, a large amount of clinical variables were recorded prospectively. These are listed in table [1](#T1){ref-type="table"}. Some of them were pre-processed to reduce the load for the multifactorial statistical analysis. E.g., the patients\' weight and height were used to calculate the body-mass-index (BMI). Furthermore, factors that were observed with a low incidence and therefore having no realistic chance to provide statistical significance in the univariate and in the multivariate analysis (history of infectious disease of the skin (n = 8) and infection with other catheter material in the past (n = 3) were analyzed separately and after merging them into an additional dummy variable. The same strategy was used for factors known to provoke surgical wound infection \[[@B11]\]. These were diabetes mellitus (n = 24), chronic steroid medication (n = 9), and cancerous disease (n = 49) \[[@B12]\]. The anatomical site of catheter insertion was grouped using the incidences of catheter colonization in a descriptive univariate analysis. Several attempts to group the different catheter techniques were used but finally the best discriminating power was achieved by summarizing catheters located in the groin (femoral nerve catheters and sciatic nerve catheters inserted by the anterior approach described by Meier et al\[[@B13]\]) against all other techniques.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Results of the stepwise logistic regression analysis, where all potential risk factors are included during the first step and subsequently removed if not significant (p \> 0.05). \*Please note that p-values at removal of the parameter must not necessarily have the same order than the step at which the factor was removed, since the logistic model recalculates at each step.
:::
Clinical parameter Removed at step Removed with an odds-ratio of Removed with a p-value of
------------------------------------------------------------------------------------- -------------------------------------------------------------------------- ------------------------------- --------------------------- -------
Sex (female vs. male gender) 2 0.89 0.88
Age (per decade) 12 0.84 0.34
Body mass index (per unit kg·m^-2^) 15 0.95 0.57
ASA-status III and IV versus I and II 18 1.48 0.22
Diabetes mellitus 7 3.72 0.99
Cancerous disease 19 0.48 0.18
Chronic systemic corticosteroid medication 20 1.89 0.13
\"immune suppression\" (any of the three afore mentioned diseases) 5 1.35 0.99
Skin abscess in the past 3 0.74 0.87
Infection with other catheter material in the past 16 0.17 0.26
Easy perspiration^1^ 11 2.33 0.32
Any of the three afore mentioned factors 14 1.64 0.43
Localization of the catheter (\"clean\" vs. \"dirty\")^2^ 8 0.38 0.37
Puncture site in the groin Not removed 3.39 0.004
Epidural catheter vs. peripheral nerve catheter 17 1.24 0.11
Disinfection technique (wiping versus spraying) 9 1.88 0.50
Number of attempts during catheter placement 10 1.49 0.38
Fixation technique (tunneling versus suturing) 1 0.97 0.97
Intraoperative corticosteroid medication 4 1.40 0.75
Intraoperative single shot antibiotic administration 6 2.34 0.47
Postoperative antibiotic therapy (at least 3 days or until removal of the catheter) Not removed 0.41 0.05
Perioperative antibiotic therapy (intra- & postoperative) 23 0.43 0.09
Duration of catheter use (per day) 21 1.17 0.16
Accidental disconnection of the catheter 22 0.78 0.19
Accidental removal of the catheter dressing (n =) 15 0.50 0.37
Intentioned replacements of the catheter dressing (n =) 13 0.83 0.31
Removal of the catheter dressing (intentionally or unintentionally; n =) Not removed 2.12 0.001
1\) Easy perspiration (e.g. getting bathed in sweat every night, sweating without moving very much) with the danger of easy removal of the dressing
2\) Localization of the catheter (\"dirty\" = groin, axilla, interscalene, epidural below Th3; \"clean\" = epidural above Th3, paravertebral, psoas compartment, posterior and distal sciatic nerve, infraclavicular) based on the distribution of sebaceous glands on the skin \[21\]
:::
Statistical analysis
--------------------
Twenty six potentially relevant variables were entered into a stepwise backward logistic regression analysis using the maximum likelihood method. They are listed in table [1](#T1){ref-type="table"} with the order of removal from the model and the odds-ration and p-value, respectively. The goodness of fit of the regression model was judged using Nagelkerkes\'s R^2^. All analyses were performed using JMP 5.1 for Windows (SAS Institute Inc., Cary, NC, USA) and SPSS 11.5 for Windows (SPSS Inc., Chicago, Ill, USA).
Results
=======
Demographic data
----------------
A total of 200 catheters from 191 patients were initially enrolled in the study. Two catheters were excluded because they were not removed in accordance with the aseptic technique. Thus, 198 catheters from 189 patients could be analyzed. Five patients for total knee replacement had double catheterization (sciatic and femoral nerve catheter) and four patients had a repetitive intervention within a couple of months with the same technique.
Catheters were removed between day 0 (if a planned extensive surgery was modified intraoperatively into a smaller one not requiring postoperative analgesia via a catheter), and day 31 after an invasive procedure. In mean, catheters were in use for 3.7 days (standard deviation: 3.0). The median and the 25^th^/ 75^th^percentile were: 3; 2 / 5. This time period was not different in colonized catheters (mean: 3.8 ± 2.1 days) and uncolonized catheters (mean: 3.7 ± 3.1 days).
Bacteriological results
-----------------------
Of 198 catheters analysed, 47 (23.7%; 95%-confidence interval: 18--30%) were not sterile. A heterogeneous flora of bacteria could be detected. In most cases (78.7%) these were normal non pathogenic skin flora. Coagulase negative staphylococci were most often detected, and only 21.3% were optional pathogenic microorganisms (table [2](#T2){ref-type="table"}). In 33 patients (16.7%; 95%-confidence interval: 12--23%) there were more than 15 CFU detectable. In table [3](#T3){ref-type="table"} the latter are listed according to the different insertion sites. Of these patients, 18 showed additional signs of local inflammation, indicating local infection.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Bacterial contamination after removal of the catheter (total number of catheters: n = 198). CFU = colony forming units; the presence of less than 15 CFU of a single organism per catheter is considered to indicate catheter contamination, higher counts are defined as catheter colonization.
:::
No bacterial catheter contamination n = 151 (76.3%)
------------------------------------------------ ---------------------------------------- ----------------------- -----------------------
[Bacterial catheter contamination]{.underline} n = 47 (23.7%)
One organism per catheter n = 31 (66% of contaminated catheters)
Two or more organisms per catheter n = 16 (34% of contaminated catheters)
[Total organisms]{.underline} n = 66 (= 100 %) [≤15 CFU]{.underline} [≥15 CFU]{.underline}
[Normal skin flora]{.underline}
Coagulase negative staphylococci n = 40 (60.6%) 10 30
Bacillus species n = 9 (13.6%) 5 4
Enterococcus species n = 3 (4.5%) 2 1
[Optional pathogenic]{.underline}
Escherichia coli n = 5 (7.8%) 5 0
Enterobacter species n = 3 (4.5%) 2 1
Klebsiella species n = 3 (4.5%) 3 0
Morganella morganii n = 1 (1.5%) 1 0
Nonfermenter species n = 1 (1.5%) 1 0
Pseudomonas aeruginosa n = 1 (1.5%) 0 1
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Catheter colonization rate at the different sites of insertion. The presence of more than 15 CFU (colony forming units) of a single organism per catheter is defined as catheter colonization, and more than 15 CFU accompanied by local signs of inflammation indicate local infection. Values are expressed as median (25. / 75. percentile) or absolute and relative incidences; n = (%).
:::
Localization Total number of catheters Duration of catheter use (days) Not colonized n (%) Colonized n (%) Colonized *and*local infectious signs; n (%)
---------------------------------------- --------------------------- --------------------------------- --------------------- ----------------- ----------------------------------------------
Epidural catheter (C6/7 - T2/3) 29 6 (5 / 8) 29 (100%) 0 (0%) 0 (0%)
Epidural catheter (T3/4 - T 12/L1) 59 4 (3 / 6) 51 (86.4%) 8 (13.6%) 4 (6.8%)
Epidural catheter (L1/2 - L4/5) 8 4 (3 / 6) 7 (87.5%) 1 (12.5%) 0 (0%)
Paravertebral catheter (level T3 - T8) 4 1 (1 / 3) 4 (100%) 0 (0%) 0 (0%)
Psoas compartment 5 4 (3 / 4) 4 (80%) 1 (20%) 1 (20%)
Posterior sciatic (gluteal) 1 2 1 (100%) 0 (0%) 0 (0%)
Distal sciatic (popliteal) 1 2 1 (100%) 0 (0%) 0 (0%)
Anterior sciatic (proximal) 7 2 (1 / 3) 4 (57.1%) 3 (42.9%) 1 (14.3%)
Femoral nerve 68 3 (1 / 3) 50 (73.5%) 18 (26.5%) 10 (14.7%)
Interscalene plexus 9 2 (1 / 3) 8 (88.9%) 1 (11.1%) 1 (11.1%)
Infraclavicular plexus 5 3 (2 / 6) 4 (80%) 1 (20%) 1 (20%)
Axillary plexus 2 3 (1 / 4) 2 (100%) 0 (0%) 0 (0%)
Total 198 3 (2 / 5) 151 (83.3%) 33 (16.7%) 18 (9.1%)
:::
Results of the logistic regression analysis
-------------------------------------------
The stepwise logistic regression analysis revealed that out of the 26 potentially relevant parameters only three independent factors remained in the final model as statistically significant (table [4](#T4){ref-type="table"}). Catheter placement in the groin was associated with a significant higher incidence of catheter colonization (p = 0.004). The odds-ratio was 3.4 (95%-confidence interval: 1.5 -- 7.8) compared to all other anatomical sites. No other potential risk factor that can be determined preoperatively came out as statistically significant. Postoperatively, removal of the catheter dressing, either intentionally or unintentionally, was associated with an increased risk for colonisation. Using the graphical exploratory tools in the JMP 5.1 software, there was an almost linear increase of the rate of colonization with an increasing number of changes of the dressings. Results of the logistic regression analysis revealed that each attempt to change the dressing increased the risk with an odds ratio of 2.1 (95%-CI: 1.4 -- 3.3; p = 0.001). There was a maximum number of changing the dressing of five times.
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Results of the final step of the backward logistic regression analysis. Three factors remained in the final model and are presented with their odds ratio and 95%-confidence interval. The coefficient of each factor can be used with the constante of the model to calculate a predicted risk for each patient
:::
Coefficient Standard error P Odds ratio (95%-confidence interval)
----------------------------------------------------------------------- ------------- ---------------- ----------- --------------------------------------
Catheter placement in the groin 1.222 0.424 0.004 3.39 (1.48 -- 7.79)
Changing the dressing (per attempt) 0.753 0.222 0.001 2.12 (1.37 -- 3.29)
Postoperative administration of an antibiotic (at least for 24 hours) \- 0.896 0.516 0.05 0.41 (0.12 -- 1.02)
Intercept \- 2.634 0.414 \< 0.0001
:::
Postoperative administration of an antibiotic drug at least for 24 hours significantly reduced the risk of catheter colonization. The odds-ratio was 0.41 (95%-CI: 0.12 -- 1.0; p = 0.05). The constant of the equation of the regression analysis (- 2.63) and the coefficients for each risk factor can be used to calculate a predicted risk for each patient. This theoretical risk can vary between 2.8% (when a catheter is not placed in the groin, no changing of the dressing is performed, and the patients receives postoperative antibiotic treatment) and 91% (in a patients receiving a femoral nerve catheter, with no postoperative antibiotic treatment, and where the dressing was removed five times or more). These calculations are performed for demonstration in the appendix. The goodness of fit was moderate but acceptable (Nagelkerke\'s R^2^= 0.20).
Clinically infected catheters
-----------------------------
Despite the high rates of catheter colonization and superficial local infection, only two clinical infections occurred.
On the fourth postoperative day (the dressing was changed once on the second postoperative day) a patient with an interscalene plexus catheter developed pain at the insertion site, neuropathic pain of the arm and a reddish swelling of 4 cm in diameter, temperature of 38.6°Celsius, and a leukocyte count of 16.7 G·l^-1^within a few hours. Until then, the patient did not receive any prophylactic antibiotic except for the intraoperative single-shot administration of 1.5 g intravenous cefuroxim. The catheter was immediately removed and antibiotic therapy with cefuroxim 1.5 g intravenously three times daily was initiated. All symptoms disappeared within the following two days. Two different species of coagulase negative staphylococci (staphylococcus epidermidis) were found on the catheter tip, both of them with CFU \> 15. One kind of staphylococcus epidermidis was resistant to cefuroxim, but since the symptoms resolved quickly, the antibiotic regimen was not changed.
The other patient presenting with an infectious complication had an epidural catheter at T7/8. The dressing was changed three times. Only the perioperative single-shot antibiotic with a fix combination of 2 g ampicillin + 1 g sulbactam had been administered, and no further antibiotic treatment was necessary. On the fifth postoperative night he developed very intensive pain and a dark red swelling of 8 cm in diameter superficially just underneath the skin. Until then, a continuous infusion of 4--6 ml/h ropivacaine 0.2% was infused. Neurological examination was normal, neither were there signs of systemic reaction like fever or leukocytosis. The catheter was removed and a local disinfectant ointment was applied. Within 36 hours all symptoms had resolved. The bacterium found on the catheter tip was again staphylococcus epidermidis with \> 15 CFU.
Discussion
==========
In this study, catheter colonization occurred with an incidence of almost 17%. More than half of these colonized catheters also presented with local signs of inflammation (9%). In contrast to these high colonization rates real catheter related infections (local complications, bacteriaemia and / or systemic reactions like fever and leukocytosis) are quite rare. Cuvillon found only three out of 208 femoral catheters with transitory bacteriemia likely related to the catheter, and no abscess occurred, despite the high colonization rate of 57% \[[@B14]\]. Steffen et al. reported a low incidence of colonization in a series of 502 epidural catheters. Several large studies reported epidural abscesses with a varying incidence between 0% and 3% \[[@B5],[@B15]-[@B17]\].
In our trial only two catheter related local infections occurred. Both resolved completely within two days with only local ointment or intravenous antibiotics. No serious complication occurred at all during our observational period.
Using a multifactorial statistical model, three independent factors could be identified that were associated with bacterial colonization. However, only one factor (anatomical localization of the insertion site) can be used as a \"true risk factor\" since the other risk factors are \"postoperative\" variables. E.g., the decision to perform antibiotic therapy is often performed by the surgeon and the number of changing of the dressing is not easy to foresee.
All other potential \"true risk factors\" that are patient related factors (e.g. gender, age, pre-existing diseases), puncture site and technical details of catheter placement and fixation (e.g. number of attempts until successful placement, catheter tunnelling) were removed as insignificant. This means that it is not possible to discriminate which patient will or will not develop catheter colonization preoperatively. This result highlights the need for a close postoperative evaluation of every patient even if no factor is present that has been described as a risk factor in previous reports.
Age, preexisting diseases or medical treatment which compromise the immune response have been discussed as potential risk factors and in part are proven risk factors for surgical wound infection \[[@B18]\]. In our trial, neither age nor preexisting diabetes mellitus, cancer disease, infectious disease, abscess in the past, infection with other catheter material in the past, prolonged corticosteroid therapy or short term corticosteroid therapy perioperatively were indicators for an increased risk. Furthermore, combining disease states that occurred too infrequent to have a realistic change to achieve statistical significance did also not lead to variables with significant impact.
The site of catheter insertion is another potential influencing factor in previous studies. The femoral site was associated with a rate of bacterial colonization as high as 57% \[[@B14]\], whereas the popliteal insertion site had a very small bacterial colonization rate of 7.5% \[[@B19]\]. Epidural catheters revealed catheter colonization in 6 to 35% \[[@B2],[@B20]\]. One possible explanation for these differing results might be the great variations with respect to the density of sebaceous glands in the different insertion sites that has been shown to impact the ability of local disinfectants to reduce the number of microorganisms \[[@B21]\]. For example Steffen et al. \[[@B2]\] reported a higher incidence of colonized catheters in patients where the epidural catheters were placed at a thoracic level compared to the lumbar route. However, a variable that should distinguish between potentially more contaminated and clean puncture sites based on the latter hypothesis was early removed as insignificant in our analysis. Catheter placement in the groin (femoral nerve catheters and sciatic catheters advanced via the anterior approach) was associated with a significantly higher incidence of colonization than all other anatomical landmarks.
Technically difficult catheter insertion may cause asymptomatic haematoma that may later become the focus of bacterial colonization \[[@B6]\]. However, this theory was not supported by other authors \[[@B5]\]. In our trial the numbers of skin perforations with the needle during catheter placement did not increase the occurrence of catheter colonization.
The repetitive administration of antibiotics during the postoperative period reduced the incidence of catheter colonization. Reports from the literature support the view that antibiotic therapy during the perioperative period lowers the risk for infectious catheter complications. A relatively high rate of epidural abscess occurred in a population that apparently did not receive perioperative antibiotics routinely \[[@B5]\]. Furthermore, in a series of 405 axillary catheters, the only abscess occurred in a patient who had not received an antibiotic \[[@B22]\]. It is interesting to notice that intraoperative single dose antibiotic treatment did not provide sufficient protection. However, this single shot treatment was usually administered 30--60 minutes after the insertion of the epidural or peripheral nerve catheter. Thus, we can not answer the question, whether antibiotic prophylaxis *before*catheter placement might be able to reduce the incidence of colonization.
Concerning the possible routes for catheter colonization, Hunt et al. demonstrated that the catheter hub represented the main route for catheter colonization \[[@B11]\]. Therefore disconnection of the closed system or filter changing maneuvers should be avoided if possible \[[@B11],[@B23]\]. We analyzed the situations where catheters were accidentally disconnected assuming that the unprotected end was open for an indefinite time and could let microorganisms pass through. In the multifactorial analysis, accidental disconnection of the catheter was removed at a late stage of the stepwise logistic regression procedure. Thus, this potential risk factor was insignificant but is a candidate for further investigations.
Local anaesthetic solutions with bacteriostatic effect like bupivacaine, prilocaine, lidocaine and tetracaine \[[@B24]\] have shown to decrease the risk of infection \[[@B8],[@B9]\]. In our trial, only ropivacaine 0.2% was used postoperatively for continuous administration and thus this potential influencing factor could not be included in the statistical model.
Duration of catheter use has been found to increase the risk of infectious complications in a Danish study with epidural catheters \[[@B5]\]. No epidural abscess was found with use of catheters ≤ 2 days, but one third of the abscesses were found in patients who had the catheter in situ for three days only. This implicates that even a short catheterization time of three days does not eliminate the risk of infection. In another observational trial, there was a very strong correlation between duration of catheter use and infectious complications in patients with perfusion disorders, but not in the other subgroups \[[@B25],[@B26]\]. In our own trial we could not observe a statistically significant time dependency, but the variable was late removed at step 21 of 23. In our trial catheters were removed between day 0 and day 31. The decision to withdraw a catheter was primarily based on the daily pain evaluation by the patient. However, also local signs of the insertion site influenced the decision to remove the catheter. Thus, it is important to notice that duration of catheter use is not a risk factor only under the strict assumption that the site of insertion is evaluated at least once a day and the catheter is immediately removed if there are any signs of local redness, swelling or pain at the insertion site. This is in agreement with a recent study showing that the duration of use of an epidural catheter was not different in colonization and in sterile catheters \[[@B2]\]. Attention should be paid to the fact that duration of catheter placement has some correlation with the number of removal of catheter dressing (Pearson correlation coefficient r = 0.50; Spearman correlation coefficient rho = 0.35). Only the latter factor remained statistical significant in the final model, and thus some of the predictive information provided by the duration of catheter placement was virtually transferred. This phenomenon of co-linearity is discussed in more detail in the following paragraph.
Limitations of the study
------------------------
Several of the potential risk factors that were evaluated using a stepwise logistic regression analysis are correlated with each other. For example there is a strong correlation between all factors describing the pre-anaesthetic health condition of a patient summarized as the ASA-status on one hand, and cancer disease or age on the other hand. Thus, if one or several of such factors are removed during the stepwise exclusion procedure a certain part of the information of the removed factor is transferred to the correlated factors still in the model. This can lead to artifacts by increasing the reputed impact of a risk factor. Thus, it is important to notice that the risk factors that remained in the final model are not necessarily causal for the rate of catheter colonization but are maybe only associated with an increased incidence. Example given: it could be misleading to reason that the underlying reason why the number of changing the dressing increases the risk for catheter colonization is simply the manipulation of the catheter (with subsequent contamination etc.). Rather it is also possible that other correlated factors (contamination with blood, local secretion at the insertion site of the catheter, intensive sweating of the patient, and the site of catheter placement) are the \"true\" (causal) factors for colonization and were erroneously removed during the logistic regression procedure transferring their predictive power to other correlated factors.
Another limitation of the study is the lack of a sufficient number of all catheter techniques at all anatomical sites. E.g., there were only two axillary plexus catheters and from a theoretical point of view this anatomical site has similar problems as the groin with respect to difficulties in fixation of the catheter, intense sweating, a high rate of sebaceous glands etc. Thus, the fact that the axilla and other anatomical sites underrepresented in this study were not identified as a relevant risk factor, does not necessarily mean that these locations are not associated with a higher incidence of catheter colonization.
Conclusion
==========
Summarizing the present results, three independent risk factors could be detected applying a stepwise logistic regression procedure to a great number of potential risk factors for bacterial catheter colonization. Catheter localisation in the groin, removal of the dressing and omission of postoperative antibiotics were associated but not necessarily causal for postoperative catheter colonization.
Appendix
========
Calculation of the theoretical risk for bacterial colonization of a peripheral or epidural nerve catheter, defined as more than 15 colony forming units.
Patients 1 (low risk) received an epidural catheter for five days after hemicolectomy. During this time, removal of the dressing was not necessary. He received an antibiotic prophylaxis during the first three postoperative days.
The risk for colonization of the epidural catheter can be calculated as follows:
\- z = - (-2.63~constante~+ (-0.90 × 1~antibiotic\ treatment~+ 0.75 × 0~number\ of\ dressing\ replacement~+ 1.22 × 0~localisation:\ not\ groin~) = - 3.53 Risk \[%\] = 100% / (1+e^-z^) = 100% / (1+e^-3.53^) = 2.8%
Patient 2 (high risk) received a sciatic nerve catheter for five days using the anterior approach (in the groin). Postoperatively, the dressing had to be changed every day. The patient received only a single dose of an antibiotic intraoperatively.
\- z = - (-2.63~constante~+ (-0.90 × 0~antibiotic\ treatment~+ 0.75 × 5~number\ of\ dressing\ replacement~+ 1.22 × 1~localisation:\ groin~) = - (-2.34) Risk \[%\] = 100% / (1+e^-z^) = 100% / (1+e 2.34) = 91.2%
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
AMM designed the study, collected the clinicaldata and wrote the manuscript.
KMK collected the clinical data.
MK performed the culture techniques.
RN collected the clinical data.
VER collected the clinical data.
HW participated in the conception of the study.
SZ performed the culture techniques.
LHJE designed the study, performed the statistical analysis and extensively revised the manuscript.
All authors read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2253/5/1/prepub>
|
PubMed Central
|
2024-06-05T03:55:55.162200
|
2005-3-17
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079795/",
"journal": "BMC Anesthesiol. 2005 Mar 17; 5:1",
"authors": [
{
"first": "Astrid M",
"last": "Morin"
},
{
"first": "Klaus M",
"last": "Kerwat"
},
{
"first": "Martina",
"last": "Klotz"
},
{
"first": "Roswitha",
"last": "Niestolik"
},
{
"first": "Veronika E",
"last": "Ruf"
},
{
"first": "Hinnerk",
"last": "Wulf"
},
{
"first": "Stefan",
"last": "Zimmermann"
},
{
"first": "Leopold HJ",
"last": "Eberhart"
}
]
}
|
PMC1079796
|
Background
==========
Phosphorylation emerged as a major post-translational mechanism that is well suited for effecting a rapid change in the activity of a transcription factor in response to an extracellular signal \[[@B1],[@B2]\]. During periods of physical or chemical stress, transcription of genes encoding cytoprotective heat shock or stress proteins (HSPs) is increased. This enhanced expression is primarily mediated by heat shock factor 1 (HSF1) in vertebrate cells or by a homologous factor (HSF) in non-vertebrate cells. HSF/HSF1 is continuously present in cells but is only activated when the cells experience a stress. It was long known that HSF/HSF1 is hyperphosphorylated in stressed cells \[[@B3]-[@B5]\].
Activation of human HSF1 occurs in at least two steps. A first step results in formation of factor homotrimers that are capable of binding so-called heat shock element (HSE) sequences present in *hsp*genes but essentially lack transcriptional activity. In a second step, these HSF1 homotrimers are converted to a transcriptionally competent form \[[@B6]-[@B8]\]. In cells exposed to heat, acquisition of HSE DNA-binding activity was observed to precede hyperphosphorylation of HSF1 \[[@B9]\]. This result suggested that hyperphosphorylation could play a regulatory role in the second activation step that renders the factor transactivation-competent. Several additional observations are compatible with the hypothesis that hyperphosphorylation of HSF1 is required for or enhances induction of the transcriptional competence of the factor: (i) To the extent this was examined, all conditions that resulted in activation of HSF1 also induced hyperphosphorylation of the factor. (ii) Conversely, compounds such as salicylate, indomethacin, menadione and hydrogen peroxide that were only capable of triggering the first step of HSF1 activation also failed to prompt factor hyperphosphorylation \[[@B8],[@B10],[@B11]\]. (iii) Inhibitors of Ser/Thr protein kinases reduced, and inhibitors of Ser/Thr phosphatases enhanced, HSF1 activity \[[@B11]-[@B17]\]. For the inhibitors investigated it was found that they did not affect HSF1 DNA-binding activity \[[@B11]\] (see also \[[@B18]\]).
To date, stress-induced phosphorylation of HSF/HSF1 has not been comprehensively analyzed. However, phosphorylation of Ser^230^of human HSF1 was reported to contribute to heat activation of the factor by enhancing its transcriptional competence \[[@B19]\]. It was also proposed that phosphorylation of Thr^142^of human HSF1 may be essential for factor activity \[[@B20]\]. Furthermore, several HSF/HSF1 residues whose phosphorylation repressed factor activity were identified \[[@B9],[@B21]-[@B30]\]. In human HSF1 these residues are Ser^303^, Ser^307^and Ser^363^. The present study sought to combine systematic mutagenesis and physical analyses to provide a broad accounting of phosphorylation of HSF1 in heat-stressed cells.
Results
=======
Validation of a transactivation assay for testing HSF1 mutants
--------------------------------------------------------------
In chimeric transcription factor LEXA-(human)HSF1 the DNA-binding domain (residues 1--78) of the 529-residue human HSF1 polypeptide is substituted with that of bacterial repressor LEXA (residues 1--87) \[[@B7]\]. LEXA-HSF1 is known from previous studies to be regulated similarly as HSF1 \[[@B7]\]. Transactivation by LEXA-HSF1 was assessed by dual luciferase assay of cells co-transfected with a firefly luciferase gene responsive to LEXA-HSF1 (LEXA-fLUC) and a constitutively expressed *Renilla*luciferase gene (pRL-TK or pRL-CMV). Reporter activity was expressed as the ratio of firefly and *Renilla*luciferase activities. To permit the identification of mutants of LEXA-HSF1 with only minor functional impairments, the transactivation assay needed to sensitively detect changes in transcription factor activity. To find out whether the assay had this capability under the chosen experimental conditions, 96-well cultures were transfected with different amounts of expression construct LEXA-HSF1 (Figure [1A](#F1){ref-type="fig"}). At amounts below about 2 ng/culture, reporter activity increased proportionally with the amount of expression construct transfected. Hence, at these low DNA concentrations changes in LEXA-HSF1 activity were certain to be reflected in proportional changes in reporter activity. Most subsequent transfections were carried out with 0.5 ng or less of expression construct per 96-well culture. Transfection with 0.5 ng of construct LEXA-HSF1 resulted in a 30% increase in total HSF1 concentration as estimated by western blot one day after transfection (Figure [1B](#F1){ref-type="fig"}). Assuming a typical transfection efficiency of 20--40%, this result implied that transfected cells on average expressed comparable amounts of LEXA-HSF1 and endogenous HSF1. Hence, under these conditions, under which an exogenous HSF1 form was not or was only minimally overexpressed, the regulatory environment encountered by the exogenous HSF1 form was likely to be similar to that to which endogenous factor is exposed. This is a departure from several earlier studies relating to HSF1 phosphorylation, including one from our own group \[[@B28]\], in which studies exogenous HSF1 forms were substantially overexpressed. A known result of such overexpression is that a significant fraction of exogenous factor accumulates as DNA-binding trimers in the absence of a stress \[[@B7]\]. At the low levels of expression construct transfected in the assays used herein, this result did not occur (as documented, e.g., by the experiment shown in Figure [3A](#F3){ref-type="fig"}).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Validation of the transactivation assay used, transactivation analysis of mutant S326A, and isolation of FLAG-HSF1 for phosphorylation analyses**. (**A**) Transactivation assay of cells in 96-well dishes co-transfected with reporter gene mixture (LEXA-fLUC and pRL-CMV) and 0--40 ng LEXA-HSF1. C: Control unheated cells; HS: Cells subjected to standard heat-treatment for 30 min at 44°C. (**B**) Relative amounts of HSF1-immunoreactive protein in cells transfected with 0.5 ng β-galactosidase expression construct (B-GAL) or 0.5 ng LEXA-HSF1. Extracts prepared one day after transfection were analyzed by anti-HSF1 western blot. A quantitation of the HSF1 signals, shown in the inserts on top, is presented. Note that the insert only shows the relevant (scanned) portion of the blot, i.e., polypeptide sizes from about 60--100 kD. HSF1 forms appear as several closely spaced bands above a single band representing a nonspecific signal. (**C**) Coomassie Blue-stained gel containing immune-isolated FLAG-HSF1. M: markers having molecular weights in kD as indicated on the left. Note that the cluster of bands appearing at and below the 98 kD marker was identified as HSF1 forms by a parallel anti-HSF1 western blot (not shown). (**D**) Transactivation assay comparing activities of HSF1 forms in cells co-transfected with luciferase reporter gene mixture and expression constructs (0.5 ng) for FLAG-LEXA-HSF1 substitution mutant S326A, parent FLAG-LEXA-HSF1 or B-GAL. One day after transfection, cells were exposed either to 300 μM CdCl~2~for 2 h (Cd) or to standard heat treatment (HS), or were left untreated (C). Relative firefly luciferase activities assayed 6 h after treatments are shown. (**E**) FLAG western blot of parallel (heat-treated) cultures from the experiment analyzed under C that compares expression levels of FLAG-LEXA-HSF1 mutant S326A and parent FLAG-LEXA-HSF1 one day after transfection. The tubulin signal was used as loading control. The numbers below the blots represent a quantitative comparison of the mutant S326A and parent FLAG-LEXA-HSF1 signals in the FLAG blot.
:::
:::
Alanine scan of LEXA-HSF1
-------------------------
Substitution mutants of LEXA-HSF1 or FLAG-LEXA-HSF1 (FLAG-tagged LEXA-HSF1) were prepared that collectively covered all 92 Ser, Thr and Tyr residues of the HSF1 sequence present in LEXA-HSF1. Transactivation assays were carried out to compare transcription-enhancing abilities of mutant and parent factors in cells that either had been exposed to a 44°C/30 min heat treatment or had not been heat-treated. It is noted that this heat treatment resulted in only minimal cell death. Results revealed that most mutants retained the strong heat inducibility of the respective parent factor (Table [1](#T1){ref-type="table"}). Nine mutants (highlighted) had an induced activity that was lower than that of parent factor by more than two average standard deviations (2 × 10.3%). Five of these mutants were only modestly impaired, but four mutants, S326A/T328A, T400A/S(403/404)A, S(457/458/461)A and T511A/S513A, had substantially reduced activities. Mutants S326A/T328A and T511A/S513A were expressed to normal levels. However, mutant T400A/S(403/404)A accumulated to a noticeably lower level, and mutant S(457/458/461)A was cleaved proteolytically (western blot data not shown). These results suggested the possibility that phosphorylation of residues 326, 328, 511 and/or 513 played an important role in activation of HSF1 by heat stress. Mutants S(303/307)A and S307A were about 150% and 50%, respectively, more active than parent factor in heat-treated cells (Table [1](#T1){ref-type="table"}).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Alanine scan of LEXA-HSF1
:::
Construct Relative luciferase
---------------------------- --------------------- --------- -------- ----------
LEXA-HSF1 0.50 0.1 100 13.5
LEXA-HSF1(F) 0.32 0.1 100 6.3
T97A 0.83 0.3 122 5.0
T120A/S121A(F) 0.59 0.1 115 4.7
S123A/T124A 0.65 0.6 95 6.6
S127A 0.50 0.0 84 5.1
S136A/T138A(F) 0.23 0.1 112 11.6
T142A(F) 0.22 0.0 83 11.0
S156A(F) 0.29 0.0 102 6.1
S(174/195/199)A 0.30 0.0 105 8.2
**S216A** **0.52** **0.1** **77** **10.5**
S218A(F) 0.31 0.0 117 9.5
**S221A(F)** **0.64** **0.4** **70** **8.3**
Y225A/S(226/230)A(F) 0.39 0.0 92 5.8
S(237/241/244)A/Y240A 0.54 0.1 126 6.4
Y247A/S(248/249)A(F) 0.72 0.1 127 2.2
S(250/251)A/Y253(F) 1.21 0.3 99 16.6
S(260/261)A 0.84 0.2 119 19.1
S266A/T269A(F) 0.95 0.3 83 11.8
**S275A** **1.05** **0.3** **75** **5.5**
S279A 0.51 0.0 103 24.7
S290A 0.32 0.0 101 13.5
S291A 1.43 0.4 80 4.1
S292A 1.47 0.6 84 3.1
S303A 0.97 0.3 81 13.9
S307A(F) 0.36 0.0 148 33.7
S(303/307)A(F) 1.27 0.5 249 16.9
S314A(F) 0.34 0.0 131 18.3
S(319/320)A/T323A 0.83 0.2 78 10.8
**S326A/T328A** **2.15** **1.0** **46** **1.9**
S333A(F) 0.37 0.0 103 9.3
S338A(F) 0.51 0.1 88 10.1
S344A(F) 0.47 0.1 101 13.4
T346A(F) 0.60 0.1 108 15.0
T349A(F) 0.35 0.1 95 12.6
T(355/357)A(F) 0.34 0.0 129 12.6
S(363/368)A/T(367/369)A(F) 1.05 0.6 92 14.4
S375A 2.13 1.9 73 13.6
S385A(F) 0.57 0.1 121 5.7
S393A 1.38 0.6 83 14.9
**T400A/S(403/404)A(F)** **0.57** **0.2** **63** **2.9**
T411A/S412A(F) 0.57 0.0 85 6.3
S(419/421)A/T423A(F) 0.48 0.0 101 6.7
**S428A** **0.88** **0.1** **75** **9.6**
S(434/435/438)A 1.39 0.1 81 12.9
S444A 1.27 0.2 88 13.2
**S(457/458/461)A(F)** **0.49** **0.1** **7** **0.3**
Y468A/T469A 2.12 1.9 103 16.6
S(480/485)A/T483A 0.47 0.1 80 4.0
S(498/501)A/Y499A 1.08 0.1 80 9.2
**T511A/S513A(F)** **0.44** **0.1** **22** **1.0**
T516A/S518A(F) 0.87 0.1 78 8.1
T527A/S529A 1.14 0.2 91 4.6
S121A 1.73 0.2 106 12.1
S230A 1.64 0.3 101 22.7
S319A 0.60 0.1 105 14.3
S320A 0.36 0.1 91 12.8
T323A 1.97 0.5 102 17.6
S326A 1.66 0.3 40 8.1
S419A 1.67 0.2 104 17.8
C: not-heated; SD: standard deviation; HS: heat-treated. Transactivation assays were carried out as discussed in Methods and Results. (F) refers to the presence of an amino-terminal FLAG tag (see Methods). The activities of FLAG-tagged mutants are expressed relative to the activity of FLAG-LEXA-HSF1, and those of not-FLAG-tagged mutants to the activity of LEXA-HSF1.
:::
Residues phosphorylated in HSF1 isolated from heat-treated cells
----------------------------------------------------------------
In a first series of experiments, cultures transfected on the previous day with expression construct FLAG-HSF1 were pre-equilibrated with ^32^PO~4~and then heat-treated for 45 min at 44°C. Exogenous HSF1 was immunoprecipitated using an anti-FLAG resin. In a stained, high-resolution Tris-Tricine SDS-PAGE gel (Figure [1C](#F1){ref-type="fig"}), immune-isolated HSF1 appeared as two sharp bands and a slower migrating diffuse region that contained the most highly phosphorylated forms (that were also most intensely radiolabeled; data not shown). Protein from the latter region (see bracket in Figure [1C](#F1){ref-type="fig"}) was subjected to trypsin or trypsin/chymotrypsin digestion. Phosphopeptides were separated by HPLC, tentatively identified by mass spectrometry (MALDI-MS) and confirmed by regular and/or radiochemical sequencing. In subsequent experiments, peptides from unlabeled, purified FLAG-tagged HSF1 digested with trypsin, trypsin/chymotrypsin, or endoproteinases Glu-C or AspN were analyzed by tandem mass spectrometry (LC/MS/MS). These analyses resulted in an unambiguous identification of phosphoserines at positions 121, 230\*, 292, 303\*, 307\*, 314, 319, 326, 344, 363\*, 419, and 444 of the HSF1 sequence (Table [2](#T2){ref-type="table"}; \*previously reported sites). No phosphorylated Thr or Tyr residues were discovered. This analysis covered \>90% of the HSF1 sequence. No information was obtained about residues 176--184, 202--206, 225--227, 241--256, 270--284 and 427--432.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Residues phosphorylated in HSF1 from heat-treated cells.
:::
Peptide N Phospho-residue Dig.
------------------------------------------------------------------------ --- ----------------- --------------
VEEA**pS**PGRP**pS**[S]{.underline}VD[T]{.underline}LL**pS**PTALIDSILR 4 314, 319, 326 Trp.
VKEEPP**pS**PPQ**pS**PR 2 303, 307 Trp.
GHTDTEGRPP**pS**PPP[TST]{.underline}PEK 3 363 Trp.
VVHIEQGGLVKPERDD[T]{.underline}EFQHPCFLR 1 (97) Trp.
GHTDTEGRPP**pS**PPP[TST]{.underline}PEK\* 2 363 Trp.
VKEEPP**pS**PPQ**pS**PR\* 2 303, 307 Trp.
QF**pS**LEHVHGSGPY\* 1 230 Trp.
VEEA**pS**PGRP**pS**SVDTLL**pS**PTALIDSILR\* 3 314, 319, 326 Trp, Chytrp.
VEEASPGRPSSVDTLL**pS**P[T]{.underline}ALIDSILR 2 326 Trp, Chytrp.
VEEA**pS**PGRPSSVDTLL**pS**PTALIDSILR 2 314, 326 Trp, Chytrp.
VKEEPP**pS**PPQSPR 1 303 Trp, Chytrp.
GHTDTEGRPP**pS**PPPTSTPEK 2 363 Trp, Chytrp.
KVT**pS**VSTLKS 1 121 Trp, Chytrp.
PLSS**pS**PLVR 1 292 Trp, Chytrp.
**pS**LEHVHGSGPY 1 230 Trp, Chytrp.
DARGHTDTEGRPP**pS**PPPTSTPEKCLSVACL 2 363 AspN
DERPL[SS]{.underline}**pS**PLVRVK 2 292 AspN
DLF**pS**PSVTVP 1 419 AspN
DSSLASIQELL**pS**PQEPPRPPEAENSSP 1 444 AspN
DTLL**pS**PTALI 1 326 AspN
PASVTALTDARGHTDTEGRPP**pS**PPPTSTPE 1 363 Glu-C
LL**pS**PQEPPRPPEAEN 1 444 Glu-C
SEPAPA**pS**VTALTDARGHTDTE 1 344 Glu-C
A**pS**PGRPSSVDTLL**pS**PTALID 2 314, 326 Glu-C
LF**pS**PSVTVPD 1 419 Glu-C
N: Unambiguously identified phosphoresidues are bolded and are preceded by p. Where identification was ambiguous, candidate residues are underlined. Dig.: proteinase used for digestion of HSF1. \*Radiolabeled peptide. Trp: trypsin; Chytrp: chymotrypsin. For experimental details relating to analysis of HSF1 phosphopeptides and identification of phosphorylated residues see Methods.
:::
Phosphorylation of HSF1 residue Ser^326^
----------------------------------------
Based on results from the above-described alanine scan HSF1 residues 326, 328, 511 and/or 513 were considered potential sites for regulatory phosphorylation. Of these residues only Ser^326^was actually found phosphorylated in HSF1 from heat-shocked cells. Not all residues that were identified as targets of phosphorylation by mass spectrometry and/or sequencing had been substituted individually in the earlier alanine scan. To rule out the possibility that an effect of substitution of a phosphorylated residue had somehow been masked or otherwise modulated (in the case of Ser^326^) by other substitutions present in the same mutant, single substitutions were prepared and examined in the transactivation assay. All substitutions of phosphorylated serines except for the S326A substitution displayed heat-induced activities comparable to that of parent factor (Table [1](#T1){ref-type="table"}, bottom). The S326A substitution was only about 40% as active as the parent factor (35--55% in individual experiments). To test whether substitution of Ser^326^affected not only heat-induced but also chemically induced HSF1 activity, the activity of the S326A mutant was tested in cells exposed to CdCl~2~. Induction by CdCl~2~was found to be similarly impaired as induction by heat (Figure [1D](#F1){ref-type="fig"}). Note that this reduced activity phenotype was not due to a reduced level of accumulation of the mutant as evidenced by the anti-FLAG western blot shown in Figure [1E](#F1){ref-type="fig"}. Nucleotide sequencing of the entire S326A-coding sequence confirmed that it did not contain any additional mutation. Furthermore, a second copy of mutant S326A obtained in a separate mutagenesis experiment had a similarly impaired stress-induced activity as the original copy.
Rapid phosphorylation of HSF1 residue Ser^326^during heat stress
----------------------------------------------------------------
A phosphopeptide antibody (referred to below as pSer^326^antibody) was prepared to specifically monitor phosphorylation of Ser^326^. Heat-induced rates of overall phosphorylation of HSF1 (i.e., phosphorylation of all available sites), of specific phosphorylation of Ser^326^and of factor oligomerization were compared in the experiment shown in Figure [2](#F2){ref-type="fig"} (panels A and B). Cultures were either not heat-treated or heat-treated at 44°C for 5, 10, 15 or 30 min. HSF1 trimerization was assessed by native anti-HSF1 blot. As best seen by following the disappearance of the monomeric form, most HSF1 was or was in the process of becoming oligomeric within 5 min of heat treatment (Figure [2A](#F2){ref-type="fig"}). For examining overall and Ser^326^-specific phosphorylation, HSF1 was immunoprecipitated using pSer^326^antibody. Immunoprecipitated HSF1 (Figure [2B](#F2){ref-type="fig"}, \"ip\" lanes) and HSF1 in extract (\"lysate\" lanes) were detected by anti-HSF1 western blot. To better resolve different HSF1 species, electrophoresis was continued until the 50 kDa marker had migrated to the end of the gel (only the region containing HSF1 forms, i.e. from about 70 kD protein size, is shown in Figure [2B](#F2){ref-type="fig"}). Results suggested that unstressed cells contained a small amount of HSF1 that could be immunoprecipitated by the pSer^326^antibody (\"ip\", first lane). Note that, as would have been expected, the most rapidly migrating immunoprecipitated HSF1 form was slower than the fastest species present in extract (compare 0-min \"ip\" and \"lysate\" lanes). Upon heat treatment of the cells, amounts of HSF1 precipitatable from extracts by the pSer^326^antibody increased about 3 fold (see quantitation below \"ip\" lanes). This increase was essentially complete after only 5 min of heat treatment. Overall phosphorylation as revealed by a shift from faster- to slower-migrating HSF1 species occurred more slowly and continued over the entire 30-min course of heat treatment (\"lysate\" lanes). A comparable gradual shift to slower-migrating species was also observed for pSer^326^-containing HSF1 species (\"ip\" lanes). Thus, heat treatment appeared to result in a substantial increase in the level of phosphorylation of Ser^326^. This increase in phosphorylation occurred about as rapidly as factor oligomerization and preceded phosphorylation of most other sites.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Heat-induced oligomerization and phosphorylation of HSF1**. (**A**) Native anti-HSF1 blot showing heat-induced oligomerization of endogenous HSF1 in response to heat treatment at 44°C (HS) for the times indicated on top of the blot. T: HSF1 trimers; D: heterodimers; M: monomers. (**B**) Parallel cultures to those used in A for an analysis of HSF1 oligomerization were employed here for an examination of global or Ser^326^-specific phosphorylation of endogenous HSF1. The anti-HSF1 western blot shown reports the distribution of HSF1 forms of different apparent size (reflecting different levels of phosphorylation) in extract samples (lysate) or in protein immunoprecipitated from the same extracts by pSer^326^antibody (ip). The portion of the blot depicted only shows protein signals larger than about 70 kD. (**C**) Detection of heat-induced phosphorylation of Ser^326^by western blot using anti-pSer^326^antibody. Parallel cultures were transfected with small amounts of FLAG-HSF1. One day later, the cultures were either left untreated (C) or were heat-treated for 30 min at 44°C (HS) and were processed for western blot immediately following the heat treatment. The anti-pSer^326^blot reports on induction of Ser^326^phosphorylation by heat, and the parallel anti-FLAG blot shows that similar amounts of FLAG-HSF1 were compared in the anti-pSer^326^blot. Data from densitometry are shown below the blots. Brackets shown on the side of blots indicate lengths of regions scanned. In A, the monomer signal was quantitated.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Comparative analyses in HSF1-negative mouse embryo fibroblasts**. (**A**) Comparison of DNA-binding abilities of HSF1 and HSF1 mutant S326A. An electrophoretic mobility shift assay was carried out using an HSE DNA probe and extracts from heat-treated or not-heat-treated cells that had been transfected one day earlier with the constructs indicated on top of the gel. An anti-HSF1 western blot of the same samples reporting on the relative levels of expression of HSF1 forms from the different constructs is shown below the gel. HSF1/HSE: HSF1-DNA complex; NS: nonspecific signal, serving as a loading control for the group of samples from not-heated cells, and, independently, for the group of samples from heat-treated cells; HS: heat-treated for 30 min at 43°C; C: not heat-treated. (**B**) Heat-induced transactivation of endogenous *hsp70*gene(s) in cells transfected with small amounts of the constructs indicated above the blots and with luciferase reporters HSP70-fLUC and pRL-CMV. One day after transfection, cultures were either left untreated (C) or were heat-treated for 30 min at 43°C (HS). Extracts were prepared after 6 h of further incubation at 37°C and were used for western blots. The top blot was probed with an antibody recognizing HSP70, whereas the bottom blot was probed with HSF1 antibody. The bottom blot demonstrates that mutant and wildtype HSF1 forms accumulated to similar levels. Note the presence of a weak nonspecific signal present in all lanes that happened to co-migrate with the transfected HSF1 forms. (**C**) Quantitative comparisons of data of B. (**D**) Relative luciferase reporter activities in the same extracts that were analyzed for HSP70 in B. Representative results from one of several independent experiments are shown in this Figure.
:::
:::
Heat-induced phosphorylation of Ser^326^was confirmed by a second experiment, in which HSF1 phosphorylated at Ser^326^was detected by anti-pSer^326^western blot. Because of the low avidity of the antibody, HSF1 needed to be enriched prior to western blot. Large cultures (in 100 mm plates) were transfected with small amounts of expression construct FLAG-HSF1. One day later, half of the cultures were heat-treated at 44°C for 30 min, and tagged HSF1 was immunoprecipitated using an anti-FLAG resin. Immune-isolated material was then analyzed by western blot using pSer^326^and FLAG antibodies (Figure [2C](#F2){ref-type="fig"}). Recovery of FLAG-HSF1 from heat-treated and untreated cells was comparable (anti-FLAG blot on top). A substantially larger fraction (2.5 fold) of factor from heat-treated cells than from not-heat-treated cells reacted with the pSer^326^antibody.
Phosphorylation of Ser^326^specifically enhances HSF1 transactivation competence
--------------------------------------------------------------------------------
To examine the effects of substitution of Ser^326^in an otherwise wildtype HSF1 background, use was made of an HSF1-negative mouse cell line prepared previously by McMillan et al. \[[@B31]\]. Analyses were carried out one day after transfection of expression constructs for HSF1, substitution mutant S326A (in wildtype HSF1 not LEXA-HSF1 background) and control protein β-galactosidase. First, it was examined whether phosphorylation of Ser^326^affected the first step of HSF1 activation, which step involves acquisition of HSE DNA-binding activity and nuclear localization. Electrophoretic mobility shift assay revealed that HSE DNA-binding activities of wildtype HSF1 and mutant S326A were heat-induced to similar levels (Figure [3A](#F3){ref-type="fig"}). Nuclear localization was assayed by standard fractionation of cell extracts and anti-HSF1 western blot. Comparable amounts of wildtype HSF1 and mutant S326A were present in the nuclear fraction of heat-treated cells (data not shown). Thus, phosphorylation of Ser^326^did not affect the first step of HSF1 activation. To probe the second activation step, i.e., acquisition of transactivation competence, HSF1-negative cells were co-transfected with the above expression constructs and with reporter constructs HSP70-fLUC and rLUC. Cultures either were left untreated or were heat-treated at 43°C for 30 min and further incubated for 6 hours. Transactivation competence was estimated by western blot of endogenous HSP70 (Figure [3](#F3){ref-type="fig"}, panels B and C) and by luciferase assay (Figure [3D](#F3){ref-type="fig"}). Heat-induced expression of endogenous HSP70 was reduced by 80% in the cells expressing the S326A mutant of HSF1 when compared to the cells expressing wildtype HSF1. The transfected luciferase reporter was reduced by 50%. The greater effect on HSP70 expression is likely explained by the difference in RNA/protein stability between HSP70 and luciferase.
Discussion
==========
The present study attempted for the first time to examine phosphorylation of HSF1 in cells exposed to a stress in a comprehensive fashion. Although HSF/HSF1 is activated in cells exposed to various types of stressful events, in the interest of being able to complete a thorough analysis, we decided to focus on phosphorylation of HSF1 in cells responding to a single type of stress, i.e., a heat stress. Our study identified twelve serine residues in human HSF1 that are phosphorylated in heat-stressed cells. Eight of these residues represent phosphorylation sites that were not previously known, i.e., Ser^121^, Ser^292^, Ser^314^, Ser^319^, Ser^326^, Ser^344^, Ser^419^, and Ser^444^. Phosphorylation of all residues previously found to be phosphorylated *in vivo*, i.e., Ser^230^, Ser^303^, Ser^307^and Ser^363^, was confirmed. No phosphorylation of Thr^142^or any other Thr or Tyr residue was observed.
Because the importance of the various phosphorylation events for the activation of human HSF1 could not be predicted, an effort was made to ensure that the basic transactivation assay used in the present study was capable of reporting even relatively minor impairments in the activity of HSF1 mutants as well as examined the exogenous HSF1 forms under conditions that differed as little as possible from those encountered by endogenous factor. Although HSF1-deficient mouse cells were available and were used in later experiments, our initial goal was to identify mutants of human HSF1 that were functionally deficient in human cells. Therefore, mutants were prepared in the LEXA-HSF1 background, allowing us to test effects of mutations in cells containing endogenous HSF1. Because hyperphosphorylation was expected to affect HSF1 transcriptional competence rather than HSE DNA-binding ability, use of an HSF1 form identical to HSF1 except for a substituted DNA-binding domain appeared justified. For obvious reasons, an HSF1 mutant with an impaired activity was not available when assay conditions needed to be established. In the absence of such a mutant, assay conditions were defined, under which reporter activity increased proportionally with amounts of LEXA-HSF1 expression construct transfected. Under the chosen conditions, exogenous HSF1 (i.e., LEXA-HSF1) was expressed at a comparable level as endogenous HSF1. Hence, these conditions also satisfied our second criterion that was to assay exogenous HSF1 in an intracellular situation that closely resembled that encountered by endogenous HSF1. The importance of the latter criterion is exemplified by the previous observation that substantially overexpressed exogenous HSF1 is trimeric and DNA binding in the absence of a stress, whereas endogenous HSF1 is not trimeric and DNA binding under the same conditions \[[@B7]\]. In the transactivation assays used in the present study, exogenous HSF1 did not specifically bind DNA in the absence of a stress.
The above-discussed differences between the transactivation assay used in the present study and assays employed in earlier studies provide a ready explanation for the observed differences in phenotypes of mutants S(303/307)A and S307A. The latter substitutions (also in LEXA-HSF1 background) had been examined in a previous study by our laboratory and were found to be active in the absence of a stress \[[@B28]\] (see also \[[@B9],[@B23],[@B25]\]). In this earlier study, HSF1 forms were substantially overexpressed, resulting in accumulation of homotrimeric factors in the absence of a stress. Therefore, the experiments were only capable of assessing effects of mutations on HSF1 transcriptional competence. When examined using the assay of the present study, in which assay oligomerization of exogenous HSF1 forms is regulated, the Ser^307^and Ser^303^/Ser^307^substitutions could be expected to be inactive in the absence of a stress, provided that the mutations only affected HSF1 transactivation competence and not also oligomerization. As shown in Table [1](#T1){ref-type="table"}, this expected result was observed. In heat-stressed cells, however, the Ser^307^and Ser^303^/Ser^307^substitutions exceeded the activity of the parent factor. This finding is consistent with a role of phosphorylation at Ser^303^and Ser^307^in down-modulation of HSF1 activity during a heat stress or, more likely, during recovery from the stress. Such a role has been proposed previously by others (e.g., \[[@B23],[@B27],[@B32]\]). Also compatible with this hypothesis is that, in a tryptic digest of HSF1 isolated from cells pulse-labeled with ^32^PO~4~during a 44°C/45 min heat treatment, peptide 297--309 was among the most intensely radiolabeled peptides (not shown). This finding implied that phosphorylation of Ser^303^and/or Ser^307^occurred during heat treatment. Hietakangas et al. recently confirmed that Ser^303^is inducibly phosphorylated by western blot experiments using a phosphopeptide antibody recognizing pSer^303^\[[@B33]\]. As Ser^303^phosphorylation may require prior phosphorylation of Ser^307^\[[@B23]\], phosphorylation of Ser^307^is likely also heat-inducible.
The present study identified HSF1 residue Ser^326^as a dominant target of regulatory phosphorylation during activation of the factor by a heat stress. Luciferase reporter assays indicated that phosphorylation of Ser^326^causes the heat-induced activity of HSF1 to at least double. The transactivation assays in which endogenous HSP70 was used as the endpoint revealed that this enhancement of HSF1 activity translates into a fivefold increase in accumulation of HSP70. While our study did not address this issue, it seems likely that the observed five-fold enhancement of HSP70 expression resulting from phosphorylation of Ser^326^is physiologically important. A previous study demonstrated that a five to eight fold impairment in heat-induced HSP70 expression led to substantially diminished thermotolerance of the affected mouse embryo fibroblast cells \[[@B34]\]. Phosphorylation of Ser^326^also appears to significantly contribute to HSF1 activity induced in cells stressed by exposure to CdCl~2~.
Our analyses suggested that, individually, phosphorylation of none of the other residues identified as being phosphorylated in heat-treated cells significantly contributes to HSF1 activity. The possibility was considered that phosphorylation of several of these residues may be required for producing a clearly detectable effect. Although this possibility was not examined exhaustively, several combinations of substitutions including or excluding the S326A substitution were tested by transactivation assay (data not shown). These experiments failed to uncover evidence for a functional effect of phosphorylation on residues other than Ser^326^. It is puzzling that HSF1 is phosphorylated on a number of residues (e.g., Ser^121^, Ser^230^, Ser^292^, Ser^314^, Ser^319^, Ser^344^, Ser^419^, and Ser^444^) whose phosphorylation does not appear to affect factor activity. The possibility cannot be formally ruled out that phosphorylation of some of these residues may reflect artifacts due to differences in phosphorylation of exogenous and endogenous HSF1. A more reasonable explanation may be that this phosphorylation may be relevant under conditions not tested in the present study. Such conditions may include different types of stresses or different levels of stresses used to activate HSF1. They may even relate to differences in the transactivation assays used that may result in preferential assessment of different facets of HSF1 activation. This latter explanation may apply to phosphorylation of Ser^230^that was previously reported to contribute to activation of HSF1 by heat stress \[[@B19]\]. Another likely possibility is suggested by the fact that HSF1 not only transactivates HSP genes but also participates in the regulation of several important signaling pathways (e.g., \[[@B35]-[@B38]\]). Phosphorylation of Ser residues that appears gratuitous with respect to regulation of HSP expression may affect interactions of HSF1 with components of these other pathways and alter their activity.
In agreement with earlier work, heat stress induced rapid trimerization of HSF1. The substantial enhancement of phosphorylation of Ser^326^that was induced by heat stress occurred within a similar time frame. This rapid rate of phosphorylation of Ser^326^was commensurate with what was expected for a phosphorylation event that was critical for heat stress activation of HSF1. Most other phosphorylation events that could be monitored by their effect on gel mobility occurred more slowly. Hence, it was possible that some of these later events required prior phosphorylation of Ser^326^. However, that at least some of this phosphorylation occurred independent of Ser^326^phosphorylation was suggested by the observation of a heat-induced SDS-PAGE mobility shift for mutant S326A (data not shown).
The present study provides evidence that phosphorylation of Ser^326^stimulates the transcription-enhancing activity of HSF1 but not its DNA-binding activity. How this phosphorylation results in increased transcriptional competence of HSF1 remains to be elucidated. The observation that substitution of Ser^326^with neither Asp nor Glu reproduced the effect of phosphorylation on factor activity (data not shown) suggested that the mechanism is not based on simple charge repulsion. Perhaps, phosphorylation of Ser^326^induces a local conformational change that affects binding of a chaperone complex or another regulatory protein to the nearby regulatory domain that is known to be involved in repression of transcriptional competence \[[@B6]\]. Alternatively, pSer^326^may be a critical aspect of a binding site for an unknown co-activator. Identification of the protein kinase that phosphorylates Ser^326^in heat-stressed cells would be helpful for determining whether the level of phosphorylation of the residue is actively regulated and, if this were the case, by what stress-induced mechanism. Unfortunately, a search of the sequence within which Ser^326^is embedded for protein kinase sites using the NetPhosp program \[[@B39]\] did not provide any useful information about candidate protein kinases.
Conclusion
==========
The present article is concerned with regulation of human HSF1, which is a key factor mediating the transcriptional response of human cells to physical and chemical stresses and a coregulator of other important signaling pathways (e.g.. \[[@B35]-[@B38]\]). HSF1 has even been discovered to regulate aging and age-related disease \[[@B40],[@B41]\]. To arrive at a better description of the mechanisms that enable cells to respond to various stresses by transiently upregulating HSP gene expression, it will be important to learn about the extent to which phosphorylation of HSF1 modulates these responses as well as to discover how phosphorylation/dephosphorylation of HSF1 itself is regulated by stresses. Furthermore, it can be expected that a thorough understanding of regulatory phosphorylation of HSF1 will advance our knowledge about what controls the interactions of this factor with other pathways as well as likely will, through the eventual identification of regulated protein kinases and phosphatases involved in HSF1 phosphorylation and dephosphorylation, lead to the identification of new connections with additional regulatory systems. The present study represents an initial contribution towards these larger goals. Our systematic analysis of HSF1 phosphorylation in heat-stressed cells identified twelve phosphorylated Ser residues, of which eight were not previously known. Mutagenesis and functional experiments revealed that newly identified phosphoSer^326^plays an important role in heat activation of HSF1 transcriptional activity as evidenced by the fact that substitution of this residue reduced HSP70 accumulation several fold. Phenotypes for substitutions of Ser^303^and Ser^307^were observed that are consistent with the previously proposed function of phosphorylation of these residues in HSF1 deactivation. Although no evidence for functional roles of other phosphoserines could be obtained in this study, knowledge of the identity of most or all residues phosphorylated in heat-activated HSF1 should greatly facilitate further directed experiments to test the potential importance of their phosphorylation in the various processes and interactions in which HSF1 is known to participate. It cannot be excluded that through the use of different transactivation assays functions of the latter phosphoresidues in heat regulation of HSF1 activity may be discovered that escaped detection in this study.
Methods
=======
Antibodies
----------
PhosphoSer^326^-specific rabbit polyclonal antibody was raised against peptide CSVDTLLpSTAL. The antiserum was positively and negatively affinity-purified on immobilized phosphorylated and unphosphorylated peptide. For immunoprecipitations, purified antibody was cross-linked to a resin using the Seize Primary Immunoprecipitation Kit (Pierce). HSF1 antiserum was from StressGen Biotechnologies; FLAG antibody M5, FLAG resin M2 and tubulin antibody were from Sigma; HSP70 antibody 4G4 was from Affinity Bioreagents. Mouse monoclonal antibody 4G4, which antibody was raised against human Hsp70, also recognizes mouse Hsp70. Antibody signals were detected by chemifluorescence and were quantitated on a Molecular Dynamics Storm system.
Cell culture
------------
Hela-CAT cells \[[@B42]\] were maintained at 37°C and 5% CO~2~in DMEM containing 10% fetal bovine serum, 100U/ml penicillin and 100 μg/ml streptomycin. HSF1-negative mouse embryo fibroblasts were grown in supplemented DMEM as described by McMillan et al. \[[@B31]\].
Expression constructs and site-directed mutagenesis
---------------------------------------------------
Sequences coding for complete (human) HSF1 (residues 1--529), LEXA-HSF1 (containing residues 1--87 of LEXA and residues 79--529 of human HSF1) and amino-terminally FLAG-tagged derivatives were subcloned into pcDNA3.1(+) (Invitrogen), placing the sequences under the control of a CMV promoter \[[@B7]\]. These constructs were named HSF1, LEXA-HSF1, FLAG-HSF1 and FLAG-LEXA-HSF1, respectively. For the alanine scan, LEXA-HSF1 and FLAG-LEXA-HSF1, were used as templates for QuikChange^R^site-directed mutagenesis (Stratagene Instruction Manual). Complementary primer pairs replaced single or multiple Ser, Thr or Tyr codons with Ala codons. Potential mutant genes were characterized by restriction analysis, expression in a rabbit reticulocyte lysate-based transcription and translation system (T7 Quick TNT system, Promega) and nucleotide sequence analysis. Several mutations were also introduced into construct HSF1 using the same approach. The β-galactosidase expression construct (B-GAL) used was pcDNA3.1/His/ LacZ (Invitrogen).
Transactivation assay
---------------------
Reporter construct LEXA-fLUC was described previously \[[@B42]\]. Reporter gene HSP70-fLUC was obtained by subcloning promoter and RNA leader sequences of the human HSP70B gene into a plasmid containing a firefly luciferase gene. Constructs containing a constitutively expressed *Renilla*luciferase gene (pRL-TK, pRL-CMV) were obtained from Promega. Cultures in 96-well plates were transfected using a rapid transfection protocol for Lipofectamine 2000 (Gibco). Typically, each well received 0.75--1.0 μl of Lipofectamine 2000 in 25 μl Opti-MEM and a DNA master mixture (88.25 ng) in 25 μl Opti-MEM consisting of 80 ng of LEXA-fLUC or HSP70-fLUC, 0.25 ng of pRL-Tk or 0.1 ng of pRL-CMV, 0.1--0.5 ng of LEXA-HSF1 or HSF1 (or a mutant) and 7.5--8.05 ng of B-GAL. 80,000 cells in 100 μl DMEM were added subsequently. Typically, transfections were carried out in triplicate. Transfected cells were incubated for 16--20 hours, heat-treated at 44°C for 30 min (unless indicated otherwise) or not heat-treated and harvested 6--7 hours later. The length of the period between heat treatment and cell harvest was optimized for expression of firefly luciferase and recovery of *Renilla*luciferase activity. Luciferase activity was measured using the Dual Luciferase Kit (Promega) and a Stratec plate luminometer. Typically, luciferase activity assays were performed in triplicate.
Electrophoretic mobility shift assay
------------------------------------
Cells were transfected in 100 mm-dishes with Lipofectamine PLUS, 25 ng of HSF1 or mutant HSF1 expression construct and 2.975 μg B-GAL according to the manufacturer\'s instructions, incubated for 24 hours and then either heat-treated or left untreated. PBS-washed cells were resuspended in buffer C (20 mM Hepes, pH7.9, 0.42 M NaCl, 1.5 mM MgCl~2~, 0.2 mM EDTA, 0.5 mM dithiothreitol, \"Complete\" Protease Inhibitor Cocktail (Roche), 25% glycerol). Extracts were prepared by three cycles of quick-freezing and thawing, and clarification by centrifugation. The same protocol was also employed for preparing extracts used for native gel electrophoresis. Extracts were either used immediately or were stored at -70°C. Electrophoretic mobility shift assays were carried out essentially as described before \[[@B7]\]. Signals were detected and quantified using a Molecular Dynamics PhosphorImager.
Subcellular fractionation
-------------------------
Cells were transfected as described under the previous section. Fractionation was performed using NE-PER Nuclear and Cytoplasmic Extraction Reagents from Pierce according to the manufacturer\'s protocol. Correct operation of the protocol was verified by following endogenous HSF1 whose localization had been determined previously.
Analysis of HSF1 phosphopeptides and identification of phosphorylated residues
------------------------------------------------------------------------------
9.0 × 10^6^Hela-CAT cells in 150-mm dishes were transfected with 32 μg of FLAG-HSF1 expression construct. For experiments involving analysis of radiolabeled phosphopeptides, 20 hours after transfection each culture was washed once with 25 ml phosphate-free DMEM containing 5% FCS and was incubated for 1 hour in the same medium. Medium was replaced by fresh medium further containing 2 mCi of ^32^P-orthophosphate (NEX011, NEN), and cells were incubated for 3 hours at 37°C. After heat treatment for 45 min at 44°C, cells were washed with ice-cold PBS and lysed by incubation for 15 min at room temperature in 3 ml Mper buffer (Pierce) supplemented with 0.5 mM NaV~3~, 5 mM NaF, 150 mM NaCl, 1 μM ocadaic acid and \"Complete\" Protease Inhibitor Cocktail (Roche). After removal of debris, extract was incubated overnight at 4°C with 80 μl of FLAG M2 resin. Resin was washed extensively with Mper buffer containing 150 mM NaCl and Mper buffer alone, and FLAG-HSF1 was eluted with 120 μl of 6X SDS-PAGE sample buffer. Subsequent to electrophoresis on a Tris-Tricine high-resolution SDS-PAGE gel \[[@B43]\], FLAG-HSF1 was Coomassie-stained, and gel pieces containing the most highly phosphorylated species were processed by the Keck Foundation Biotechnology Resource Laboratory at Yale (Kenneth Williams) for proteolytic digestion, mass spectrometric analysis, and radiochemical and normal peptide sequencing. In other experiments, unlabeled, purified FLAG-HSF1 was prepared using a similar protocol. Sequence analysis of these preparations was performed at the Harvard Microchemistry Facility using microcapillary reverse-phase HPLC nano-electrospray tandem mass spectrometry on a Finnigan LCQ DECA quadrupole ion trap mass spectrometer.
Other methods
-------------
For most immunoprecipitation and HSF1 expression experiments, cells were lysed in MPer-buffer supplemented with 150 mM NaCl and \"Complete\" Protease Inhibitor Cocktail from Roche. Protein concentrations in extracts were measured using a Bradford assay (Protein Assay reagent from Bio-Rad Laboratories). Results were used to adjust protein concentrations in extracts to be compared.
Authors\' contributions
=======================
TG carried out the bulk of the experimental work presented herein. He also prepared all tables and designed all figures. FB prepared a first set of LEXA-HSF1 substitution mutants. He also designed all oligonucleotide primers used in the study as well as analyzed all sequence information. WSL carried out or supervised all work concerning the identification of phosphorylated residues of HSF1 using LC/MS/MS, interpreted all data obtained and consulted on experimental design. RV conceived of the study, participated in its design and coordination, and drafted the manuscript. All authors read and approved the manuscript.
Acknowledgements
================
We thank Ivor J. Benjamin for providing the HSF1-deficient cell line, Corneliu Sologon for confirmatory experiments, Guenther Kraus and James Hnatyszyn for nucleotide sequencing, and Kenneth Williams for mass spectrometric analyses. Lawrence Boise and Gennaro D\'Urso critically read the manuscript. This work was supported by NIH Grant GM31125.
|
PubMed Central
|
2024-06-05T03:55:55.167335
|
2005-3-11
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079796/",
"journal": "BMC Biochem. 2005 Mar 11; 6:4",
"authors": [
{
"first": "Toumy",
"last": "Guettouche"
},
{
"first": "Frank",
"last": "Boellmann"
},
{
"first": "William S",
"last": "Lane"
},
{
"first": "Richard",
"last": "Voellmy"
}
]
}
|
PMC1079797
|
Background
==========
Fatty acid composition of cell membrane is an important determinant of cell function \[[@B1]\]. Manipulation of cellular fatty acid composition has been a widely used approach to modulating the biological responsiveness of different cell types. Recently, fatty acid profile, particularly the ratio of omega-6 (n-6) to omega-3 (n-3) polyunsaturated fatty acids, of cells or tissues has become a biomarker for monitoring the outcome of dietary interventions (i.e., fatty acid supplementation) and for identifying the risk factors for lipid related diseases (e.g. cardiovascular disease) \[[@B2]\]. Measurement of the n-6/n-3 fatty acid ratio can be also used to identify animal phenotypes, such as the fat-1 transgenic mice that we created recently \[[@B3]\]. Thus, analysis of fatty acid composition is a commonly used technique in lipid research.
Analysis of fatty acid composition is usually carried out by gas chromatography (GC). Conventionally, preparation of samples for GC involves two separate procedures: extraction and methylation. Lipids are usually extracted from cells or tissue homogenates by using organic solvents such as chroloform/methanol \[[@B4]\]. This procedure is time and material consuming, potentially causes sample loss and contamination, and generates organic wastes. These problems become more apparent when the number of samples to be extracted and analyzed is large and/or the specimen size is small.
This study validated a method that combines the extraction and methylation into a single step. Our results showed that this simplified method without the need of prior extraction yielded desirable outcome.
Results and discussion
======================
The samples tested, including cultured cells, tissue homogenates, and red blood cells, were divided into two aliquots. One was analyzed by the conventional method and the other analyzed by the simplified method, under the same chromatographic conditions. The results obtained using the conventional and simplified methods were compared. Figure [1](#F1){ref-type="fig"} shows a representative set of the fatty acid profiles of mouse heart tissue derived from the two methods. Generally, the results obtained using the simplified method without prior extraction are as good as or even better than those obtained using the standard method, particularly when the sample size is small. As shown in Table [1](#T1){ref-type="table"}, the recovery of long chain fatty acids from tissue samples by the simplified method is significantly higher than that by the traditional method, but there is no difference in relative fatty acid composition between the two methods. However, it was noted that some short and medium chain fatty acids (\<16C) could be lost, but none of those fatty acids with a chain length of 16 or more carbons was affected (Fig. [1](#F1){ref-type="fig"}). Similar outcomes were found in the tests using cultured cells, red blood cells and mouse-tails. These results indicate that the extraction of lipid prior to methylation is not necessary for the analysis of total long chain fatty acid composition by GC. We also found that the water in liquid samples, if its content is not more than 5% of the volume of BF3 solution added, did not affect significantly the analytic outcome. Therefore, the simplified method is suitable for both dried and liquid samples. However, the result obtained using the simplified method is limited to fatty acid profile of total lipids in the analyzed specimens. For studies requiring fatty acid composition of individual lipid class (e.g. phospholipid or triglyceride), prior extraction and separation of lipids are still necessary for their analysis.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Comparison of the results (gas chromatographs) of lipid analysis by the conventional and simplified methods. Mouse heart tissue was homogenized by grinding it up in liquid nitrogen and same amount (2 mg/sample) of the homogenate was used for analysis by each method under the same chromatographic conditions, as described in the Methods. Panels A: The GC result obtained using the conventional method. Panel B: The result obtained using the simplified method. Note, the fatty acid 23:0 (1 μg) was added to the sample before extraction as an internal standard.
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Quantitative results of lipid analysis by the conventional and simplified methods. Sample preparation and lipid analysis were performed as described in Fig.1. The quantity (absolute amount) of each fatty acid is represented by its peak area. (The initial weight of tissue samples for the two methods was the same.) The area percent of each fatty acid was calculated by dividing its peak area by the total peak area of the 8 fatty acids identified (excluding 23:0). Values are means ± SD of five (n = 5) measurements. Values for each fatty acid with the same letter do not differ significantly (p \< 0.05) between conventional and simplified methods.
:::
**Fatty Acids Identified** **Quantity**(Area of Peak: × 1,000 counts) **Composition**(% of total LCFA identified)
---------------------------- -------------------------------------------- --------------------------------------------- --------------- ---------------
**16:0** 26.3 ± 0.8^a^ 30.4 ± 0.6^b^ 14.6 ± 0.1^a^ 14.5 ± 0.6^a^
**18:0** 34.5 ± 1.2^a^ 37.5 ± 0.4^b^ 18.9 ± 0.2^a^ 18.5 ± 0.3^a^
**18:1(9)** 18.1 ± 0.6^a^ 21.6 ± 0.3^b^ 10.1 ± 0.1^a^ 10.3 ± 0.1^a^
**18:2(6)** 29.3 ± 0.9^a^ 35.3 ± 0.5^b^ 16.3 ± 0.2^a^ 16.8 ± 0.3^a^
**20:4(6)** 12.7 ± 0.5^a^ 15.0 ± 0.2^b^ 7.0 ± 0.2^a^ 7.1 ± 0.1^a^
**22:5(3)** 4.1 ± 0.2^a^ 5.0 ± 0.3^b^ 2.1 ± 0.1^a^ 2.3 ± 0.2^a^
**22:6(3)** 55.1 ± 1.6^a^ 64.4 ± 1.3^b^ 30.6 ± 0.2^a^ 30.7 ± 0.3^a^
**23:0 (std)** 42.9 ± 1.6^a^ 47.7 ± 1.0^b^
:::
Conclusion
==========
The present study has demonstrated that the lipid extraction procedure prior to methylation employed conventionally in lipid analysis of long chain (≥ 18 C) polyunsaturated fatty acids in biological samples can be omitted, without affecting the recovery of long chain fatty acids and their composition. This simplified method is suitable for analysis of long chain fatty acid composition in a variety of biological specimens, but not appropriate for quantification of medium and short chain (\<16C) fatty acids. Because the modified method is relatively simple and sensitive, it has a number of advantages including saving time and easy-to-use, reducing the potentials of sample loss and contamination, and requiring only small quantities of specimens and solvents. These advantages become more obvious in the case where the number of samples to be analyzed is large and/or the specimen size is small. For example, use of the simplified method would make the task of lipid analysis much easier in the large clinical trials that need to monitor polyunsaturated fatty acid composition (especially n-3 fatty acid contents) of red blood cells or other specimens from a large number of subjects. This new method is particularly useful for phenotype analysis of the transgenic animals that exhibit a unique fatty acid profile, such as the fat-1 transgenic mice that we generated recently \[[@B3]\].
Methods
=======
Conventional method
-------------------
Cell or tissue lipids were extracted by the procedures similar to the Folch method \[[@B4]\]. Chloroform/methanol (2:1, v/v) containing 0.005% butylated hydroxytoluene (as antioxidant) was added (usually 5 ml solvent added to 50--100 μl sample) and mixed vigorously for 1 min then left at 4°C overnight. One ml of 0.9% NaCl was added and mixed again. The chloroform phase containing lipids was collected. The remains were extracted with another 2 ml chloroform. The chloroform was pooled and dried under nitrogen and subjected to methylation. To monitor the recovery rate, the fatty acid C23:0 was added to the samples (usually 1 μg added to 2 mg tissue sample) as an internal standard.
Fatty acid methyl esters were prepared by methods similar to those described previously \[[@B5],[@B6]\] using BF~3~/methanol reagent (14% Boron Trifluoride). Lipid sample was mixed with 1 ml hexane in 16 ml glass tubes with Teflon-lined caps. BF~3~/MeOH reagent (1 ml) was added and the mixture was heated at 90--110°C in a metal block or a sand bath for 1 hour, cooled to room temperature and methyl esters extracted in the hexane phase after addition of 1 ml H~2~O. Samples were allowed to stand for 20--30 min, and then the upper hexane layer was removed and concentrated under nitrogen.
Fatty acid methyl esters were analyzed by gas chromatography using a fully automated HP5890 system equipped with a flame-ionization detector, as described previously \[[@B7]\] The chromatography utilized an Omegawax 250 capillary column (30 m × 0.25 mm I.D.). Peaks were identified by comparison with fatty acid standards (Nu-chek-Prep, Elysian, MN), and area and its percentage for each resolved peak were analyzed using a Perkin-Elmer M1 integrator.
Simplified method
-----------------
An aliquot of cell pellet or tissue homogenate (\<50 μl) in a glass methylation tube was mixed with 1 ml of hexane and 1 ml of 14% BF~3~/MeOH reagent. After blanketed with nitrogen, the mixture was heated at 100°C for 1 hour, cooled to room temperature and methyl esters extracted in the hexane phase following addition of 1 ml H~2~O. The samples were centrifuged for 1 minute, and then the upper hexane layer was removed and concentrated under nitrogen. Fatty acid methyl esters were analyzed by gas chromatography as described above.
Statistical analysis
--------------------
The results from the two methods were compared by using unpaired t-test, and *P*values of \<0.05 were considered significant. Results are means ± SD.
Authors\' contributions
=======================
**JXK**conceived of the study, participated in design and conduct of the experiments, and prepared the manuscript.
**JW**carried out sample preparation, GC analysis and data collection.
Acknowledgements
================
The authors thank Kelley Kallin for her technical assistance. This study was supported by grants from the American Cancer Society (RSG-03-140-01-CNE) and the American Institute for Cancer Research (2A017-REV) to J.X.K.
|
PubMed Central
|
2024-06-05T03:55:55.172920
|
2005-3-24
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079797/",
"journal": "BMC Biochem. 2005 Mar 24; 6:5",
"authors": [
{
"first": "Jing X",
"last": "Kang"
},
{
"first": "Jingdong",
"last": "Wang"
}
]
}
|
PMC1079798
|
Background
==========
Familial adenomatous polyposis (FAP) is an autosomal dominantly inherited disease and is caused by germline mutations in the adenomatous polyposis coli gene (APC) in chromosome 5q21 \[[@B1]\]. Somatic mutations of the APC gene occur in about 80% of sporadic colorectal cancers. APC encodes for a multimodal protein that plays an important role in the wnt-signalling pathway and in intercellular adhesion \[[@B2],[@B3]\]. The APC germline mutation has a penetrance which is close to 100% \[[@B4]\]. Untreated, the disease usually leads to the appearance of hundreds of adenomatous polyps in the colorectum between puberty and age 20 and to cancer by the early forties at the latest which is the most frequent reason for death in patients with FAP \[[@B5]\]. Attenuated forms of FAP (AFAP) are variations in phenotype. AFAP with less than 100 adenomatous polyps is diagnosed at a mean age of 44 years, and cancer is diagnosed at a mean age of 56 years \[[@B6]\]. Congenital hypertrophy of retinal pigment epithelium, upper gastrointestinal polyps, desmoid tumors, adrenal adenomas and osteomas are extracolonic FAP phenotypic features whose expression depends upon the mutated region of the APC gene \[[@B7]-[@B11]\].
The incidence of FAP is approximately 1 in 8000 \[[@B12]\]. Early detection, prophylactic surgery and lifelong surveillance are essential to prevent the development of colorectal cancer \[[@B13]\]. When an index patient is newly diagnosed in a center with an established FAP register, screening of family members with molecular diagnostics is carried out as a matter of routine and can detect the condition before the development of symptoms. It is rare for patients outside this system to be diagnosed prior to the development of symptoms. Also in patients with spontaneous new mutations of the APC gene, early clinical diagnosis can increase their descendants\' chance of survival. Lack of awareness and the absence of systematic family screening may cause diagnosis to be delayed in patients at risk of inherited colorectal cancer \[[@B14]\]. According to the literature, new mutations occur in 25% of all FAP cases and this predominantly young patient population is burdened with a high risk of colorectal cancer. It is therefore desirable that there are defined clinical algorithms for detecting the characteristic clinical signs of FAP, ideally within a multidisciplinary setting in which patients are entered on an FAP register.
In a retrospective analysis, we evaluated all clinical symptoms of FAP prior to surgery. We categorized these according to patient age and into colorectal and extracolonic manifestations of the disease. Our aim was to detect specific characteristics in each age group which could lead to an early clinical diagnosis and prevent the development of colorectal cancer.
Methods
=======
Definition
----------
The diagnosis of FAP was established clinically if more than 100 polyps were detected endoscopically in the colorectum.
Patient population
------------------
Between 1971 and 2000, we registered 157 patients (66 women, 85 men, 6 unknown) with an endoscopic diagnosis of FAP. The mean age at primary diagnosis was 34 (± 15 SD) years with a range of 5 to 73 years. In 151 cases, surgery was performed. Sixty-seven patients had colectomy with ileorectal anastomosis, 55 patients underwent restorative proctocolectomy with ileal pouch-anal anastomosis, and 10 had proctocolectomy with an end-ileostomy. Nineteen patients underwent various other surgical procedures. Presently, 83 patients remain alive and 28 patients have been lost to follow-up. Forty patients are known to have died: 20 of colorectal cancer, 5 due to desmoid tumors, 10 of other causes and 5 for unknown reasons.
A predominant early symptom could be defined and documented in 143 patients. Family screening revealed that 29 of these patients had affected relatives, while no FAP was detected in the families of the remaining 114 patients.
Evaluation of symptoms
----------------------
The predominant symptoms of FAP were analyzed and documented retrospectively in all 143 cases. The entire patient population was divided into three age groups according the time of primary diagnosis: less than 20, 20 to 40, and more than 40 years of age. Abdominal pain, pain during defecation, bloating, fatigue and loss of weight were categorized as non-specific symptoms (figure [1](#F1){ref-type="fig"}). Colonic bleeding, anemia, mucous discharge, diarrhea, paradoxical diarrhea and constipation were categorized as bleeding and stool disorders (figure [2](#F2){ref-type="fig"}).
Evaluation of colorectal and extracolonic manifestations
--------------------------------------------------------
The endoscopic reports were obtained and the pattern of polyps in the colorectum was divided into several groups: no rectal polyps, less than 10 rectal polyps, 10--100 polyps, and more than 100 polyps. We recorded the prevalence of carcinoma at the primary FAP diagnosis (table [1](#T1){ref-type="table"}). In addition, we documented any extracolonic manifestations of FAP: desmoid tumors (abdominal wall, mesenteric, retroperitoneal), gastric polyps, duodenal polyps and polyps of the small bowel (table [2](#T2){ref-type="table"}).
Data presentation
-----------------
The relative prevalence of symptoms in each age group is shown in figures [1](#F1){ref-type="fig"} and [2](#F2){ref-type="fig"}. Molecular findings are grouped according to the mutations detected in the APC gene (table [3](#T3){ref-type="table"}). Colorectal and extracolonic manifestations at primary diagnosis are presented for each of the recorded symptoms. Values in brackets indicate the percentage of affected patients in each subgroup (tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}).
Results
=======
Age and symptoms
----------------
The diagnosis of FAP was made on the basis of clinical symptoms in 66% of patients (94/143). Diagnosis was incidental in 6% (9/143) and the reasons are unknown in 8% (11/143).
Most patients (53%, 76/143) were between 20 and 40 years of age at primary diagnosis, of whom 75% (58/76) had symptoms. 32% (46/143) of our patients were older than 40 years at primary diagnosis, of whom 65% (30/46) had symptoms. Only 15% of patients (21/143) were less than 20 years old at diagnosis. Within this group five patients were under 10 years old. Only 33% (7/21) of patients under 20 years presented with symptoms.
Molecular diagnostics
---------------------
Table [3](#T3){ref-type="table"} shows the results of molecular diagnostics in 42 patients (42/143, 29%). These were carried out in 43% (9/21) of patients less than 20 years old, 28% (21/76) of patients 20--40 years old and 26% (12/46) of patients over 40 years of age. Eighteen out of 42 mutations in the APC gene (42.9%) were detected in exon 15, six (14.3%) were detected in exon 14, and six (14.3%) in exon 5. Exon 3 and exon 10 showed one mutation each (1/42, 2.3%).
Colorectal manifestations at primary diagnosis of FAP
-----------------------------------------------------
Table [1](#T1){ref-type="table"} shows the colorectal findings at primary diagnosis, when 63% (90/143) of our patients had more than 100 polyps in the colon and 67% (96/143) between 10 and 100 polyps in the rectum. Among the patients younger than 20 years, 57% (12/21) had more than 100 colonic polyps and 67% (14/21) had between 10 and 100 rectal polyps. Patients between the age of 20 and 40 years had the highest prevalence of rectal polyps. In 79% (60/76) of this age group 10--100 rectal polyps could be detected and only 1.3% (1/76) had no rectal polyps. Among patients over 40 years of age, 24% (11/46) had no rectal polyps but 63% (29/46) had more than 100 colonic polyps.
Carcinomas of the colon were detected in 17% (25/143) and rectal carcinomas in 12% (17/143). No patient younger than 20 years had developed colorectal cancer. In the group between 20 and 40 years old, rectal carcinoma was detected in 21% (10/76) and colon carcinoma in 17% (13/76). In patients older than 40 years of age, 15% (7/143) had rectal carcinoma and 50% (23/46) had colonic carcinoma.
Extracolonic manifestations at primary diagnosis of FAP
-------------------------------------------------------
Table [2](#T2){ref-type="table"} shows the extracolonic manifestations at primary diagnosis. Information about the absence or presence of desmoid tumors of the abdominal wall and the mesenterium was aviable in 119 patients and about retroperitoneal desmoids in 143 patients. Desmoid tumors of the abdominal wall were detected in 7% (8/119), mesenteric desmoids in 11% (13/119) and retroperitoneal desmoids in 4% (5/143). Information about the absence or presence of gastric gland polyps was aviable in 84 patients, about duodenal polyps in 79 and about polyps of the small bowel in 32 patients. Gastric gland polyps were found in 46% (39/84), duodenal polyps in 33% (26/79) and polyps of the small bowel in 22% (7/32).
No patients younger than 20 years suffered from desmoid tumors, but in 20% (3/15) gastric gland polyps and in 13% (2/15) duodenal polyps were discovered. The highest incidence of desmoid tumors (29%, 18/62) was found in the group between 20 and 40 years of age (8% abdominal wall, 15% mesenteric, 5% retroperitoneal). In two cases within this age group the diagnosis of FAP was incidental because the patient presented with a desmoid tumor of the abdominal wall. In patients between 20 and 40 years, gastric gland polyps were detected in 52% (27/52), duodenal polyps in 31% (16/52) and polyps of the small bowel in 14% (4/29). Only 14% (5/36) of patients older than 40 years had desmoid tumors at primary diagnosis of FAP (8% abdominal wall, 8% mesenteric, 2% retroperitoneal). And in this age group 53% (9/17) had gastric gland polyps, 67% (8/12) had duodenal polyps.
Non-specific symptoms at primary diagnosis of FAP
-------------------------------------------------
The commonest non-specific symptom was abdominal pain, found in 16% (21/129) of patients. It was most common in patients between 40 and 50 years of age, 30% of whom had abdominal pain as predominant symptom (6/20, figure [1](#F1){ref-type="fig"}). Abdominal pain occurred in 21% (19/90) -- 31% (11/36) of patients with the predominant manifestation in the colon, 15% (14/96) -- 24% (6/25) of patients with rectal predominance and 43% (3/7) with polyps of the small bowel (tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}). Fatigue was reported by 14% (18/130), loss of weight by 12% (15/125), and bloating by 10% (13/130). Pain during defecation was only described in 0.8% (1/130). Most patients who developed these symptoms were between 40 and 50 years old. Fatigue was related to colon carcinomas in 31% (11/36) and in 29% (2/7) to a predominant manifestation in the colon, in 29% (2/7) to polyps of the small bowel and in 27% (7/26) to duodenal polyps. Bloating had the highest association to 10--100 colon polyps, which was 29% (2/7), and to colon carcinomas which was 17% (6/36). Loss of weight showed a correlation to carcinomas of the rectum in 29% (5/17), carcinomas of the colon in 17% (6/36), polyps of the small bowel in 29% (2/7) as well as to duodenal polyps in 23% (6/26) (table [1](#T1){ref-type="table"}, [2](#T2){ref-type="table"}).
It has to be mentioned that most of our patients had no non-specific symptoms on primary diagnosis of FAP, i.e. no abdominal pain (76%, 108/143), no fatigue (78%, 112/143), no loss of weight (77%, 110/143), no bloating (82%, 117/143) and no pain during defecation (90%, 129/143).
Bleeding and stool disorders at primary diagnosis of FAP
--------------------------------------------------------
In 59% (84/143) tests for fecal occult blood were positive. Microscopic bleeding was detected in 72% (55/76) of patients between 20 and 40 years old and 50% (23/46) of patients older than 40 years of age. In 90% (81/90) more than 100 colon polyps, in 82% (14/17) rectal carcinomas, in 61% (22/36) colon carcinomas, in 73% (19/26) duodenal polyps and in 72% (28/39) gastric gland polyps could be detected in association to fecal occult blood. Colonic bleeding as an early symptom was present in 52% (74/143). This was found in 68% (52/76) of patients between 20 and 40 years old and in 35% (16/46) of patients older than 40 years. Colonic bleeding was found in 80% (72/90) of patients with more than 100 colon polyps, 82% (14/17) of patients with rectal carcinoma, 64% (61/96) of patients with 10--100 rectal polyps, 64% (25/39) of patients with gastric gland polyps and 64% (25/39) of those with duodenal polyps. 23% (33/143) of our patients suffered from anemia, 21% (16/76) of patients between 20 and 40 years, and 17% (8/46) of patients older than 40 years of age. Anemia was found in 44% (16/36) of patients with colon carcinoma, 35% (11/17) of patients with rectal carcinoma and 43% (3/7) of those with polyps of the small bowel. Diarrhea was reported in 31% (45/143). 42% (32/76) of patients between 20 and 40 years but only in 24% (11/46) older than 40 years suffered from this symptom. Diarrhea showed an association to rectal carcinomas in 65% (11/17), more than 100 colon polyps in 49% (44/90), duodenal polyps in 46% (12/26) and polyps of the small bowel in 43% (3/7) (tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}). 13% (19/143) of the patients reported mucous stools, 6% (9/143) constipation and 3% (5/143) alternating bowel habit. These were the least frequent stool and bleeding disorders in our sample (figure [2](#F2){ref-type="fig"}).
Discussion
==========
Most patients with FAP (80%) will be detected by more or less specific symptoms of colorectal disorders which lead to further diagnostic evaluation. Prior to the mid-nineties, when there was no FAP register, families of patients with FAP underwent no routine molecular screening, nor any further specific clinical surveillance. The diagnosis of FAP was verified in 29% of patients by molecular diagnostics, which became available for routine clinical use in the 1990s. The disperson of APC mutations in our study was similar to evaluations concerning other patient populations \[[@B15]\]. Our study also demonstrates the importance of clinical FAP registers, with systematic family screening to identify presymptomatic carriers with APC gene mutations \[[@B16],[@B17]\]. Recent evaluations of physicians\' awareness of genetic tests for inherited colorectal cancer showed a need for information concerning the availability and appropriate use of these tests \[[@B18]\]. In principle, genetic testing of index patients and their offspring can be carried out without using a register, but the lack of awareness and possibly the financial costs prevent an effective application of existing knowledge and expertise.
Most patients in our hospital were diagnosed between age 20 and 40 (53%). In 75% (57/76) the diagnosis was made clinically. They mainly presented with fecal occult blood (50%), colonic bleeding (68%) and diarrhea (42%). Rectal polyps were present in 98.7% of patients in this age group. For this reason rectoscopy, which is widely available, should be the first-line investigation in symptomatic patients younger than 40 years of age. Complete colonoscopy should be supplemented if symptoms persist without diagnostic findings during rectoscopy. Some patients presented with extracolonic manifestations such as desmoid tumors. They could be detected in 29% of patients between 20 and 40 years old. In two patients with desmoid tumors, FAP was diagnosed incidentally. In these cases especially, rectoscopy should be performed to exclude polyps. Generally, bleeding and stool disorders were found in cases with severe manifestations in the colon and rectum and colorectal carcinoma. There is an ongoing discussion concerning the benefit of colonoscopy vs. sigmoidoscopy for patients less than 40 years of age with non-acute rectal bleeding \[[@B19]\]. In terms of our findings, FAP in patients younger than 40 years old can certainly be diagnosed with rectoscopy or flexible sigmoidoscopy. As no rectal polyps could be detected in 24% of patients older than 40 years of age, and colon carcinoma was found in 50%, we propose colonoscopy in patients of this age who present with bleeding and stool disorders. Although fecal occult blood, colonic bleeding and diarrhea make a colorectal disease highly probable, they are not specific for FAP and are found in other gastrointestinal disorders such as inflammatory bowel disease, irritable bowel syndrome and malabsorption syndromes or infection, which must always be kept in mind \[[@B20]-[@B22]\].
Non-specific symptoms were less frequent and occurred in only 0.8--16% of our patients. They were associated with pronounced colorectal manifestations and many polyps of the upper gastrointestinal tract. Non-specific symptoms were commonest in patients over 40 years old. The incidence of colonic carcinoma was 42--50% in this age group. We conclude that non-specific symptoms can imply a pronounced manifestation of FAP which usually affects patients beyond 40 years of age \[[@B12]\]. Non-specific symptoms are present in other malignancies, endocrine or neuromuscular disorders as well \[[@B23]-[@B25]\].
In summary, we found no symptoms specific for FAP in our patient population. Yet it was possible to correlate several symptoms with disease in our age groups. In a few cases, extracolonic manifestations were the first noticeable changes. Due to their young age and mostly moderate symptoms, patients younger than 40 years of age with FAP are not easily identifiable in routine gastrointestinal practice. In all patients who present with the gastrointestinal symptoms we describe, the first diagnostic step should be rectoscopy, followed by complete colonoscopy to exclude FAP. Patients over 40 years old in whom inherited colorectal cancer is suspected, even without rectal manifestations, should generally undergo colonoscopy earlier than recommended for the general population. In patients of all age groups genetic testing should be recommended in order to allow presymptomatic testing in their sibs and offspring. Family members with positive tests and all persons at risk from families where the mutation could not be identified must be examined and monitored by clinical registers.
Competing interests
===================
The author(s) declare that they have no competing interests
Authors\' contributions
=======================
RC had the idea for the manuscript, carried out the statistical analysis, collected the details regarding molecular analysis and wrote the manuscript. WB gave information of several patients concerning molecular diagnostics and symptoms and has given final approval of the version to be published. BR checked the manuscript several times for grammer and spelling failures and was involved in revising it critically for important intellectual content. WH is the director of the Department of Surgery at the University of Erlangen where all patients included in the study were treated. He made the data collection of all patients possible. KG selected the patients and collected the symptoms of each patient. He has made substantial contributions to conception and design of the study. All authors read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2407/5/24/prepub>
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
„Non-specific symptoms" on primary diagnosis of FAP (n = 143 patients, realtive values regarding the corresponding subgroup of age)
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
„Bleeding and stool disorders" on primary diagnosis of FAP (n = 143 patients, realtive values regarding the corresponding subgroup of age)
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
„Bleeding and stool disorders", „non-specific symptoms" and related colorectal manifestations on the primary diagnosis of FAP (n = 143, values in brackets are relative values)
:::
**colon** **rectum**
------------------------ ----------- ------------ ---- ------- ---- ------- ---- ------- ---- ------- ---- ------- ---- -------
**total patients** 7 (100) 90 (100) 36 (100) 14 (100) 25 (100) 96 (100) 17 (100)
**fecal occult blood** 3 (43) 81 (90) 22 (61) 7 (50) 14 (56) 61 (64) 14 (82)
**colonic bleeding** 2 (29) 72 (80) 17 (47) 4 (29) 12 (48) 56 (58) 13 (76)
**diarrhoea** 1 (14) 44 (49) 16 (44) 1 (7) 6 (24) 37 (39) 11 (65)
**anemia** 3 (43) 30 (33) 16 (44) 4 (29) 4 (16) 25 (26) 6 (35)
**abdomial pain** 2 (29) 19 (21) 11 (31) 1 (7) 6 (24) 14 (15) 3 (18)
**loss of strength** 2 (29) 16 (18) 11 (31) 2 (14) 2 (8) 14 (15) 4 (24)
**loss of weight** 1 (14) 14 (16) 6 (17) 1 (7) 3 (12) 11 (11) 5 (29)
**mucous discharge** 2 (29) 17 (19) 7 (19) 1 (7) 3 (12) 15 (16) 2 (12)
**bloating** 2 (29) 11 (12) 6 (17) 1 (7) 3 (12) 9 (9) 2 (12)
**constipation** \- \- 9 (10) 2 (6) 1 (7) 3 (12) 5 (5) 3 (18)
**paradox diarrhoea** \- \- 5 (6) 1 (3) \- \- 1 (4) 4 (4) 2 (12)
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
„Bleeding and stool disorders", „non-specific symptoms" and related extracolonic manifestations on primary diagnosis of FAP (n = 143, values in brackets are relative values)
:::
**desmoid** **gastric polyps** **duodenal polyps** **small bowel polyps**
------------------------ ------------- -------------------- --------------------- ------------------------ ---- ------- ---- ------- ---- ------- ---- -------
**total patients** 8 (100) 13 (100) 5 (100) 39 (100) 26 (100) 7 (100)
**fecal occult blood** 4 (50) 9 (69) 4 (80) 28 (72) 19 (73) 4 (57)
**colonic bleeding** 4 (50) 9 (69) 4 (80) 25 (64) 15 (58) 2 (29)
**diarrhoea** 3 (38) 4 (30) 1 (20) 16 (41) 12 (46) 3 (43)
**anemia** 2 (25) 1 (8) \- \- 10 (26) 9 (35) 4 (57)
**abdomial pain** \- \- \- \- \- \- 8 (21) 6 (23) 3 (43)
**loss of strength** 1 (13) \- \- \- \- 7 (18) 7 (27) 2 (29)
**loss of weight** \- \- \- \- \- \- 5 (12) 6 (23) 2 (29)
**mucous discharge** \- \- 5 (39) 1 (20) 7 (18) 5 (19) 1 (14)
**bloating** \- \- \- \- \- \- 4 (10) 3 (12) 1 (14)
**constipation** \- \- \- \- \- \- 1 (3) 1 (4) \- \-
**paradox diarrhoea** \- \- \- \- \- \- \- \- 1 (4) \- \-
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Distribution of age related mutations in the APC gene exons of patients which underwent molecular diagnostics (n = 42)
:::
**age; years**
------------- ---------------- --- --- ---- ------
**Exon 1**
**Exon 2**
**Exon 3** 1 1 2.3
**Exon 4** 2 1 3 7.1
**Exon 5** 2 3 1 6 14.3
**Exon 6** 2 2 4.8
**Exon 7**
**Exon 8** 1 1 2 4.8
**Exon 9**
**Exon 10** 1 1 2.3
**Exon 11**
**Exon 12**
**Exon 13** 3 3 7.1
**Exon 14** 2 3 1 6 14.3
**Exon 15** 5 7 6 18 42.9
:::
|
PubMed Central
|
2024-06-05T03:55:55.174057
|
2005-3-2
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079798/",
"journal": "BMC Cancer. 2005 Mar 2; 5:24",
"authors": [
{
"first": "Roland S",
"last": "Croner"
},
{
"first": "Wolfgang M",
"last": "Brueckl"
},
{
"first": "Bertram",
"last": "Reingruber"
},
{
"first": "Werner",
"last": "Hohenberger"
},
{
"first": "Klaus",
"last": "Guenther"
}
]
}
|
PMC1079799
|
Background
==========
The phosphatidylinositol 3-kinase (PI3K)-AKT signalling pathway is involved in the regulation of diverse cellular processes, including cell growth, survival and motility. Abnormal activation of this pathway is frequently observed in various cancer types, leading to aberrant cell cycle progression, altered adhesion and motility, inhibition of apoptosis and induction of angiogenesis \[[@B1]\]. It has been previously reported that genetic alterations involving various members along this signalling pathway could lead to its activation in cancer. These include mutation, allelic loss or promoter methylation of the negative regulator *PTEN*\[[@B2]\]; or alternatively, chromosomal amplification or over-expression of the positive regulators *PIK3CA*\[[@B3]-[@B5]\] and the various *AKT*kinases \[[@B6],[@B7]\]. Furthermore, changes in other related pathways that are commonly altered in cancer, such as those involved in growth factor stimulation via the G-protein-coupled receptors or through direct interaction with the activated form of small GTPase RAS, can also lead to PI3K-AKT pathway activation \[[@B8]\]. Activation of this pathway results in the phosphorylation of AKT at Thr-308/309 and Ser-473/474. These phosphorylated forms of AKT proteins have been detected by Western blot or immunohistochemistry in various cancer types, suggesting the frequent activation of PI3K-AKT pathway in the carcinogenic process \[[@B7],[@B9]\].
Although genetic changes along the PI3K-AKT pathway have been repeatedly documented in brain, ovarian, endometrial, breast, prostate and thyroid cancers \[[@B1],[@B2]\], reports on its mechanism of activation in gastric cancer are limited. Gastric cancer is the second most common cancer worldwide but its molecular basis of tumourigenesis is still poorly understood. Previous immunohistochemical study has demonstrated the presence of the phosphorylated form of AKT in 78% of gastric cancer \[[@B10]\], suggesting that activation of this pathway may also be common in gastric cancer. Though loss of heterozygosity (LOH) involving the *PTEN*locus has been demonstrated in 47% of gastric cancer in a recent study, mutation or promoter methylation was absent even in cases with LOH \[[@B11]\]. Thus data from this study could not support the two-hit inactivation of *PTEN*in gastric cancer, while the biological significance of *PTEN*haploinsufficiency remains controversial. Alternatively, amplification of *AKT1*has been reported in a single case of gastric cancer \[[@B12]\], and amplification of *PIK3CA*associated with elevated mRNA levels has been found in 36% of gastric cancer \[[@B11]\]. More recently, Samuels *et al.*screened a diverse spectrum of human cancers for mutation in 16 *PI3K*or *PI3K*-like genes and found a high frequency of somatic mutation in *PIK3CA*, which encodes the p110α catalytic subunit. Major screening in colorectal cancer (CRC) identified *PIK3CA*mutations in 74 out of 234 (32%) cases, while mutations were also noted in 3 out of 12 (25%) gastric cancers. Reported mutations were mostly of missense type, and clustered within 2 regions in the helical and kinase domains. Expression of a \"hot-spot\" mutant, H1047R, conferred a significant up-regulation of lipid kinase activity of PIK3CA, suggesting it as an activating mutation \[[@B13]\]. In this study, we have examined a series of 94 human gastric adenocarcinomas for *PIK3CA*mutation. We have also examined *PIK3CA*expression level by extracting data from a large-scale gene expression profiling study previously performed for these cases \[[@B14],[@B15]\]. Using SAM, genes with significant correlating expression with *PIK3CA*have also been identified.
Methods
=======
Patient samples preparation
---------------------------
DNA samples used for sequencing were prepared from frozen tumour and non-tumour gastric mucosae from 94 gastric cancer patients who underwent gastrectomy in the Department of Surgery, Queen Mary Hospital, The University of Hong Kong, as previously described \[[@B16]\]. Majority of the frozen samples (n = 81) showed tumour component of over 70%, whereas in 13 cases a lower proportion between 50 to 70% was accepted due to the tumours\' inherent diffuse infiltrative nature with entrapment of non-neoplastic components. Analysis for microsatellite instability (MSI), *BRAF*and *KRAS*mutation have been performed and reported previously \[[@B16]\]. RNA preparation and gene expression profiling using a cDNA microarray containing 44,500 cDNA clones, representing around 30,300 unique genes, has been performed and reported in 90 of these tumours in comparison to 22 non-tumour gastric mucosae \[[@B14],[@B15]\]. This study was approved by the Ethics Committee of the University of Hong Kong.
Mutational screening
--------------------
Mutation screening of *PIK3CA*was performed for exons 9 and 20, covering the mutational hotspots; and for exon 18, from which a mutation was found in a gastric cancer. Mutations in these 3 exons constituted 80% of all *PIK3CA*mutations detected in the previous study \[[@B13]\]. *PIK3CA*intron-specific external amplification primers and internal sequencing primers were designed according to the previous study \[[@B13]\] with some modifications \[see [Additional file 1](#S1){ref-type="supplementary-material"}\]. In particular, primers for exon 9 have been modified to avoid amplification of homologous sequences located in other chromosomes. PCR products were generated using the external primers and directly sequenced using the internal primers with the DYEnamic™ ET Terminator Cycle Sequencing Kit (Amersham Biosciences, Freiburg, Germany) according to the manufacturer\'s instruction. Electrophoresis was performed in the ABI Prism^®^3700 DNA Analyzer (Applied Biosystems, Foster City, CA, USA). For each exon, PCR products were generated from 2 independent PCR reactions for sequencing of the forward and reverse strands. For exon 9, 2 independent PCR followed by sequencing of the forward strand were performed. Analysis of the chromatograms was performed using the mutation analysis software Mutation *Explorer*™ (SoftGenetics, State College, PA, USA).
Extraction of expression data and statistical analysis
------------------------------------------------------
Gene expression data were extracted from the microarray database containing 126 samples (90 gastric cancers, 14 lymph node metastasis and 22 non-tumour gastric mucosae) based on a 3-fold signal above background ratio for either channel and with 80% good data \[[@B14]\]. Gene expression data from 20,336 cDNA clones satisfied this selection criteria and were extracted, which included a cDNA clone corresponding to *PIK3CA*(IMAGE clone number 345430, GenBank accession no. W72473). Expression data for *PIK3CA*was extracted and the differences in expression levels between tumour and non-tumour tissues were examined using the Student\'s *t*-Test. SAM was performed to identify genes with significant correlating expression with *PIK3CA*\[[@B17]\]. The missing values in the dataset were estimated by a K-nearest neighbours impute algorithm using 10 nearest neighbour \[[@B18]\] followed by 5000 permutations in the SAM analysis.
Results
=======
Among the 94 gastric adenocarcinoma analysed, we have detected *PIK3CA*mutation in 4 cases. Two cases harboured the mutation A3140G (H1047R) in exon 20, and the other 2 cases with mutations G1624A (E542K) and G1633A (E545K) in exon 9. Representative sequence chromatograms are shown in figure [1](#F1){ref-type="fig"}. All four mutations were absent in the corresponding non-neoplastic mucosae and thus were confirmed as somatic mutations. Though the overall mutation frequency (4.3%) was lower than that of the previous study, the nature of the 4 mutations found were consistent with those identified at the reported hotspots. In particular, the H1047R mutation has been reported in 2 gastric cancers and 15 colorectal cancers \[[@B13]\]. While the E542K and the E545K mutations were not found in gastric cancer in the previous series, a large number of colorectal tumours did harbour these 2 mutations.
*PIK3CA*mutation spectrum and their corresponding clinico-pathological features were listed in Table [1](#T1){ref-type="table"}. We noted a higher tendency of high-level MSI in gastric cancers with *PIK3CA*mutations (3 in 4, 75%) than in those without (18 in 90, 20%). Moreover, though the overall incidence of *KRAS*mutation in the studied population was low (8 in 94), 2 of the 4 gastric cancers with *PIK3CA*mutation also harboured a *KRAS*mutation.
Since over-expression of *PIK3CA*has been reported in gastric cancer \[[@B11]\], we have also extracted *PIK3CA*expression data from our previous cDNA microarray study of these cases \[[@B14],[@B15]\]. We have confirmed that expression level of *PIK3CA*was significantly higher in gastric cancers (n = 87, mean = 0.099, SD = 0.428) when compared with non-neoplastic gastric mucosae (n = 22, mean = -0.418, SD = 0.426; Student\'s *t*-Test, *p*\< 0.001). Using *PIK3CA*expression level as a continuous variable for SAM analysis \[[@B17]\], we found 2910 cDNA clones (corresponding to about 2546 unique genes) whose expression associated positively with *PIK3CA*expression (median number of false significant = 0.372, Delta = 1.107) \[see [Additional file 2](#S2){ref-type="supplementary-material"}\]. Interestingly, no gene was found to be negatively associated with *PIK3CA*expression.
Discussion
==========
In this study, we have reported the presence of *PIK3CA*gene mutation in 4.3% of gastric cancer. A high tendency (3 in 4) of mismatch repair deficiency was noted in cases harbouring *PIK3CA*mutation. Though the small number of *PIK3CA*mutations in our study may not justify statistical claim of significance; suggestion of such, despite of its not being mentioned by the authors, can be found from a previous study in CRC by Samuels *et al.*. From their study of 33 MSI and 201 microsatellite stable (MSS) CRC cases, *PIK3CA*mutation was present in 48% of the MSI tumours, but only in 29% of the MSS tumours. A significant association would have been revealed if statistical analysis had been applied (Fisher\'s exact test, *p*= 0.014) \[[@B13]\]. Gastrointestinal tract cancers with MSI are known to have a different molecular pathway of tumour evolution compared with their MSS counterparts \[[@B19],[@B20]\]. This can be attributed to their propensity for frameshift mutations in repeat sequences, resulting in selective disruption of genes with such sequences within their coding regions. With 2 poly-adenine tracts within its coding region, *PTEN*can be inactivated through frameshift mutations in MSI CRC, resulting in the selective targeting of the PI3K-AKT signalling pathway \[[@B21],[@B22]\]. It is also known that mismatch repair deficiency would lead to an elevated rate of missense mutation due to impaired single nucleotide mismatch repair \[[@B23]\]. Thus, the observed higher incidence of *PIK3CA*missense mutation in MSI colorectal and gastric cancers suggests yet another mechanism for the activation of the PI3K-AKT signalling pathway through mismatch repair deficiency.
Our data also showed a higher tendency of *KRAS*mutation in cases with *PIK3CA*mutations (2 in 4) than in those without (6 in 90). Yet again due to the low incidence of both mutations in our samples, statistical significance may not be claimed. In the study by Samuels *et al.*, some of the colorectal tumours with *PIK3CA*mutation also harboured *KRAS*or *BRAF*mutation \[[@B13]\]. The PI3K-AKT pathway is known to have a close association with the RAS-MEKK signalling pathway \[[@B8]\]. Constitutively active RAS can interact with the catalytic subunit of PI3K and lead to its activation. Ras-dependent PI3K activation contributes to the transforming phenotype by mediating anchorage-independent growth, cytoskeletal reorganisation and apoptosis evasion. It has been observed that genes involved in the same signalling pathway may manifest mutations in cancer cells in a mutually exclusive manner, presumably due to the lack of selective growth advantage in having a second hit in the already altered pathway. A prominent example is the mutually exclusive occurrence of the *BRAF*hotspot mutation (V600E) and *KRAS*mutations in colorectal cancer \[[@B24],[@B25]\]. However, there exist other examples of alterations in multiple components of the same signalling pathway that may lead to a multi-level modulation of its activity. For example, non-V600E *BRAF*mutations tend to occur together with *KRAS*mutations \[[@B26]\], and inactivation of the secreted frizzled-related proteins (antagonists of WNT) by promoter methylation frequently coincides with mutations in the Adenomatous Polyposis Coli gene to achieve multi-level activation of the WNT signalling pathway in colorectal cancers \[[@B27]\]. Whether PIK3CA functions independently from RAS, or acts synergistically with RAS to produce additive effects on the activation of the same pathway awaits further clarification.
By extracting data from microarray, we have confirmed the up-regulation of *PIK3CA*expression in gastric cancer tissues compared with the non-neoplastic gastric mucosae and identified a large number of genes that showed a significant positive correlation in expression level with *PIK3CA*. These genes participate in diverse cellular processes with 177 as putative cell cycle-regulated genes \[[@B28]\] and 126 mapped to genes with known functions in cell cycle regulation, cell proliferation or DNA replication \[see [Additional file 2](#S2){ref-type="supplementary-material"}\]. While some of these genes maybe induced by *PIK3CA*, others maybe co-ordinately regulated by common upstream signals. Expression data set at one point was limited in differentiating the above cause and consequence, yet it certainly revealed the complexity of the carcinogenic process and the intricate relationship of *PIK3CA*signalling with other cellular processes.
Contrary to our expectation, the incidence of *PIK3CA*mutation found in the current study (4%) is much lower compared with that observed by *Samuel et al.*(25%) \[[@B13]\]. The reason for discrepancy may simply be a result of sample bias as the previous study involved only a small number of gastric cancers (n = 12). However, ethnic differences can also be another possibility. The diverse pathological spectrum and aetiological factors of gastric cancers in different geographical locations may be paralleled by differences in molecular pathway of tumour development. Since our current study is only based on a Chinese population with an intermediate gastric cancer incidence, further studies involving patients from different ethnic groups will be able to address this possibility.
Conclusion
==========
Large-scale screening of gastric adenocarcinomas for *PIK3CA*mutations revealed a mutation incidence of 4.3%. Increased *PIK3CA*expression level was observed in gastric tumours compared with non-neoplastic mucosae. This increase in *PIK3CA*level was associated with the elevated expression of a large number of genes, which may constitute the upstream regulators or downstream targets of *PIK3CA*along the PI3K signalling pathway.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
VSWL carried out the molecular analysis, performed data analysis and drafted the manuscript. CWW, TLC, WZ assisted in the molecular analysis. KMC provided the clinical data. ASWC assisted in data analysis and edited the manuscript. SS and XC participated in the microarray study and data analysis. STY and SYL conceived of the study, participated in its design, coordination and data analysis, and edited the manuscript. All authors read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2407/5/29/prepub>
Supplementary Material
======================
::: {.caption}
###### Additional File 1
Primers and PCR conditions used for *PIK3CA*sequencing. Conditions for PCR and sequences of primers used in *PIK3CA*sequencing
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 2
SAM significant genes list associated with *PIK3CA*expression. SAM detected 2910 genes whose expression levels were positively associated with the *PIK3CA*expression level. Gene expression data were extracted from the microarray database containing 126 samples (90 gastric adenocarcinomas, 14 lymph node metastasis and 22 non-tumour gastric mucosae) based on a 3-fold signal above background ratio for either channel and with 80% good data \[14\]. Gene expression data from 20,336 cDNA clones satisfied this selection criteria and were extracted, which included a cDNA clone corresponding to *PIK3CA*(IMAGE clone number 345430, GenBank accession no. W72473). Expression value of this clone was imputed as a quantitative variable for SAM analysis. The missing values in the dataset were estimated by a K-nearest neighbours impute algorithm using 10 nearest neighbour followed by 5000 permutations in the SAM analysis.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
This work was supported by the Research Grants Council of the Hong Kong Special Administrative Region, HKU 7313/02M.
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Representative sequence chromatogram.**(a) Sequence chromatogram from a case of gastric cancer with H1047R (A3140G) mutation within exon 20 of *PIK3CA*. Site of mutation is denoted by the red arrow. (b) Sequence chromatogram of the corresponding normal mucosa from the same case showing the absence of mutation (red arrow), and thus confirming the somatic nature of the mutation.
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Spectrum of *PIK3CA*mutations in gastric adenocarcinoma
:::
Case Nucleotide Substitution^a^ Amino acid change Age Sex MSI Status^b,c^ *KRAS*Mutation^c^ Tumour site Tumour type IM^d^
------ ---------------------------- ------------------- ----- ----- ----------------- ------------------- ------------- ------------- -------
28 A3140G H1047R 37 M Stable \- Body Intestinal 0
100 G1633A E545K 79 F High \- Antrum Intestinal 1
240 G1624A E542K 74 M High G12D Antrum Intestinal 1
310 A3140G H1047R 72 F High G13D Antrum Intestinal 1
^a^Nucleotide change at the position within coding sequence, where position 1 corresponds to the first position of the start codon
^b^High, high level of MSI; Stable, microsatellite stable
^c^Analysis of *KRAS*mutations have been performed and reported previously \[16\]
^d^IM, the presence of intestinal metaplasia at tumour edge; 1, present; 0, absent
:::
|
PubMed Central
|
2024-06-05T03:55:55.176627
|
2005-3-23
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079799/",
"journal": "BMC Cancer. 2005 Mar 23; 5:29",
"authors": [
{
"first": "Vivian Sze Wing",
"last": "Li"
},
{
"first": "Chi Wai",
"last": "Wong"
},
{
"first": "Tsun Leung",
"last": "Chan"
},
{
"first": "Agnes Sze Wah",
"last": "Chan"
},
{
"first": "Wei",
"last": "Zhao"
},
{
"first": "Kent-Man",
"last": "Chu"
},
{
"first": "Samuel",
"last": "So"
},
{
"first": "Xin",
"last": "Chen"
},
{
"first": "Siu Tsan",
"last": "Yuen"
},
{
"first": "Suet Yi",
"last": "Leung"
}
]
}
|
PMC1079800
|
Background
==========
Breast cancer is a very frequent malignancy among elderly patients. Its incidence increases with age, up to 80 years of age \[[@B1]\]. Furthermore, there are convincing data from the Surveillance, Epidemiology, and End Result (SEER) Program \[[@B2]\] showing that the proportion of elderly patients with breast cancer has been increasing during the last decades; patients aged 65 or more, indeed, accounted for 37% of breast cancer cases diagnosed in 1973 and for 47% of those diagnosed in 1995.
Nowadays, adjuvant chemotherapy for breast cancer is a standard clinical practice. This strategy has solid bases in the data of the Oxford Overview by the Early Breast Cancer Trialists\' Collaborative Group \[[@B3]\] that showed unequivocal benefit of adjuvant chemotherapy and suggested that its positive effect could be additive to that of tamoxifen. Thus, during recent years there has been a trend toward giving chemotherapy to the majority of women with early breast cancer, including the elderly ones if their risk of relapse is high and their clinical conditions are not deteriorated; thus, without considering elderly patients like a special subgroup for which specific recommendations are needed \[[@B4]-[@B6]\]. However, data collected in the Overview do not directly demonstrate the usefulness of adjuvant chemotherapy in older patients. Larger number of patients are needed, possibly enrolled into clinical trials specifically planned for elderly women, like the recently published study conducted by the French Adjuvant Study Group \[[@B7]\]. This study, indeed, evaluated the efficacy of the addition of weekly epirubicin to tamoxifen for patients older than 65 with node-positive early breast cancer, finding an advantage in terms of disease-free survival for patients receiving chemotherapy. Studies like this are needed to make the evidence stronger and to overcome selection biases possibly deriving from severe selection of elderly patients entered in clinical trials designed for adult ones. Indeed, it is well known that the proportion of elderly patients enrolled in clinical trials is far more lower than the proportion they represent in population statistics \[[@B8]\], barriers to their enrollment being multidimensional \[[@B9]\].
In clinical practice another problem due to the lack of reliable trials is the choice of which type of chemotherapy can be given to elderly patients. While it is now recognized \[[@B6]\] that two general levels of cytotoxic therapy regimens exist, one less aggressive that includes four courses of doxorubicin and cyclophosphamide or six courses of classical CMF (cyclophosphamide, methotrexate and fluorouracil), and another more aggressive including cyclophosphamide, epirubicin, and fluorouracil \[[@B10],[@B11]\] and experimental associations or sequences with taxanes \[[@B12],[@B13]\], many of these schemes have never been tested in prospective trials in elderly patients. With paucity of experimental data, CMF is felt as the more reasonable standard schedule for such patients \[[@B14]\].
While planning a randomized clinical trial, we reviewed the records of early breast cancer patients who received adjuvant CMF (days 1 and 8 every 4 weeks for a planned number of 6 cycles) at the National Cancer Institute of Naples, Italy, during the last decade, to describe compliance and toxicity in patients 65 or more years old; as a comparative group, we collected data of patients of the immediately lower age cohort (older than 60 and younger than 65).
Methods
=======
We selected charts of patients older than 60 (i.e.: those diagnosed breast cancer after the 60^th^birthday) who had received CMF from March 1991 to March 2002, outside clinical trials. All the patients had undergone mastectomy or quadrantectomy plus axillary lymphadenectomy. Patients undergoing conservative surgery were scheduled for radiotherapy of the residual breast with a standard scheme (5000 cGy plus 1000 cGy of boost on the quadrant) that was planned either concomitantly with adjuvant chemotherapy or immediately after the end of it, according to waiting lists or patient\'s or physician\'s preference. When radiotherapy was planned concomitantly, it was administered between the 3^rd^and the 4^th^cycle of CMF, in a sandwich scheme, producing a 6 weeks prolongation of the overall duration of CMF treatment.
In the period of interest, at our Institution, criteria on which therapeutic decisions regarding adjuvant chemotherapy were taken outside clinical trials varied according to the evolution of international guidelines and the results of the Oxford metanalysis. Overall, we recommended chemotherapy to patients aged less than 75, who had undergone surgical treatment (mastectomy or quadrantectomy, plus axillary lymphnode dissection), with a high risk of recurrence (based on common features like positive nodes, large primary tumor, high grade of malignancy, estrogen-receptor negative), good performance status (ECOG PS 0 or 1), normal function of vital organs (bone marrow, kidney, liver and heart) and no other severe comorbidity contraindicating chemotherapy. As older the patients and as lower the estimate of recurrence risk, the higher was the chance of choosing CMF rather than an anthracycline-containing regimen. Estrogen receptor positive patients, treated with chemotherapy, received tamoxifen after the end of chemotherapy.
For this analysis, we focused on patients aged over 60, who had received CMF according to the following schedule: cyclophosphamide 600 mg/m^2^, methotrexate 40 mg/m^2^and fluorouracil 600 mg/m^2^, on days 1 and 8, every 4 weeks, for 6 planned cycles. With this schedule, common supportive care was applied consisting of antiemetic medication with 5HT~3~antagonist on the days of chemo, antibiotics in case of known bacterial infection associated with granulocyte colony-stimulating factor (G-CSF) in case of febrile neutropenia; G-CSF was also sporadically used in case of grade 4 neutropenia without fever, and prophylactically in cycles following febrile or grade 4 neutropenia.
Pre-chemotherapy evaluation included medical history, physical examination, haematology, serum biochemistry tests, and tumor staging with chest radiography, abdominal ultrasonography, nuclidic bone scan with X-ray of bone hot-spots, ECG and, in selected cases, an evaluation of left ventricular ejection fraction. Toxicity was usually assessed with blood counts preceding each administration of chemotherapy, and complete serology before day 1 administration of chemotherapy, unless clinical need for more frequent control. Reviewing the charts, we codified toxicity according to National Cancer Institute Common Toxicity Criteria (version 2.0, 1998). Toxicity is reported with details of type and grade, including all grades. Anyway, for descriptive purposes we also summarised toxicity into three major groups: haematological (leukopenia, neutropenia, anemia, thrombocytopenia, febrile neutropenia, bleeding, infection), other (including all the other toxicities) and any (including both haematological and other) and we summed up grades 0 to 2 as non severe and grades 3 or 4 as severe.
To analyse compliance, the distribution of several indices was compared between the two age subgroups. A cycle was considered as delivered if day 1 chemotherapy had been administered. Indices used for compliance description were: a) the number of delivered cycles, b) the occurrence of withdrawal of chemotherapy on day 8, c) the occurrence and the percentage of dose-reductions, d) the occurrence of treatment delay that was expressed as relative duration by dividing actual and planned duration (1 means no delay, higher values mean progressively longer delays), e) the need for G-CSF treatment according to previously reported guidelines, and f) the cause of treatment discontinuation if earlier than planned. After description of single indices, a binary index was built, weak compliance being defined as the occurrence of at least one of the following events: less than 6 cycles administered, withdrawal of more than one day-8 treatment, dose reduction in more than 25% of cycles, relative duration higher than 1.25, need of G-CSF treatment, treatment interruption because of patient\'s refusal or toxicity. Because of possible confounding effect of radiotherapy associated to CMF, both toxicity and compliance were analysed not only by age subgroups but also according to presence or absence of concomitant radiotherapy.
Disease-free survival was defined as the time elapsed from surgery to the date of assessment of local or distant or contralateral breast cancer or the date of death for patients dying without disease recurrence. The Kaplan-Meier method was applied to draw disease-free survival curves.
Statistical significance of associations among variables was tested by the Fisher\'s exact test, or Wilcoxon-Mann Whitney test for naturally ordered variables (i.e. number of cycles, no. of day 8 omitted, grades of toxicity). Disease-free survival curves were compared by the Log-rank test. Analysis were done with S-PLUS (6.0 Professional Release 1, Insightful Corporation) and StatXact-5 (release 5.0.3, Cytel software Corporation).
Results
=======
From March 1991 to March 2002, 180 patients were identified, 80 aged 65 or older, and 100 older than 60 and younger than 65. The baseline characteristics of patients are reported in table [1](#T1){ref-type="table"}. Overall, 87 patients received radiotherapy on the residual breast; among these, 21 patients, who received radiation therapy after the end of CMF, have been considered in the no radiotherapy subgroup for all subsequent analyses. Sixty-six patients receiving concomitant radiotherapy were equally distributed in the two age subgroups, 36 (36%) being younger than 65 and 30 (38%) being older.
Treatment toxicity scattered by age subgroups is reported in table [2](#T2){ref-type="table"}. There was no toxic death. The only type of toxicity that significantly differed between the two age groups was febrile neutropenia that occurred in 3 cases among older patients and never in the younger group (p = 0.05). Among toxicities that did not show a statistically significant difference between the two groups, leukopenia, neutropenia, nausea, cardiac toxicity and thrombophlebitis tended to be more frequent or severe among elderlies, while mucositis tended to be more evident among younger patients. As shown in table [3](#T3){ref-type="table"}, haematologic toxicity was adversely affected by concomitant radiotherapy, leukopenia, neutropenia, febrile neutropenia and fever being significantly worse among patients who received concomitant radiotherapy. As reported in table [4](#T4){ref-type="table"}, incidence of grade 3--4 toxicity (either haematological or other or any type) did not vary according to age, while severe haematological toxicity was more frequent among patients who received concomitant radiotherapy (35% vs 8%, p \< 0.0001); almost one half (47%) of the older patients receiving concomitant radiotherapy experienced grade 3--4 hematological toxicity.
Focusing on compliance, we found that CMF scheduling was modified (from days 1 and 8 every 4 weeks to day 1 every 3 weeks) because of toxicity in 7 patients (2 in the older and 5 in the younger group, p = 0.46); none of these patients did receive concomitant radiotherapy (p = 0.05). Distribution of compliance indices is summarized in table [5](#T5){ref-type="table"}. Seventy-six percent of older patients received six cycles of chemotherapy as compared with 81% of those in the younger group. Day 8 of chemotherapy was withdrawn at least once in 36% and 31%, and 18% and 14% of the patients had more than 25% of cycle at reduced doses, in the older and younger groups, respectively. Treatment was administered without significant delay (i.e. 24 weeks without concomitant radiotherapy or 30 weeks with) in almost all the patients. Haematopoietic support with G-CSF was required by 18% and 9% of older and younger patients. Causes of treatment discontinuation were similar in the two groups, protocol completion accounting for more than three-fourths, and refusal and toxicity for about 10% each (p = 0.39). Eighty-eight percent of the 66 patients who received concomitant radiotherapy were able to receive 6 cycles of CMF and completed treatment according to the protocol, as compared with 74% of those not irradiated concomitantly; concomitant irradiation prolonged significantly duration of CMF in 5% of patients and produced a more frequent need of G-CSF. Overall, a weak compliance (i.e. the occurrence of at least one of the following events: less than 6 cycles received, more than 1 day 8 omission, dose reduction in more than 25% of cycles, relative duration higher than 1.25, use of G-CSF or treatment interruption because of toxicity or refusal) was more frequent among older patients (58% vs 46%, p = 0.02) and did not vary according to concomitant radiotherapy.
As of March 2004, 41 patients had an event and disease-free survival at 5 and 8 years was 76% and 71%. As shown in figure [1](#F1){ref-type="fig"}, there was no statistically significant difference in disease-free survival in the two age groups (p = 0.84).
Discussion
==========
The present study was planned being aware that a selection bias, typical of retrospective data collection, could affect the results; namely, outside clinical trials, the therapeutic strategy applied in our Institution during the time period of interest for this study was prevalently conservative, foreseeing adjuvant chemotherapy in elderly patients when either a very high risk of relapse or a very strong patient\'s motivation existed, no severe or untreatable comorbid conditions being evident. Overall, we anticipated that, under such conditions and because of selection bias, our study should produce underestimation of toxicity and lack of compliance, this effect possibly being as more pronounced as older the patients because of the application of more restrictive criteria.
We selected for this study a consecutive series of 180 early breast cancer patients aged 60 to 73 that received, outside clinical trials, adjuvant CMF, with all drugs intravenously on days 1 and 8 every 4 weeks for 6 cycles; this schedule resulted as the most commonly used regimen in a survey circulated among the members of the Breast International Group \[[@B14]\]. Overall, 18% of the patients suffered grade 3--4 toxicity, and 47% had some problem of compliance, either because dosing and scheduling rules could not be followed appropriately, or because costly supportive drugs (G-CSF) were required. Both these figures are much higher than what we expected. In addition, problems were more frequent among the older subgroup (for compliance) and when radiotherapy was given concomitantly with CMF (for haematological toxicity). The sense of such finding is even strengthened by the fact that very old patients, i.e. those over 75, were not considered for treatment. Particularly, concomitant radiotherapy significantly prolonged duration of chemotherapy over the planned time in 5% of patients, haematological toxicity being the most reasonable cause, as suggested by a more frequent need of G-CSF. Paradoxically, because of more prolonged time and less intensive administration, the number of CMF cycles fared better among patients who received concomitant radiotherapy.
In a retrospective analysis of adjuvant CMF for elderly breast cancer patients enrolled in the International Breast Cancer Study Group (IBCSG) trial VII \[[@B15]\], 76 women 65 years of age or older, compared with 223 postmenopausal women younger than 65, showed higher grades of severe toxicity either considering any type (17% v 7%, respectively), or haematologic (9% v 5%, respectively). In addition, only 48% of the older patients as compared with 65% of the younger ones received at least 85% of the planned dose. Our findings in terms of toxicity and compliance are overall worse than in the IBCSG trial VII. The most reasonable explanation for this is the duration of CMF, that was planned for 3 cycles in the IBCSG trial and for 6 cycles in our routine practice; also, we cannot exclude effects of the type of schedule (all drugs intravenously versus cyclophosphamide orally on days 1--14) and the effect of the timing of radiotherapy, given at the end of chemotherapy in the IBCSG trial VII, while it was concomitant in 37% of the patients in our study. In addition, we hypothesize that selection bias, arising from typical exclusion criteria of randomized clinical trials, could also have lowered toxicity rate in the IBCSG trial VII. Anyway, the conclusion of the Authors that less toxic chemotherapy regimens are required for high-risk elderly patients is completely supported by our findings.
For elderly patients with breast cancer, adjuvant chemotherapy should be considered both when the tumor does not express hormonal receptors, because tamoxifen would be ineffective, and in many cases in which the tumor is receptor-positive, because several indirect, but reliable, evidences suggest that positive effects of chemotherapy and tamoxifen are not mutually exclusive but can sum up, at a given extent. Based on current knowledge, the problem could be structured saying that the choice of not giving adjuvant chemotherapy must be supported by valid reasons, inside the decision making process. Factors that can affect such decision are risk of relapse, life expectancy and patient\'s preferences. As for risk of relapse, several subsequent editions of the St.Gallen guidelines \[[@B4],[@B6],[@B16]\], widely accepted as a reasonable approach to the treatment of early breast cancer, rule out that patients with ER/PgR-negative can be considered at minimal risk. In addition, within endocrine responsive tumors, the minimal risk category is limited to a quite small group of patients. A large amount of elderly patients fall within prognostic categories at average or high risk of relapse. Extermann et al. \[[@B17]\] faced the problem of coupling the risk of cancer relapse or death with life expectancy in order to define a utility measure that can assist decision making. Based on current estimates of efficacy of adjuvant chemotherapy, they found that a risk of 12% at 10 years is the threshold value to produce an absolute 1% survival gain for a 75 years old patient, who has approximately 12 years of life expectancy, if not particularly sick; as for the risk of relapse, estimates show that older breast cancer patients can expect advantages from treatment fairly similar to that of younger patients, both for tamoxifen and for chemotherapy. Finally, patient\'s preferences are not an easy matter of study \[[@B18]\], and, particularly in the setting of adjuvant chemotherapy of elderly breast cancer patients, data are very poor. Some studies, mostly including patients younger than 70, have shown that age does not affect patient\'s preferences \[[@B19]\] and that most patients consider 6 months of adjuvant CMF chemotherapy worthwhile for relatively modest survival gains \[[@B20]\]. Furthermore, indirect data, regarding elderly cancer patients \[[@B21]\] with various cancer types, suggest that more than 70% of elderly patients are willing to receive effective chemotherapy even if toxicity is relevant; such data, however, should be confirmed in a particular clinical setting like that of adjuvant chemotherapy, where the treatment is given to reduce an invisible risk, rather than to cure a visible disease.
Anyway, while growing the scientific evidence that adjuvant chemotherapy should be considered for elderly patients with breast cancer at risk of relapse, several reports show that the use of adjuvant chemotherapy in clinical practice does dramatically decline with age. De Michele et al. showed that in a tertiary cancer centre the rate of women eligible for adjuvant chemotherapy according to international guidelines who were actually offered chemotherapy dramatically decreased from 77%, among those aged 61 to 69, to 23%, among those aged 70 ore more \[[@B22]\]. Mandelblatt et al., from the Outcomes and Preferences for Treatment in Older Women Nationwide Study (OPTIONS) project, reported that, within a cohort of 718 patients, treated at public hospitals in several US states between 1995 and 1997, only 7% of those aged 67 to 79 years received adjuvant chemotherapy \[[@B23]\]. Woodard et al., after adjusting for a number of possible confounders, found that the odds of not receiving adjuvant chemotherapy for patients older than 65 is 62 times higher than that of patients younger than 50, and, even if the analysis is limited to ER-negative cases (i.e. those not amenable to hormonal adjuvant treatment), an odds of 7 is calculated \[[@B24]\].
There can be a lot of reasons for the discrepancy between guidelines or clinical trials evidences and practical application. One we believe is very important, together with other Authors \[[@B25]\], is the paucity of clinical trials specifically addressing adjuvant chemotherapy of elderly patients, because it is conceivable that lack of reliable evidences does not encourage physicians to suggest treatments that are potentially toxic and dangerous \[[@B26]\]. The consequences of such phenomenon are relevant, considering that sensible differences in patients outcome can be produced by varying treatment approaches \[[@B27]\].
Conclusion
==========
In conclusion, the present data, together with other retrospective evidences, strongly highlight the need of clinical trials of adjuvant chemotherapy specific for elderly early breast cancer patients, possibly looking for drugs or schemes that have shown in previous feasibility studies a favourable profile of safety. In this light, in a randomised clinical trial open to early breast cancer patients aged 65 to 80 with an average/high risk of relapse according to St.Gallen criteria, we are actually comparing CMF with weekly docetaxel, a treatment scheme that has shown sufficient activity and excellent tolerability in a phase II study conducted in elderly advanced breast cancer patients \[[@B28]\].
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
EDM, FP and AdM conceived of the study, participated in its design and coordination and drafted the manuscript;
AG, CP, GA, VL, GL, FN, ER, GD\'A, IC, MR, BM, ME and MDM treated the patients, collected the clinical data useful for the analysis and revised the article critically for important intellectual content;
All authors read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2407/5/30/prepub>
Acknowledgements
================
The authors wish to thank Giuliana Canzanella, Fiorella Romano and Federika Crudele for data management, Alfonso Savio for software assistance, and Giovanni de Matteis for secretarial activities. The Clinical Trials Unit is partially supported by AIRC (Associazione Italiana per la Ricerca sul Cancro), CTPG (Clinical Trials Promoting Group) and GIOGer (Gruppo Italiano di Oncologia Geriatrica). Ermelinda De Maio and Massimo Di Maio are recipients of a FIRC (Fondazione Italiana per la Ricerca sul Cancro) fellowship.
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Kaplan-Meier estimated disease-free survival curves by age categories (solid line: \>60 and \<65 years; dashed line: ≥ 65 years; crosses indicate censoring)
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Characteristics of Patients (N = 180)
:::
Median age (range), yrs 64 (60--73)
--------------------------------- -------------
pT-category, n (%)
pT1 73 (41)
pT2 91 (51)
pT3 5 (3)
pT4 3 (2)
unknown 8 (4)
pN-category, n (%)
pN0 66 (37)
pN1 97 (54)
pN2 10 (6)
unknown 7 (4)
Receptor status, n (%)
ER or PgR positive 82 (46)
Negative 78 (43)
Unknown 20 (11)
Histotype, n (%)
Ductal 132 (73)
Lobular 33 (18)
Other 15 (8)
Histologic grade, n (%)
G1 5 (3)
G2 53 (29)
G3 96 (53)
Unknown 26 (14)
Concomitant radiotherapy, n (%)
No 114 (63)
Yes 66 (37)
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Worst degree of toxicity by age group (% of patients)
:::
Type of toxicity \>60 and \<65 years (n = 100) ≥ 65 years (n = 80) *P*\*
--------------------- ------------------------------- --------------------- ------- ---- --- ----- ---- ----- ----- --- --------
Anemia 81 10 8 0 1 86 6 4 4 0 *0.39*
Leukopenia 54 15 19 7 5 45 11 26 11 6 *0.13*
Neutropenia 73 10 8 5 4 68 9 7 10 6 *0.32*
Febrile Neutropenia 100 \- \- 0 0 96 \- \- 4 0 *0.05*
Infection 97 0 1 2 0 95 0 2.5 2.5 0 *0.50*
Fever 98 0 2 0 0 98 2 0 0 0 *0.83*
Thrombocytopenia 90 7 2 0 1 91 4 1 3 1 *0.83*
Bleeding 98 1 0 1 0 100 0 0 0 0 *0.21*
Nausea 81 8 6 5 0 69 11 12 8 0 *0.06*
Vomiting 71 5 11 13 0 69 9 16 6 0 *0.98*
Mucositis 84 5 8 3 0 92 1 5 1 0 *0.09*
Skin 99 1 0 0 0 98 0 2 0 0 *0.43*
Diarrhea 96 1 0 3 0 96 3 1 0 0 *0.91*
Constipation 99 0 0 1 0 98 1 1 0 0 *0.45*
Hepatic 88 3 7 2 0 90 6 4 0 0 *0.60*
Cardiac 100 0 0 0 0 96 0 2.5 1 0 *0.09*
Fatigue 96 1 3 0 0 95 0 2.5 2.5 0 *0.72*
Pulmonary 100 0 0 0 0 99 1 0 0 0 *0.27*
Abdominal pain 99 0 0 1 0 99 0 0 1 0 *0.88*
Chest pain 99 0 0 1 0 100 0 0 0 0 *0.37*
Heartburn 99 0 1 0 0 98 0 2 0 0 *0.44*
Headache 99 0 0 1 0 100 0 0 0 0 *0.38*
Confusion 99 1 0 0 0 100 0 0 0 0 *0.38*
Depression 100 0 0 0 0 99 0 0 1 0 *0.27*
Thrombophlebitis 97 0 0 3 0 94 0 0 6 0 *0.30*
Dysuria 99 0 0 1 0 99 0 0 1 0 *0.88*
Allergy 99 0 0 1 0 100 0 0 0 0 *0.37*
Weight gain 100 0 0 0 0 99 0 1 0 0 *0.27*
Weight loss 99 0 1 0 0 99 1 0 0 0 *0.89*
\* Wilcoxon-Mann Whitney test
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Worst degree of hematological toxicity by radiotherapy (% of patients)
:::
Type of toxicity No (n = 114) Yes (n = 66) *P*\*
--------------------- -------------- -------------- ------- --- --- ---- ----- ----- ----- ----- ------------
Anemia 87 6 4 3 0 77 12 9 0 1.5 *0.12*
Leukopenia 58 15 22 3 3 36 11 23 20 11 *0.0001*
Neutropenia 82 8 6 2 2 50 12 11 17 11 *\<0.0001*
Febrile Neutropenia 100 0 0 0 0 95 0 0 5 0 *0.05*
Infection 96 0 2 2 0 95 0 2 3 0 *0.89*
Fever 100 0 0 0 0 94 6 0 0 0 *0.02*
Thrombocytopenia 90 6 2 1 1 91 4.5 1.5 1.5 1.5 *0.96*
Bleeding 99 0 0 1 0 98 2 0 0 0 *1.00*
\* Wilcoxon-Mann Whitney test
:::
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Summary of toxicity by age, radiotherapy and combined subgroups expressed as row percentages)
:::
**Haematologic** **Other** **Any**
---------------------------------- ------------------ ----------- ---------- ----- ----- ------ ----- ----- ------
**Age** 0.17 0.49 1.00
\>60 and \<65 years (*n = 100*) 86% 14% 72% 28% 62% 38%
≥ 65 years (*n = 80*) 78% 22% 78% 22% 61% 39%
**Radiotherapy** \<0.0001 0.21 0.27
No (*n = 114*) 92% 8% 71% 29% 65% 35%
Yes (*n = 66*) 65% 35% 80% 20% 56% 44%
**Age and radiotherapy** \<0.0001 0.42 0.08
\>60 and \< 65, no RT (*n = 64*) 92% 8% 67% 33% 59% 41%
≥ 65, no RT (*n = 50*) 92% 8% 76% 24% 72% 28%
\>60 and \< 65, RT (*n = 36*) 75% 25% 81% 19% 67% 33%
≥ 65, RT (*n = 30*) 53% 47% 80% 20% 43% 57%
G = grade; RT = radiotherapy.
:::
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
Pattern of compliance to chemotherapy by age and radiotherapy
:::
Variable Age Radiotherapy
----------------------------------------- --------- -------------- ------ ----------- --------- ------
N. of cycles administered, no. (%) 0.42 0.01
6 cycles 81 (81) 61 (76) 84 (74) 58 (88)
5 cycles 5 (5) 4 (5) 4 (3) 5 (8)
4 cycles 4 (4) 4 (5) 6 (5) 2 (3)
3 cycles 5 (5) 6 (8) 10 (9) 1 (1)
2 cycles 4 (4) 4 (5) 8 (7) 0
1 cycle 1 (1) 1 (1) 2 (2) 0
N. of day 8 omitted, no. (%) 0.37 0.86
0 69 (69) 51 (64) 77 (68) 43 (65)
1 23 (23) 18 (22) 24 (21) 17 (26)
2 5 (5) 8 (10) 9 (8) 4 (6)
3 3 (3) 3 (4) 4 (3) 2 (3)
Rate of cycles at reduced dose, no. (%) 0.54 0.83
≤ 25% 86 (86) 66 (82) 97 (85) 55 (83)
\>25% 14 (14) 14 (18) 17 (15) 11 (17)
Actual/planned duration, no. (%) 0.59 0.05
≤ 1.25 99 (99) 78 (98) 114 (100) 63 (95)
\>1.25 1 (1) 2 (2) 0 3 (5)
G-CSF utilization, no. (%) 0.12 \<.0001
No 91 (91) 66 (82) 109 (96) 48 (73)
Yes 9 (9) 14 (18) 5 (4) 18 (27)
Treatment discontinuation, no. (%) 0.52 0.10
protocol completion 81 (81) 61 (76) 84 (74) 58 (88)
patient\'s refusal 6 (6) 9 (11) 12 (11) 3 (4)
treatment toxicity 11 (11) 7 (9) 13 (11) 5 (8)
other 2 (2) 3 (4) 5 (4) 0
Low compliance^†^, no. (%) 0.02 0.88
No 61 (61) 34 (42) 61 (54) 34 (52)
yes 39 (39) 46 (58) 53 (46) 32 (48)
\* including 21 patients who received radiotherapy after the end of chemotherapy
§ Fisher exact test or Wilcoxon-Mann Whitney test for naturally ordered variables)
† at least one of the following features: less than 6 cycles administered, withdrawal of more than one day-8 treatment, dose reduction in more than 25% of cycles, relative duration higher than 1.25, use of G-CSF treatment, treatment interruption because of patient\'s refusal or toxicity
:::
|
PubMed Central
|
2024-06-05T03:55:55.178286
|
2005-3-24
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079800/",
"journal": "BMC Cancer. 2005 Mar 24; 5:30",
"authors": [
{
"first": "Ermelinda",
"last": "De Maio"
},
{
"first": "Adriano",
"last": "Gravina"
},
{
"first": "Carmen",
"last": "Pacilio"
},
{
"first": "Gerardo",
"last": "Amabile"
},
{
"first": "Vincenzo",
"last": "Labonia"
},
{
"first": "Gabriella",
"last": "Landi"
},
{
"first": "Francesco",
"last": "Nuzzo"
},
{
"first": "Emanuela",
"last": "Rossi"
},
{
"first": "Giuseppe",
"last": "D'Aiuto"
},
{
"first": "Immacolata",
"last": "Capasso"
},
{
"first": "Massimo",
"last": "Rinaldo"
},
{
"first": "Brunello",
"last": "Morrica"
},
{
"first": "Massimo",
"last": "Elmo"
},
{
"first": "Massimo",
"last": "Di Maio"
},
{
"first": "Francesco",
"last": "Perrone"
},
{
"first": "Andrea",
"last": "de Matteis"
}
]
}
|
PMC1079801
|
Background
==========
Intraperitoneal (IP) administration of antineoplastic agents has been studied in many investigative trials and has been used in clinical practice for more than 40 years \[[@B1]\]. Serosal invasion occurring in spite of curative resection is an unfavourable prognostic factor in gastrointestinal malignancies \[[@B2]\]. IP chemotherapy is now viewed as an interesting approach to treating tumors with peritoneal, local lymphatic or hepatic dissemination \[[@B3]\] such as gastric or colon cancer. The main goal of IP administration is to increase the exposure of cancer cells in the peritoneal cavity to the drugs with minimal potential toxic effects on the internal organs \[[@B1],[@B4],[@B5]\]. Various chemotherapeutic agents such as mitomycin-C (MTC), cisplatin and 5-fluorouracil, administered intraperitoneally during the early postoperative period, have been shown to be safe and to improve prognosis in patients with gastrointestinal cancers \[[@B6],[@B7]\].
The timing as well as the route of administration of the chemotherapeutic agent plays a major role in determining the response. Since drug resistance is often related to the number of neoplastic cells present when the chemotherapeutic treatment is initiated, early administration of the agent in conjunction with surgery may give better results \[[@B8]\]. However, the effect of chemotherapeutic agents, delivered intravenously or intraperitoneally, on the healing of intestinal anastomosis is still controversial \[[@B9]-[@B16]\]. Weiber \[[@B14]\] investigated the influence of postoperative 5-fluorouracil on intestinal anastomotic wound healing in relation to timing of administration, but no other chemotherapeutic agents have been studied in this way. Our aim in the present study was to investigate the effects of IP MTC, administered at various times postoperatively, on intestinal anastomotic wound healing. MTC has been extensively used for treating gastric cancers.
Methods
=======
Eighty-four Wistar-Albino female rats weighing 275 ± 25 g were acclimatized for 10 days before the study. All the animals were in good health and were kept under the same physical and environmental conditions. They were fed with standard pellet diet and tap water and kept at 23 ± 2°C, with a 12 h light/dark cycle, throughout the study period. They received human care and the study protocol complied with the institutions\' guidelines. All the rats underwent abdominal surgery under ether anesthesia. Standardized laparotomy was performed through a 3-cm midline incision. A 2-cm ileal segment was resected up to 10 cm proximal to the ileocecal valve. An end-to-end anastomosis was made using interrupted 5/0 polypropylene sutures as a continuous single layer. The fascia was closed using interrupted 3/0 polypropylene sutures. Subsequently, the rats were randomly assigned to two groups: group A, MTC group(n = 65); control (saline) group (n = 13). The MTC group was sub-divided into 5 equal (n = 13) groups, A1-A5, and IP mitomycin-C was administered postoperatively as follows: day 0 (A1), day 3 (A2), day 5 (A3), day 7 (A4) and day 10 (A5). Each MTC dosage was 2 mg/kg with 10 ml saline. All the rats were sacrificed with an overdose of ether on the 14th postoperative day. Postmortem exploration of the abdomen to detect anastomotic complications was performed by an observer blinded to the treatment arm of the study. The entire anastomotic segment of ileum with 2-cm margins on either side was removed for study.
Anastomotic bursting pressure measurement
-----------------------------------------
Anastomotic bursting pressure was measured in all rats. A catheter was inserted intraluminally into the resected segment of the ileum and secured using pursestring sutures. A mercury manometer connected to an inflating apparatus was attached to the other side of the catheter and the open end of the ileum was ligated. The resected segment of the intestine was then totally submerged in a water pot and gently distended with air. The bursting pressure was recorded as mmHg at the time of appearance of the first gas bubble.
Hydroxyproline content determination
------------------------------------
Tissue hydroxyproline content was determined in all rats by the Reddy method \[[@B17]\] and the results were recorded as μg hydroxyproline per mg dry tissue weight.
Statistical analysis
--------------------
Non-parametric continuous variables were compared by the Kruskal-Wallis one-way analysis of variance with post-hoc analysis using a Mann-Whitney U test. Parametric variables were compared using one-way analysis of variance with post-hoc analysis using the Tukey test. Data are presented as medians and ranges for nonparametric variables and means ± SD for parametric variables. Differences were regarded as significant at p \< 0.05 for parametric and 0.05/6 for nonparametric analyses.
Results
=======
Postoperative mortality was observed in five animals: 2 (15.4%) in group A1, 2 (15.4%) in group A2 and 1 (7.7%) in group A3. The deaths in group A1 occurred on the 1st postoperative day; the other deaths occurred before administration of MTC at the 2nd postoperative day. Macroscopic examination during post-mortem laparotomy showed anastomotic leakage communicating with abdominal abscesses in all deaths. Because the deaths in group A2 an A3 occurred before MTC administration and those in group A1 occurred earlier, the anastomotic leakage was attributed to operative complication and the dead rats were excluded from the study. No mortality occurred in any of the remaining animals before the completion of the study. During post-mortem abdominal explorations at the end of the study, non-lethal anastomotic leakage was observed in 5 animals: 1 in group A1, 2 in group A2, 1 in group A5 and 1 in the control group.
Anastomotic bursting pressure
-----------------------------
Anastomotic bursting pressures (mmHg) were 86 (0--102), 96 (0--116), 149.5 (116--196), 175 (108--241), 193 (0--243) and 198 (0--231) in groups A1, A2, A3, A4, A5 and control, respectively (Fig. [1](#F1){ref-type="fig"}). The values in groups A1 and A2 were significantly lower than in groups A3, A4, A5 and control (P values were \<0.05 for all these comparisons). There was no statistically significant difference between groups A1 and A2 (p \> 0.05). The anastomotic bursting pressures in groups A3, A4 and A5 were not significantly different from those in the control group (P values were \>0.05/6 for these comparisons).
Tissue hydroxyproline content
-----------------------------
The mean tissue hydroxyproline contents (μg/mg) of groups A1, A2, A3, A4, A5 and control were 6.54 ± 1.9, 8.14 ± 2.3, 14.77 ± 3.4, 14.80 ± 4.3, 17.01 ± 3.9 and 18.28 ± 4.3, respectively (Fig. [2](#F2){ref-type="fig"}). The values for groups A3, A4 and A5 were comparable with the controls (P values were \>0.05 for these comparisons). The tissue hydroxyproline levels of groups A1 and A2 were significantly lower than the controls or the other MTC sub-groups (P values were \<0.05 for these comparisons).
Discussion
==========
Recurrence and IP dissemination of gastrointestinal cancers is common despite curative resection, particularly when the serosa has been infiltrated \[[@B2]\]. It has been suggested that IP administration of anticancer agents before closure of the laparotomy, and during the first few postoperative days, may prevent the dissemination of cancer cells \[[@B18]-[@B20]\]. A potential criticism of delayed postoperative IP administration is inhomogeneous drug distribution caused by early postoperative adhesions \[[@B21]\]. Entrapment of the tumor cells by fibrin in the dissected areas decreases their exposure to the chemotherapeutic agents \[[@B22],[@B23]\]. Thus, the use of IP chemotherapy at the time of surgery provides better drug distribution without the risk of interference from postoperative adhesions \[[@B21]\]. However, although the administration of chemotherapeutic agents in the immediate postoperative period may decrease the local cancer recurrence rate, this must be weighed against the possible impairment of wound healing \[[@B12],[@B24],[@B25]\].
In the present study, the animals that received IP MTC perioperatively and on the 3rd potoperative day had significantly lower mean tissue hydroxyproline contents and anastomotic bursting pressures than the controls. No difference was observed between the controls and the other MTC sub-groups. Collagen synthesis, which may be reflected by tissue hydroxyproline content, is an essential feature of anastomotic healing in the intestine \[[@B17],[@B26]\]. We therefore measured the tissue hydroxyproline content to determine whether MTC is detrimental to anastomotic wound healing. The data showed that IP MTC administered perioperatively or on the 3rd postoperative day was detrimental to the healing of intestinal anastomosis, but MTC administered on or after the 5th postoperative day had no significant effect on the process. The effects of IP administration of various chemotherapeutic agents on intestinal anastomotic wound healing have been investigated in many studies and the results have been conflicting \[[@B12]-[@B14],[@B16],[@B24],[@B27],[@B28]\]; some have shown detrimental effects \[[@B12],[@B28]\]. However, the effect on wound healing of peri- and post-operative intraperitoneal MTC administration has not been described previously.
Tissue injury initiates a cycle of inflammation, cellular migration and replication, and connective tissue deposition and remodeling, which restores tissue integrity. Successful wound healing depends on the formation of a strong and stable scar. Wound strength is determined by the amount and quality of newly-synthesized and deposited collagen, and by the degradation of preformed collagen \[[@B29]\]. Collagenase activity, which is important in determining anastomotic integrity and suture-holding capacity during the first few days of healing, increases significantly after 3 days of colonic anastomosis; at this time the suture holding capacity of the anastomosis decreases by up to 80% \[[@B30]-[@B33]\]. MTC arrests the proliferation of fibroblasts, which are responsible for several crucial aspects of the wound-healing process outlined above \[[@B34]\]. Thus, IP administeration of MTC within three days of the operation may be detrimental to intestinal anastomotic wound healing, as shown in the present study.
Ineffective colonic wound healing can have devastating consequences e.g. abscesses, fistula formation, or death secondary to overwhelming sepsis. Therefore, although early IP administration of MTC following cytoreductive surgery is effective in preventing the dissemination of cancer cells, the side effects discussed here have considerable importance and should not be overlooked.
In conclusion, certain interventions are needed to avoid the potentially lethal side effects of chemotherapy on intestinal wound healing in subjects receiving early IP administration. Additional therapeutic approaches such as anti-inflammatory or fibrinolytic agents should be introduced to reduce adhesion formation, and thus preclude inhomogeneous drug distribution, in patients for whom IP administration has been postponed beyond the early postoperative period to avoid the undesirable side effects.
Abbreviations
=============
IP, intraperitoneal; MTC, mitomycin-C.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
AU conceived the study and participated in its design and coordination. CB, MH and FFB conceived the study, participated in the sequence alignment and drafted the manuscript. AK collected the samples and carried out the laboratory analysis.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2407/5/31/prepub>
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Anastomotic bursting pressures in mitomycin-C sub-groups and controls. Group A1 and A2 which have received mitomycin-C (MTC) in preoperative (day 0) and 3rd day of postoperative period, respectively, had significantly lower value than the other groups (both, p \< 0.001).
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Tissue hydroxyproline levels in mitomycin-C sub-groups and controls. Group A1 and A2 which have received mitomycin-C in preoperative (day 0) and 3rd day of postoperative period, respectively, had significantly lower value than the other groups (both, p \< 0.001).
:::
:::
|
PubMed Central
|
2024-06-05T03:55:55.183774
|
2005-3-31
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079801/",
"journal": "BMC Cancer. 2005 Mar 31; 5:31",
"authors": [
{
"first": "Ali",
"last": "Uzunkoy"
},
{
"first": "Cengiz",
"last": "Bolukbas"
},
{
"first": "Mehmet",
"last": "Horoz"
},
{
"first": "Fusun F",
"last": "Bolukbas"
},
{
"first": "Abdurrahim",
"last": "Kocyigit"
}
]
}
|
PMC1079802
|
Background
==========
Many developing tissues are comprised of morphologically indistinguishable cells. However, these cells are often heterogeneous with respect to gene and protein expression, as well as developmental potential. Differences that develop from initially clonal cancer cell populations underlie the emergence of cells resistant to initial therapeutic intervention, and the ability of certain cancers to spread may relate in part to the intrinsic motility of cancerous cells \[[@B1]\]. Methods that facilitate the identification and isolation of cells exhibiting specific behaviors may lead to greater understanding of molecular mechanisms underlying cancer progression.
The identification of differences in gene and protein expression that contribute to carcinogenesis depends crucially on the specific identification and isolation of abnormal cells. Although recent advances in tissue microdissection enables highly specific isolation of cells from tissue samples \[[@B2]\], the ability to identify and isolate living cells based on specific behavioral characteristics may provide valuable insights that may not be evident from static morphological analysis of tissue \[[@B3],[@B4]\].
Although several methods to examine cell motility exist, most characterize motility on a cell population basis, cannot distinguish heterogeneity within a population, and do not permit isolation of cells with specific motility. Variations of a classic chemotactic assay initially described by Boyden \[[@B5]\] have been effectively used to characterize the motility of a variety of cell populations. These assays typically monitor the movement of cells to the opposite side of a porous membrane onto which they are initially plated in high numbers. Typically, Boyden/transwell assays reveal differences in motility of the most motile fraction of the entire populations analyzed, because the vast majority of cells do not pass through the transwell membrane. Finally, although these assays have proven to be quite versatile, they require large starting numbers of cells, and isolation of cells possessing distinct motility remains challenging.
It has been observed that cells moving on substrates coated with supra-colloidal gold particles generate a record of their movements by clearing the particles from their path \[[@B6]\]. This clearing of a particle-free trail by a combination of cell locomotion and phagocytosis, described as \"phagokinetics,\" has been used to quantify the motility of a variety of cell types \[[@B6]-[@B8]\]. Here, we describe a method that enables quantitation of motility both by direct measurement of cleared area and by fluorescent signal intensity within single cells, and permits isolation of cells based on their motility.
Results
=======
Migrating cells create non-fluorescent tracks on fluorescent microsphere-coated substrates
------------------------------------------------------------------------------------------
Non-cytotoxic fluorescent polystyrene microspheres have been utilized as cell labels \[[@B9]\], microinjectable cell tracers \[[@B10]\], retrograde neuronal markers \[[@B11]\], and phagocytosis indicators \[[@B12]\]. We have taken advantage of the phagokinetic ability of migrating cells \[[@B6]\] by allowing them to ingest fluorescently labeled polystyrene microspheres coated onto a variety of migratory substrates. Tissue culture vessels prepared by pre-treatment with poly-D-lysine were coated with 1 μm diameter green fluorescent microspheres. Cells were plated onto polylysine-treated tissue culture plastic, incubated for 18--24 hours, and then fixed with 4% paraformaldehyde.
Moving cells generated microsphere-free areas in the dense fluorescent particle coat that were easily visualized using fluorescence microscopy (Figure [1](#F1){ref-type="fig"}). Differences in the motility of two glioblastoma cell lines were readily apparent by the distinctions in area of the tracks cleared (Figure [1B](#F1){ref-type="fig"}). Utilizing beads of different fluorescent emission wavelength allowed simultaneous visualization of tracks and cells stained with distinct fluorescent markers (F-actin with Alexa-Fluor 546 Phalloidin, DNA with Hoescht 33342, or transfected with green fluorescent protein, Figure [1A](#F1){ref-type="fig"}). Confocal imaging of cells stained with Alexa-Fluor Phalloidin migrating on microspheres reveals that ingested microspheres do not interfere with the resolution of the actin cytoskeleton (Figure [1D](#F1){ref-type="fig"}).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Appearance of cells migrating on fluorescent microspheres**. (A) F-actin (red), DNA (blue), microspheres (green), and merged view indicate that cells clear non-fluorescent tracks in the dense particle field as they move. Bar, 100 μm. (B) Cell lines exhibit differences in motility reflected by the area of particles cleared, highlighted by comparing T98G and U118MG glioblastoma lines. Bar, 300 μm. (C) Cells transfected with expression vector for GFP can be visualized on a blue fluorescent microsphere field. The trails from both cells converge at a common origin, suggesting that the two cells arose from the division of a common progenitor and migrated away. Bar, 20 μm. (D) Confocal section of phalloidin-stained U118MG cell migrating on field of blue microspheres. Bar, 20 μm.
:::
:::
To establish a reference for comparing the motility of two glioblastoma cell lines, U118MG and T98G, we employed a commonly used transwell filter assay. Cells were seeded on the top of the membrane and allowed to migrate to a lower compartment containing media. In contrast with traditional chemotactic transwell assays, we did not use a chemoattractant gradient, but instead utilized the normal growth media on both sides of the membrane (identical media used in fluorescent phagokinetic assay above). Thus, instead of chemotaxis, this assay measures pure cell motility. We observed that more (1.33 fold greater) U118MG cells had migrated through the transwell filter compared with the T98G cell line (Figure [2A](#F2){ref-type="fig"}).
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Characteristics of fluorescent phagokinetic assay**. (A) Transwell cell motility assay. A greater number of U118MG cells (1.33 fold) transit through the filter compared with T98G. Graph depicts the mean number of cells that transit through the filter of 5 10X fields in 4 replicate wells/cell line (20 fields/cell line/experiment) from 3 independent experiments; p = 0.0185). (B) Mean area of fluorescent particles cleared per cell reveals U118MG cell line exhibits greater intrinsic motility than T98G (2.29 fold greater area cleared/cell). Graph depicts the mean motility of cells from 3 independent experiments; 100 cells/cell line measured for each experiment; p = 0.0012. (C) Fluorescent phagokinetic assay reveals differences in motility of cell lines (primary mouse cerebral cortical astrocytes, T98G, U118MG) on different extracellular matrices (fibronectin (FN), type IV collagen, laminin). \* indicates p \< 0.001 comparing motility of U118MG on FN vs. Collagen IV or FN vs. laminin. For all pairwise comparisons of U118MG on any substratum vs. either T98G or primary astrocytes, p \< 0.001. (D) Histogram of distribution of areas cleared by U118MG vs. T98G from a representative experiment on poly-D-lysine treated tissue culture plastic (no additional substratum). Results are expressed as means +/- SEM, and statistical significance was evaluated by Student\'s t-test (A, B)or one-way ANOVA followed by Newman-Keuls *post-hoc*test (C).
:::
:::
To determine whether the fluorescent phagokinetic assay provided a measure of cell motility comparable to that of the transwell assay, we measured the fluorescence-free areas cleared by single cells plated on fluorescent microspheres. By tracing and measuring cleared areas generated by cells, we obtained a direct measure of how far individual cells moved in the time since plating. The area cleared per cell was obtained for at least 100 cells/cell line/independent experiment. The fluorescent phagokinetic assay showed that cells from the U118MG cell line cleared a greater mean area (2.29 fold greater) compared to the T98G cell line, an observation confirming that found with the transwell motility assay. (Figure [2B](#F2){ref-type="fig"}). Although these results suggest that comparable relative differences in intrinsic motility between cell lines are observed with these different assays, the phagokinetic assay has the advantage that an unbiased sample of cell motilities is measured. Unlike the transwell assay, in which only a small fraction of plated cells traverse the membrane (those with highest intrinsic motility), the motility measured with the fluorescent phagokinetic assay is more representative of the entire population of cells because the motility of all of the plated cells is recorded on the plate.
Interactions with local host microenvironment plays a crucial role in cancer spread, in part by regulating cell motility \[[@B1]\]. For cell migration to occur, complex interaction between cells and the extracellular matrix regulate local adhesion and cytoskeletal rearrangements \[[@B13]\]. To determine the utility of our approach to assess cell motility on different extracellular matrices, extracellular matrix substrates (laminin, fibronectin, type IV collagen) were coated with fluorescent microspheres, and differences in the motility of a variety of cell types on these substrates were determined (Figure [2C](#F2){ref-type="fig"}). To examine whether the motility of primary cells could be examined with this technique, we isolated and plated primary mouse cerebral cortical astrocytes onto the coated substrates. We observed that U118MG cells remained consistently more motile than T98G, as well as primary astrocytes on all of the three substrates tested (p \< 0.001 for each pairwise comparison). The motility of U118MG cells on collagen IV was significantly greater than observed on either laminin or fibronectin (p \< 0.001). These studies confirm the applicability of analyzing population motility characteristics with the fluorescent phagokinetic assay with multiple cell types and extracellular substrates.
Fluorescence of individual cells is proportional to motility
------------------------------------------------------------
The ability of tumor cells to acquire greater malignancy over time is well established, and despite the monoclonal origin of most tumors, they are heterogenous at clinical presentation \[[@B14]\]. Using the fluorescent phagokinetic assay, individual cell characteristics can be measured with fluorescent microscopy, making subtle differences in motility readily apparent. By measuring cell motility on a single cell level, we found that the motility of individual cells within clonal cell lines was distributed widely (Figure [2D](#F2){ref-type="fig"} and data not shown). Although documentation of individual cell motility by microscopy is straightforward, analyzing large number of cells remains cumbersome, and the desire to isolate and recover cells with differing motilities led us to consider alternatives to manual microscopic analysis.
Because the area of the tracks generated by cells relates directly to the quantity of fluorescent microspheres consumed, we reasoned that cell fluorescence could serve as another indicator of motility. By quantifying total cell fluorescence of individual cells, we found that single cell fluorescence was directly related to the area of fluorescence cleared (Figure [3](#F3){ref-type="fig"} and data not shown; the differences in slope measured result from differences in gain settings between experiments to maximally utilize the dynamic range of the CCD camera). This relationship between fluorescence and motility provides an approach by which cells of different motility can be isolated by fluorescent cell sorting.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Relationship of fluorescence and motility**. After migration on fluorescent beads, single cell fluorescence of T98G (A) and U118MG cells (B) is linearly related to area of fluorescent microspheres cleared by the cell. Fluorescence intensity of individual migrating cells was obtained by tracing each cell outline and measuring total fluorescent signal. Area cleared by each cell was measured by outlining cleared area. Images were collected using a 10X phase objective, 0.3 NA, and Endow GFP filter cube, and captured with a 16 bit CCD camera (Cascade 650, Roper Inc.).
:::
:::
Cells with intrinsic motility differences can be sorted and enriched by fluorescence
------------------------------------------------------------------------------------
To isolate cells based on fluorescence, U118MG cells were plated on microsphere-coated culture dishes, allowed to migrate for 20 hours, and sorted (Beckman Coulter Epics Elite ESP Cell Sorter) based on fluorescence intensity and side scatter characteristics. As might be expected, fluorescence (phagocytosed spheres) correlated with side scatter \[cell granularity (resulting from the microspheres)\] (Figure [4A](#F4){ref-type="fig"}). Fluorescence intensity is not correlated with cell size (Figure [4B](#F4){ref-type="fig"} forward scatter (cell size) vs. fluorescence), and we have not found any morphological differences between high motility and low motility cells. After sorting the cells with gates arbitrarily chosen to select for approximately the top and bottom thirds of the total population by fluorescence, we recovered and replated high fluorescence and low fluorescence U118MG cells (Figure [4D](#F4){ref-type="fig"}). A small number of sorted cells were recharacterized by flow cytometry to verify the effectiveness of the sort, and as expected from the initial sort criteria used, fluorescence did not correlate with cell size \[forward scatter (cell size) vs. fluorescence; Figure [4C](#F4){ref-type="fig"}\]. When motility of post-sorted populations was determined with the fluorescent phagokinetic assay after one to five weeks in culture post-sorting, the sorted high fluorescence cells retained significantly higher motility than the low fluorescence cells (p = 0.0230; Figure [4E](#F4){ref-type="fig"}). These observations suggest the possibility of intrinsic and long-term heterogeneity within cell lines despite their original clonal origins.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Flow-cytometric sorting of cells based on motility**. (A) Fluorescence distribution of U118MG resembles distribution of cell motility seen in phagokinetic motility assay (left). The y-axis represents the number of cells characterized in each bin; the x-axis represents cell fluorescence at 525 nm. An untransformed linear scale for fluorescence intensity is used. On a linear scale, the voltage measured (signal intensity) is directly proportional to the channel into which the event falls. A cell with a linear value of 100 is 10 times brighter than one in with a linear value of 10. Side-scatter and fluorescence characterization of U118MG after 20 hours of migration on fluorescent microspheres (middle). Cells were sorted using the gates drawn. Granularity (side scatter) reflects the quantity of ingested beads, and is directly related to cell fluorescence. 1023 channels are available to bin the signal from fluorescence, forward, and side scatter. Forward scatter (a measure of cell size) and side scatter (a measure of cell granularity) characterization of U118MG indicates that there is no relationship between cell size and bead internalization (right). (B) Cell size is not related to cell fluorescence. Forward scatter vs. fluorescent signal of analyzed cells show no relationship between cell size and fluorescence (left). (C) Flow characterization of sorted cells. A small sample of recovered cells were re-analyzed after sorting with gates shown in (A), and both gated populations (low and high fluorescence) show no relationship between cell size (forward scatter) and cell fluorescence. (D) Images of low fluorescence cells (left) and high fluorescence cells (right) 24 hours post sorting, plated onto tissue culture plastic treated with 50 μM poly-D-lysine. Bar = 20 μm. (E) Cells retain differences in motility when reassayed after one, two, three, and five weeks post-sort (n \>100 cells/cell line at each time point). Cell motility was assayed by measuring the area cleared/cell over 24 hours using the fluorescent phagokinetic assay. The overall mean fold difference between the two cell lines is 1.20, p = 0.0230. The mean motility of the high motility cell line was significantly different than the low motility cell line at each time point measured (week 1, 1.23 fold difference, p = 0.0106; week 2, 1.23 fold difference, p = 0.0105; week 3, 1.17 fold difference, p = 0.0195; week 5, 1.18 fold difference, p = 0.0201). Results are expressed as means +/- SEM, and statistical significance was evaluated by Student\'s t-test at each time point.
:::
:::
Discussion
==========
The utility of using fluorescence to sort motile cells is dependent on the close relationship of bead accumulation and motility, and a number of factors may potentially confound interpretation of the described approach. Our assay is likely affected by similar factors that limit the original colloidal gold phagokinetic motility assay. Similar to the observations of multiple studies using colloidal gold particles \[[@B15]\], we did not observe any toxicity or ill consequences from phagocytosis of beads. Cell proliferation was unaffacted, and cells plated on microspheres could be sorted, replated, expanded, and re-examined using the same assay multiple times (data not shown). After several rounds of cell division, the beads were eventually diluted away, and the cells could be re-assayed again.
If the accumulation of beads is affected by other factors in addition to motility, fluorescence will not reflect motility accurately. We believe that bead ingestion is not affected by factors other than motility for a number of reasons: 1) the beads are adherent to the substrates and thus are not free to be ingested by cells without direct contact; 2) the tracks generated by migrating cells are continuous, bead-free paths, indicating that cells do not exhibit periods of migration without bead ingestion; 3) accumulation of fluorescent beads correlates directly with the degree of cell migration as recorded by cleared areas, as well as the quantity of beads cleared (Figure [4](#F4){ref-type="fig"}). Fluorescence of the cells was identical to the fluorescence of an equal area of uncleared microspheres adjacent to the cleared area (no evidence of differences by paired t-tests, data not shown), suggesting that all of the beads that are cleared are ingested by the cell.
There is undoubtedly a maximum quantity of beads that any given cell can ingest, and as the cell approaches this limit with sustained migration, the relationship of cell fluorescence and motility may diverge as fluorescence plateaus. However, in our studies, we did not observe a plateau in fluorescence after 24 hours of migration, with a linear relationship maintained between fluorescence and motility (Figure [4](#F4){ref-type="fig"}). It is conceivable, however, that highly motile, slowly dividing (or non-dividing) cells might approach a limit to bead consumption, or that cells may migrate differently upon the microsphere substrate than on uncoated substrates. These problems could be addressed by either using a lower concentration of beads or limiting the assay time.
Existing methods to quantify motility such as transwell assays, wound healing assays, and cell outgrowth assays can be complicated by cell division, and investigators have resorted to including mitotic inhibitors in their assays. In contrast, cells that have divided in the fluorescent phagokinetic assay are easily identified by the cleared trail leading from the two cells (example shown in Figure [1C](#F1){ref-type="fig"}). As a consequence, we can easily measure the path areas of single cells to exclude cells that have divided in the period since plating. Although one approach typically used to quantify motility when utilizing the phagokinetic assay is to report a total area cleared, we feel that reporting the area cleared per cell is more representative of cell motility than reporting total area cleared, as this quantity will not be confounded by cell division.
This method can complement existing assays of cell motility. It is especially useful when starting cell numbers are limiting, or when recovery of particular populations is desirable. Furthermore, in contrast to the commonly used transwell assay (which documents the movements of the highest motility cells in a population), this assay characterizes an unselected sample of the population, and can therefore provide unbiased information about motility characteristics of the entire cell population.
Behavioral readouts are the most direct screens for molecular pathways that regulate the behavior of interest. Although traditionally, we think of screens on an organism level, here we describe screening for differences in a specific cell behavior. Populations of cells can be selected by *in vivo*behaviors as well. For example, characterization of cells selected for the ability to metastasize revealed a number of genes involved with tumor cell invasion \[[@B16]\]. As demonstrated here, further resolution on a single cell basis can be valuable even in apparently homogeneous or clonal cell populations, as individual cells can exhibit wide variations in motility. To gain additional insight onto the molecular mechanisms underlying differences in cell behavior, reporter assays for gene expression with single cell fidelity \[[@B17]\], combined with genome-wide analysis of expressed genes \[[@B18],[@B19]\] could be used to complement our behavioral assay in screens.
Here, we observe persistent differences in motility of a clonally-derived cell line. Further characterization can determine the molecular underpinnings that generate the differences in intrinsic cell motility. It is apparent that further heterogeneity exists within the sorted populations (data not shown); our approach enables 1) further enrichment of populations by re-sorting previously enriched cell populations and 2) finer resolution of cell populations by sorting into single cells with subsequent characterization using a variety of approaches \[[@B4],[@B18],[@B19]\].
Conclusion
==========
The advantages of the fluorescent phagokinetic assay relative to existing assays of cell motility are as follows. First, the preparation of substrate is straightforward, consisting merely of applying the fluorescent microspheres to the substrate, and allowing them to adhere. Second, the assay is highly sensitive- single cell characteristics, including the behavior of individual transfected cells, can be obtained in an unbiased fashion. Finally, because cell fluorescence is directly related to the area of fluorescent particles cleared, distinct subpopulations of cells can be sorted and enriched based on degree of motility. This assay could be used to identify factors that might regulate motility differences even within clonal population of cells. This approach potentially may be extended beyond two dimensions by creating three dimensional suspensions of fluorescent particles in solid matrices to also provide screens to identify additional, critical parameters of motility and oncogenesis *in vivo*.
Methods
=======
Transwell migration assay
-------------------------
Cell migration through transwell filters was analyzed as previously described \[[@B20]\], with minor modifications. Briefly, 1 × 10^5^T98G and U118MG cells were seeded on the top of transwell membranes treated with 50 μg/ml poly-D-lysine for 30 min (8 μm pore diameter; Becton Dickinson) in media (DMEM) with 20% fetal bovine serum in both upper and lower compartments and allowed to migrate. After 6 hours, filters were fixed with 4% paraformaldehyde (15 min, 4 deg C), cells from the top surface of the filters were removed by a cotton swab, and nuclei stained with Hoescht 33342. Cells that had migrated through to the bottom surface of the filter were then counted under UV fluorescence, and confirmed by visualization with phase optics (5 fields/filter using a 10X phase objective). The average number of cells in four replicate wells was determined for each cell line in each of three independent experiments.
Fluorescent phagokinetic migration assay
----------------------------------------
Tissue culture vessels were prepared by pre-treatment with 50 μg/ml poly-D-lysine for 30 min at room temperature (RT), then coated with 1 μm diameter fluorescent microspheres \[Fluospheres (carboxylate-modified, yellow-green, Molecular Probes F8815, or carboxylate-modified, blue F8814, 0.005% in Dulbecco\'s phosphate buffered saline (DPBS)\] for 2 hours (RT), and washed three times with DPBS. For migration on various extracellular matrices, tissue culture dishes precoated with fibronectin, collagen IV, laminin (Biocoat, BD Biosciences catalog numbers 354428, 354402, 354404) were rinsed and coated with beads as above. Cells were plated onto these substrates at a density of \~4 cells/mm^2^in DMEM with 20% fetal bovine serum, incubated for 18--24 hours, and then fixed with 4% paraformaldehyde, 4 deg C, 15 minutes, and washed with DPBS.
Analysis of motility
--------------------
Images of fluorescent cells on green microspheres were collected using a 10X phase objective, 0.3 NA, and Endow GFP filter cube. A 16 bit CCD camera (Cascade 650, Roper Inc.) was used to capture images. Cleared areas generated by cells were traced, and cleared area and total fluorescence of individual cells measured using the Metamorph imaging program. To prevent confounding results by overlapping paths and dividing cells, only areas cleared by single cells were measured and counted. To correlate cleared areas with fluorescence intensity, all images were collected using the same exposure time (25 ms), which was determined emperically to ensure that no pixels in the collected images were saturated. Confocal images were obtained on a Zeiss LSM510.
Transfections and staining
--------------------------
Cells were transfected with expression plasmid for EGFP using Lipofectamine 2000 following manufacturer\'s protocols. 18 hours after transfection, cells were plated onto microsphere coated dishes and allowed to migrate for 18--24 hours. Following migration assay, cells were permeablized with 0.3% Triton X-100 in phosphate buffered saline (PBS), and actin was visualized by staining with Alexa 546- labeled phalloidin (0.15 μM in PBS for 20 minutes, room temperature; Molecular Probes); DNA was visualized by incubation with Hoescht 33342 (1 μg/ml in PBS for 5 minutes, room temperature; Molecular Probes)
Flow cytometry
--------------
U118MG cells were plated on microsphere-coated 10 cm tissue culture dishes, incubated for 20 hours, removed using 0.25% trypsin -- 1 mM EDTA, and spun at 180 g for 10 minutes. Approximately 2.6 × 10^5^cells were sorted (Beckman Coulter Epics Elite ESP Cell Sorter) based on fluorescence intensity and side scatter characteristics. After sorting with gates arbitrarily chosen to select for approximately the top and bottom thirds of the total population by fluorescence, 4.1 × 10^4^highly fluorescent U118MG\'s and 2.4 × 10^4^lower fluorescence cells were recovered and returned to culture. Fluorescent signal and forward/side scatter signal is displayed on a linear scale, using 1023 available channels of resolution. Fluorescence intensity at 525 nm is displayed as typical for the flow cytometer used; there are 1023 bins into which the signal can be resolved (12 bit analog to digital converter (ADC)). An untransformed linear scale for fluorescence intensity is used. On a linear scale, the voltage measured (signal intensity) is directly proportional to the channel into which the event falls. A cell with a linear value of 100 is 10 times brighter than one in with a linear value of 10. As typical for flow cytometry, gain and threshold is set within an experiment so that the fluorescent signal is resolved into these channels. Similarly, forward and side scatter signal height is displayed in typical fashion (binned into 1023 channels). The y-axis for figure [4A](#F4){ref-type="fig"} histogram represents the number of cells in each of the histogram bins.
Primary astrocyte cell culture
------------------------------
Primary mouse cerebral cortical astrocytes were isolated from newborn mouse cortices as described in \[[@B21]\] with the following modifications. After dissection, the cells were plated on untreated tissue culture plates at 1 × 10^5^/ml in serum free DMEM with G3 glial supplement (Gibco). Cells were maintained in cultures and motility assayed within three passages.
Authors\' contributions
=======================
SH optimized the experimental protocols, participated in the design of the experiments, performed motility assays, flow sorting experiments, and assisted in drafting the manuscript. KC collected microscope images of migrating cells, quantified cleared areas, and performed the motility assay on different extracellular matrices. AC conceived of the study, participated in its design and coordination, performed the statistical analyses, quantified fluorescence vs. motility, and drafted the manuscript. All authors read and approved the final manuscript.
Acknowledgements
================
Supported by Sontag Foundation Distinguished Scientist Award and American Cancer Society, Illinois Division (AC). We thank G. Albrecht-Buehler for advice and discussion, M. Paniagua for flow cytometry, and G. A. Smith for microscopy expertise. We thank G. J. Woodhead for confocal microscopy. We thank R. J. Ferland, K. Haldar, V. L. Sheen, D. A. Sipkins, G. A. Smith, and C. N. Wrobel for helpful comments on the manuscript.
|
PubMed Central
|
2024-06-05T03:55:55.185039
|
2005-3-22
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079802/",
"journal": "BMC Cell Biol. 2005 Mar 22; 6:14",
"authors": [
{
"first": "Sarah L",
"last": "Windler-Hart"
},
{
"first": "Kwan Y",
"last": "Chen"
},
{
"first": "Anjen",
"last": "Chenn"
}
]
}
|
PMC1079803
|
Background
==========
The rab/Ypt proteins represent a large family of small GTPases involved in several transport steps of the cellular trafficking pathway. At present 11 members have been identified in yeast and more than 60 human rab proteins are supposed \[[@B1],[@B2]\]. Small GTPases are known to function as molecular switches and can cycle between an active GTP-bound and an inactive GDP-bound conformation \[[@B3],[@B4]\]. In the active conformation, the rab GTPases are localized on membranes, where they recruit specific effector molecules to mediate vesicle attachment \[[@B5]\]. The subsequent fusion step is initiated, when the membranes are forced into close proximity by the interaction of SNARES (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) presented on both membranes in concert with SM proteins (Sec1/Munc18-like proteins), thought to act as controlling elements in SNARE interaction \[[@B6]\].
The two isoforms of the small GTPase rab1 (rab1a/b) are localized on ER-Golgi membranes and regulate the exocytotic transport from the ER to the Golgi apparatus \[[@B7]-[@B9]\]. Although the function of rab1 is still not fully understood, several lines of evidence describe rab1 and its yeast homolog Ypt1p GTPase as a major component in a protein complex responsible for directing vesicular traffic between the ER and the Golgi apparatus \[[@B7]-[@B13]\]. In the past, several approaches have been established to identify effector molecules that bind specifically to the active GTP-bound form of rab1 \[[@B14]-[@B17]\]. At present, four specific effector proteins and several additional putative effector molecules are known \[[@B14]\]. The first characterized rab1 interacting protein was p115, which also binds to the Golgi-associated proteins GM130 and Giantin \[[@B14],[@B18],[@B19]\]. p115 is the mammalian homologue of the yeast protein Uso1p, which was found to be an essential component necessary for the tethering of ER-derived COP II vesicles to the Golgi compartment, a process that depends on Ypt1p, but is independent of SNARE proteins \[[@B20]\]. In addition, p115 plays a key role in coordinating sequential tethering and docking of COPI vesicles to Golgi membranes. Thus, it is thought to function as a linker protein between Giantin and GM130 \[[@B19],[@B21],[@B22]\]. Second, the cis-Golgi matrix protein GM130 was identified as a direct rab1 effector, which also binds to a complex containing GRASP65, a lipid-anchored GM130-binding protein, that regulates Golgi stack reformation following mitosis, as well as so far uncharacterized proteins \[[@B15],[@B17],[@B23],[@B24]\]. A third interacting molecule of activated rab1, golgin-84, is supposed to act as a novel mitotic target \[[@B16]\]. Overexpression or depletion of golgin-84 results in fragmentation of the Golgi ribbon and the protein is thought to be required for the incorporation of membranes into the Golgi apparatus \[[@B16],[@B25]\]. Golgin-84 is a type II C-terminally anchored cis-Golgi protein with an extensive cytoplasmic coiled-coil domain, which shows structural and sequence similarity to Giantin \[[@B26]\]. However, golgin-84 does not appear to physically interact with other cis-Golgi matrix proteins such as GM130, p115 or GRASP65 and therefore it is likely, that additional potential golgin84-binding molecules exist \[[@B16]\]. Recently, we identified MICAL as a fourth rab1 interacting partner and suggested a link between the Golgi apparatus associated rab1 and the intermediate filament cytoskeleton \[[@B27]\].
Here, we describe Iporin as a novel interacting molecule for the active GTP-bound conformation of rab1. Iporin ([I]{.underline}nteracting [p]{.underline}rotein [o]{.underline}f [r]{.underline}ab1) was identified in a yeast two-hybrid screen with rab1b Q67R as bait and sequence analyses revealed, that the encoded protein is similar to the C-terminal half of the KIAA0375 polypeptide. The rab1b-Iporin interaction is nucleotide-dependent and rab1-specific. Interestingly, the rab1 binding motif of Iporin was mapped to a region containing a RUN domain, a protein motif, which is known to be involved in the function of small GTPases. Furthermore we examined the intracellular localization and tissue-specific expression of Iporin and give insight into its domain structure. Interestingly our results also showed an interaction with the rab1 effector GM130 via its coiled-coil region 6 \[[@B15]\].
Results
=======
Identification of a novel rab1 interacting protein
--------------------------------------------------
With the rab1b Q67R mutant as bait and a human placenta cDNA library as prey we performed yeast two-hybrid screening assays to identify novel interacting partners for the active GTP-bound conformation of the small GTPase rab1b. One of the isolated clones could be identified as the human Golgi matrix protein GM130 and a second clone encoded for a fragment of MICAL1b \[[@B15],[@B27]\]. Another interesting clone was K17, which displayed a strong and specific interaction with rab1b Q67R in the yeast two-hybrid system (Table [1](#T1){ref-type="table"}). Nucleotide sequence analysis revealed that K17 represents the cDNA sequence encoding the C-terminus of KIAA0375 (GenBank Accession number: AB002373.2). We named the identified protein Iporin (for [I]{.underline}nteracting [p]{.underline}rotein [o]{.underline}f [r]{.underline}ab1) and the deleted K17 clone Iporin ΔN847. Since the rab1 effector GM130 has been shown to interact with various rab proteins (rab1, rab2, rab33b), we wanted to investigate the rab1 specificity of the Iporin interaction \[[@B16],[@B23],[@B28]\]. Thus, we used an assortment of rab cDNA molecules as bait constructs in yeast co-transformation assays (Table [1](#T1){ref-type="table"}). Iporin ΔN847 (K17) induced strong growth and β-galactosidase activity only, when co-transformed with the permanent active mutants of rab1b (rab1bQ67R/L) and rab1a (rab1a Q70L).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Interaction of Iporin ΔN847 with different rab proteins/isoforms
:::
**prey Iporin ΔN847**
------------ ----------------------- ----- -----
pAS2-1 +++ \- \-
rab1a wt +++ \+ +/-
rab1a S25N +++ \- \-
rab1a Q70L +++ +++ +++
rab1b wt +++ +/- \-
rab1b S22N +++ \- \-
rab1b Q67R +++ ++ +++
rab1b Q67L +++ ++ +++
rab2 wt +++ \- \-
rab2 S20N +++ \- \-
rab2 Q65L +++ \- \-
rab6 wt +++ \- \-
rab6 T27N +++ \- \-
rab6 Q72R +++ \- \-
Ypt1 wt +++ ++ ++
Ypt1 S22N +++ \- \-
Ypt1 Q67L +++ +++ +++
Results obtained from co-transformation assays in yeast. Clones were cultivated on selection plates lacking tryptophan and leucine (control) or lacking additionally histidine and were supplemented with 50 mM 3-aminotriazol (his3). The β-galactosidase activity was determined using X-gal as substrate on filter lift assays (lacZ). - = no growth on selection plates or no β-galactosidase activity; +/- = background growth or β-galactosidase activity appears overnight; + = slow growth; ++ = strong growth or β-galactosidase activity appears after a few hours; +++ = very fast growth or high β-galactosidase activity.
:::
Interestingly, an interaction between Iporin ΔN847 and the wild type as well as the permanently active mutant of the yeast counterpart of mammalian rab1, Ypt1p, was observed pointing to a conserved interaction of these two molecules \[[@B5]\]. The empty pAS2-1 vector, the wildtypes of the rab1 isoforms (rab1a/b), permanently inactive mutants of the rab1a/b (rab1a S25N, rab1b S22N) and Ypt1p (Ypt1 S22N) proteins were not able to significantly activate reporter gene expression (Table [1](#T1){ref-type="table"}). There were also no interactions of IporinΔN847 with the small GTPases rab6 and rab2 (Table [1](#T1){ref-type="table"}). These results indicate, that the interaction with Iporin ΔN847 is rab1-specific and requires the GTP-bound, active conformation of rab1.
Mapping the Iporin binding site of rab1b
----------------------------------------
Next, we wanted to analyze, which structural elements of rab1b are necessary or sufficient for the interaction with Iporin. Since we already demonstrated that the hypervariable regions of rab1 are necessary for an interaction with GM130, we created rab1b mutants as well as rab1/rab6 chimeras (Figure [1A](#F1){ref-type="fig"}) \[[@B15],[@B29]\]. The mutants and the chimeras enabled us to restrict specific regions for the interaction to the C- or N-terminal half of rab1. Co-transformation experiments in yeast revealed that only the rab1b deletion mutant rab1b Q67R ΔC197 lacking the prenylation site of rab1b showed a strong interaction with Iporin ΔN847, suggesting that the prenylation site is not necessary for the interaction with Iporin ΔN847 (Figure [1B](#F1){ref-type="fig"}). The deletion mutants rab1b Q67R ΔN9, rab1b Q67R ΔC163 and rab1b Q67R ΔC163-196 failed to interact with Iporin ΔN847 (Figure [1B](#F1){ref-type="fig"}). These mutants were constructed based on the model previously proposed by Bourne *et al*. and Valencia *et al*.. G1-G3 and PM1-PM3 are conserved regions in all members of the ras superfamily of GTPases, that are involved in guanine (G) and phosphate/Mg^2+^(PM) binding, respectively \[[@B3],[@B4],[@B30]\]. According to the proposed model suggested by Seabra and coworkers, which defined rab conserved sequences as rab family motifs (RabF1-5) and other regions as subfamily motifs (RabSF1-4), the generated mutants lack the RabSF1 or RabSF4 region (Figure [1A](#F1){ref-type="fig"}) \[[@B31]\]. Co-transformation with the two rab1/rab6 chimeras displayed an interaction between the C1- and C2-rab1b Q67R-rab6a chimera and Iporin/Iporin ΔN847, but not between the rab6a Q72L-rab1b chimera and Iporin ΔN847 (Figure [1B](#F1){ref-type="fig"}).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Determination of the Iporin interacting domain. (A)**Schematic overview of the domain structure of rab1b with the conserved G1-G3 boxes, PM1-3 motifs, N-terminal and C-terminal hypervariable regions (hpr) and the C-terminal prenylation site (ps). SF1-4 indicate the rab subfamily regions. The rab1b regions of the chimeras are shaded dark and the rab6 regions are light-shaded. **(B)**All constructs were co-transformed in yeast and cells were cultivated on selection plates as described above (Table 1). Rab1b deletion mutants were co-transformed with Iporin ΔN847, whereas chimeras were co-transformed with full-length Iporin. - = no growth on selection plates or no β-galactosidase activity; +/- = background growth or β-galactosidase activity appears overnight; ++ = strong growth; +++ = very fast growth or high β-galactosidase activity
:::
:::
Full-length cloning and characterization of Iporin
--------------------------------------------------
Sequence analysis of the clone K17 cDNA revealed an open reading frame of 669 amino acids, which was identical to the C-terminal half of the protein encoded by the KIAA0375 cDNA (aa 848--1516). To obtain the full-length Iporin sequence, we used the KIAA0375 cDNA for the amplification of the sequence encoding the missing N-terminal fragment by PCR.
Information about the genomic structure of Iporin protein was obtained from the human genome database. The Iporin gene consists of \~14 kb of genomic sequence and contains eleven exons (Figure [2A](#F2){ref-type="fig"}). The gene is located on the short arm of chromosome 9 (9p13.1) and comprises 4548 bp of coding DNA. Deduced from the cDNA sequence, Iporin comprises 1516 amino acids with a calculated molecular mass of 161.195 kDa and an acidic pI of 6.17. Further structure analysis revealed functional domains and motifs in the Iporin sequence (Figure [2B](#F2){ref-type="fig"}). Iporin contains some low complexity regions, which include polyproline (pP) and polyglutamic acid (pE) stretches. The first polyproline stretch is localized between aa 429 and aa 442 with the sequence PPPGPGPDPGPSQP and a second region between aa 1311 and aa 1319 with the sequence PPQAPPP. Proline-rich stretches are known to interact with SH3 or WW domains and could mediate further protein/protein interactions \[[@B32]\]. The polyglutamic acid stretch, spanning aa 1236 -- 1252 with the sequence EGGEEEEEEEETEEVAE is flanked by the two proline-rich regions. This motif is highly negatively charged and may play a role in stabilizing the protein or function as further interaction domain.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**2 Genomic structure and functional elements of Iporin. (A)**Schematic representation of the human Iporin gene located on chromosome 9. The protein coding exons of Iporin are shown as shaded boxes numbered one to eleven, non-coding sequences are white. The gene (\~14 kb) encodes a protein of 1516 amino acids. **(B)**Domains and structural motifs of Iporin. The most interesting domains within Iporin are the SH3 (aa 1447--1500) and the RUN domain (aa 1031--1175). The proline-rich motifs (pP, aa 429--442; aa 1311--1319) are possible interaction motifs for SH3 or WW domain containing proteins. The glutamic acid-rich region (pE, aa 1236--1252) is negatively charged with possibly stabilizing function. **(C)**Alignment of RUN domains from related proteins. The sequence corresponding to the RUN domain of Iporin (aa 1031--1175, human) was aligned to the RUN domains present in AI840675 (aa 309--453, mouse), NESCA (aa 53--197, human), RPIP8 (aa 52--189, human), Rabip4 (aa 31--163, mouse) and Rufy1 (aa 31--163, human) (GenBank accession numbers, XP\_131380.2, BAA77507.2, NP\_006686.1, CAC17732.1 and AAK50771.1, respectively). RUN domains were detected using the Motif Scan software <http://www.expasy.org>. Hydrophobic residues (V, L, I, F, Y, W, M, C) are show as boxes, conserved positions are shaded. A-F are the six conserved blocks, and distances between these blocks are indicated by the numbers of residues in parentheses. **(D)**Alignment of sequenced clones identified by phage display using the Iporin SH3 domain. Four different clones were isolated, whereby the number of independent clones, which contained the same sequence, are in brackets. The conserved consensus motif is shaded. Ψ = aliphatic amino acid; X~2~= two individual amino acids.
:::
:::
Using the Motif Scan Software <http://www.expasy.org> we also detected a RUN ([R]{.underline}PIP8, [U]{.underline}NC-14 and [N]{.underline}ESCA) domain in the Iporin sequence (aa 1031--1175; Figure [2B](#F2){ref-type="fig"}). This domain was shown to be involved in the function of small GTPases and can also be found in proteins which play a role in proliferation, differentiation and motility \[[@B33],[@B34]\]. The RUN domain is organized in six conserved blocks (Figure [2C](#F2){ref-type="fig"}; A-F), which are predicted to constitute the \"core\" of a globular structure. Figure [2C](#F2){ref-type="fig"} shows an alignment of different RUN domains containing proteins related to small GTPases like RPIP8, Rabip4, Rufy1, NESCA or Iporin \[[@B34]-[@B36]\]. RUN domain containing proteins enclose hydrophobic amino acids in conserved positions as shown by shaded or boxed regions (Figure [2C](#F2){ref-type="fig"}). The secondary structure of the RUN domain core was predicted to consist of predominantly α-helices and the conserved three dimensional structure is probably of importance. However, the exact function of this motif is still unclear \[[@B33]\].
Interestingly, NESCA ([n]{.underline}ew molecule containing [S]{.underline}H3 at the carboxy-terminus) and Iporin share 42% of sequence homology and contain both a SH3 domain at the C-terminus. The function of NESCA is not known, but it is thought to be involved in signal transduction pathways and was recently described as a signaling adaptor which shuttles from the cytosol to the nuclear envelope \[[@B37],[@B38]\]. We used NESCA as a prey for yeast two-hybrid assays to investigate, whether NESCA might be a further interacting partner of the small GTPase rab1. However, we were not able to show an interaction between NESCA and rab1 (data not shown).
BLAST searches revealed a mouse homolog of Iporin, EST clone AI840675, which shows 84% similarity to the human protein. This clone encodes 736 amino acids, which are homologous to Iporin aa 723 to 1455. The truncated mouse Iporin protein contains only a RUN domain and we suppose that it might function in a different, but related manner, to human Iporin. The C-terminus of Iporin contains a SH3 domain (aa 1447--1500), which is found in a variety of proteins (Figure [2B](#F2){ref-type="fig"}) \[[@B39]\]. This domain is known to interact with polyproline domains in target molecules and is involved in transmembrane signaling and cytoskeletal rearrangements \[[@B40]\]. To identify the interacting motif of the SH3 domain of Iporin we performed a phage displayed analysis. For this purpose we screened a fUSE5/15-mer M13 phage displayed random peptide library using purified GST-Iporin SH3 (Iporin ΔN1430) as a target. Sequencing of the cDNA from phages with a high and specific affinity towards GST-Iporin SH3 identified a new consensus motif (Figure [2D](#F2){ref-type="fig"}). This motif contains two prolines followed by an aliphatic amino acid (A, I, L, V), two further amino acids and a conserved methionine (PPΨX~2~M).
Mapping the rab1 binding domain of Iporin
-----------------------------------------
To investigate, which part of the Iporin sequence acts as a rab1 binding domain (R1BD), we generated several deletion mutants of Iporin (Figure [3A](#F3){ref-type="fig"}) and tested them in yeast two-hybrid assays (Figure [3B](#F3){ref-type="fig"}). As expected, the N-terminus of the protein (Iporin ΔC853), which is not present in the original prey clone as well as the SH3 domain (Iporin ΔN1430) are not necessary for the interaction with rab1b and failed to activate the reporter genes (Figure [3B](#F3){ref-type="fig"}). Only the RUN domain-containing mutants (Iporin ΔN847, ΔN847ΔC1450, ΔN847ΔC1239, ΔN991ΔC1177, ΔN991ΔC1239 and ΔN991ΔC1450) showed growth and β-galactosidase activity on selection plates (Figure [3B](#F3){ref-type="fig"}). Interestingly, the interaction of rab1b with the shortest RUN domain containing mutants (ΔN991ΔC1177, ΔN991ΔC1239, respectively) was weaker compared to fragments containing longer flanking sequences, which could result from protein misfolding.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Mapping the rab1 binding domain (R1BD) of Iporin. (A)**Overview of deletion mutants used for yeast co-transformation assays. The deletion mutants ΔNx or ΔCx do not contain the N-terminus or the C-terminus, respectively. We were able to map the rab1 binding domain (R1BD) to a RUN domain containing region. **(B)**Results obtained from yeast co-transformations. The bait construct rab1b Q67R was cloned into the pAS2-1 vector and the prey constructs into pGADT7. Clones were cultivated on selection plates as described previously (see Table 1). **(C)**GST-pulldown with coupled GST-Iporin ΔN847ΔC1239 indicated an interaction with Iporin, whereas GST-Iporin ΔN847ΔC992 and GST were negative. The lower panel shows the Coomassie blue stained blot membrane with equal amounts of GST fusion protein. - = no growth on selection plates or no β-galactosidase activity; +/- = β-galactosidase activity appears overnight; + = slow growth or β-galactosidase activity appears after several hours; ++ = strong growth; +++ = very fast growth or high β-galactosidase activity; pP = polyproline; pE = polyglutamic acid.
:::
:::
To confirm the yeast two-hybrid mapping results, we performed GST-pulldown assays. Bacterially expressed GST-Iporin ΔN847ΔC1239, GST-Iporin ΔN847ΔC992 and GST were coupled to glutathione-Sepharose and incubated with extract from BHK cells overexpressing HA-tagged rab1b Q67R (Figure [3C](#F3){ref-type="fig"}). In this assay, the GST-Iporin ΔN847ΔC1239 was able to bind to the active form of rab1b, whereas GST and GST-Iporin ΔN847ΔC992 failed to interact (Figure [3C](#F3){ref-type="fig"}). These findings confirmed the yeast data (Figure [3B](#F3){ref-type="fig"}) and indicate that the rab1 binding domain of Iporin needs the whole RUN motif.
Tissue-specific expression and cellular localization of Iporin
--------------------------------------------------------------
Northern blot analysis of various human tissues indicated, that Iporin is ubiquitously expressed. The Iporin mRNA was identified as a prominent band of approximately 5.5 kb. Highest amounts of Iporin transcripts were observed in brain and testis (Figure [4](#F4){ref-type="fig"}). These findings are consistent with the RT-PCR results, published by the Kazusa institute, Japan.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Tissue-specific expression of Iporin mRNA.**A 12-lane multiple-tissue Northern blot was hybridized with a specific ^32^P-labeled cDNA probe for Iporin. Each lane contains approximately 2 μg of poly A^+^from various tissues. The blot was exposed to an X-ray film for 17 h at -80°C with an amplifying screen. To confirm equal loading, the membrane was re-hybridized with a radiolabeled probe specific for β-actin mRNA.
:::
:::
The cellular distribution of Iporin was examined by cytosol/membrane fractionation and immunofluorescence.
Although the calculated mass of Iporin (deduced from the cDNA sequence) is 161,195 kDa, the endogenous protein is detected at about 220 kDa in Western blot analyses (Fig. [5Ac](#F5){ref-type="fig"} first row). We assume that the molecular weight shift observed in the Western blot analyses could result from posttranslational modifications within the aminoterminal part of the Iporin molecule, because transient transfections in HeLa cells with aminoterminal constructs showed the highest difference between measured and calculated weight (data not shown). However, in lysates from transfected HeLa cells expressing HA-tagged full length Iporin, the affinity-purified antibody Aip519 detected a 220 kDa protein as did the 16B12 antibody, which is directed against the HA epitop (Figure [5Aa](#F5){ref-type="fig"}).
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Cellular distribution of Iporin. (Aa)**To test specificity of the polyclonal anti Iporin antibody Aip519, HeLa cells were transiently transfected with HA-tagged Iporin. After lysis, cell extracts were blotted and incubated with the Aip519 and the monoclonal 16B12 anti HA antibody, respectively. The Western blot displays a signal of about 220 kDa for both antibodies. **(Ab)**Cytosol/membrane fractionation with extracts from transiently transfected HeLa cells. Recombinant HA-Iporin was detected with anti HA (16B12) antibody. Fractionation of full-length HA-Iporin and HA-Iporin ΔN847 revealed that most of the recombinant proteins were present in the membrane fraction (P100) with small amounts in the cytosolic (S100) fraction. In contrast, the transiently expressed HA-Iporin ΔC853 was present in a high concentration in the cytosolic (S100) fraction. **(Ac)**Comparison between untransfected and HA-rab1bQ67L transfected HeLa cells. Blots were analyzed using the Aip519 (Iporin), anti-GM130 (GM130), anti-p115 (p115), anti-calnexin and 1E7 (anti-rab1b) antibodies. **(B)**Immunofluorescence staining of recombinant HA-Iporin expressed in HeLa cells shows a cytosolic distribution with a perinuclear enrichment in spotted structures. Bar 10 μm.
:::
:::
To obtain information about the membrane association of Iporin, we performed cytosol/membrane fractionation and therefore we transiently transfected HeLa cells with vectors encoding full-length Iporin, an aminoterminal fragment (Iporin ΔC853) or the C-terminal fragment (Iporin ΔN847), respectively (Figure [5Ab](#F5){ref-type="fig"}). Transfected cells were lysed and extracts were separated by differential centrifugation steps into cytosol (S100) and membrane (P100) fractions. Equal amounts of protein were analyzed by Western blotting. Calnexin distribution was used as a marker for integral membrane proteins to verify the purity of the fractions (not shown). Our data demonstrate that full-length HA-Iporin and HA-Iporin ΔN847 seem to associate with cellular membranes or insoluble parts of cells, because major part of the proteins was detected in the P100 fractions (Figure [5Ab](#F5){ref-type="fig"}). Interestingly, the HA-Iporin ΔC853 mutant, which lacks the C-terminus was highly enriched in the cytosolic fraction, indicating that parts of the C-terminus play an important role in the cellular localization of Iporin (Figure [5Ab](#F5){ref-type="fig"}). To address the question, whether the overexpression of permanently active rab1b influences the cellular distribution of Iporin, we transiently transfected HeLa cells with vectors encoding for HA-tagged rab1b Q67L and performed further fractionations (Figure [5Ac](#F5){ref-type="fig"}). The comparison between non-transfected and transfected HeLa cells showed, that the overexpression of HA-rab1b Q67L had no remarkable effect on the distribution of Iporin. GM130, p115 and Calnexin were used as control proteins and showed as well no significant changes.
Hence our Aip519 polyclonal antibody was inapplicable for immunohistochemistry we transiently transfected HeLa cells with HA tagged Iporin and used the anti HA-antibody 16b12 for staining. The immunofluorescence shows a cytosolic pattern of HA-Iporin, enriched in small, brighter spots. However, most of the cells showed a faint perinuclear enriched fluorescence where usually the Golgi/ER structures are localized (Figure [5B](#F5){ref-type="fig"}).
Iporin interacts with the Golgi matrix protein GM130
----------------------------------------------------
Surprisingly, yeast two hybrid co-transformation experiments showed an additional interaction between the two rab1 binding proteins Iporin and GM130. Mapping of the binding site of GM130 revealed, that the coiled-coil domain 6 is responsible for the interaction with Iporin, whereas GM130 coiled-coil domain 3 mediates the interaction with rab1b (Figure [6A](#F6){ref-type="fig"}) \[[@B15]\]. To confirm the yeast data, we performed *in vitro*binding assays. Recombinant GST-Iporin ΔN847ΔC1239 as well as GST-Iporin ΔN1430 proteins were coupled to glutathione Sepharose beads and then incubated with HeLa SS6 cell extracts, which expressed the GM130 protein in considerable amounts. After washing steps, the bound proteins were separated by SDS-page and analyzed by Western blotting analysis using GM130 antibodies. As shown in figure [6B](#F6){ref-type="fig"}, only GST-Iporin ΔN847ΔC1239 is able to bind to endogenous GM130, whereas the GST-IporinΔN1430 was unable to interact (Figure [6B](#F6){ref-type="fig"}).
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**Interaction of Iporin with GM130. (A)**Yeast co-transformations revealed an interaction between Iporin and GM130. Using different GM130 deletion mutants, we were able to map the binding site to the coiled-coil region 6 (Figure 6A). The bait construct IporinΔN847ΔC1239 was cloned into the pAS2-1 vector and the prey constructs into pGADT7. Clones were cultivated on selection plates as described previously (see Table 1). = no growth on selection plates or no β-galactosidase activity; ++ = strong growth or β-galactosidase activity appears after a few hours; +++ = very fast growth or high β-galactosidase activity **(B)**GST-pulldown of endogenous GM130 with GST-IporinΔN847ΔC1239 or GST-Iporin ΔN1430. After incubation with HeLa cell extracts, GST-IporinΔN847ΔC1239 showed an interaction with GM130, whereas GST-Iporin ΔN1430 did not bind to GM130. The lower panel shows the Coomassie blue stained blot membrane with equal amounts of GST fusion protein.
:::
:::
Discussion and conclusions
==========================
The cellular function of the small GTPase rab1 is not fully understood, thus the identification of novel rab1 interacting molecules may help to elucidate its role in vesicular transport. Using the yeast two-hybrid system, we were able to isolate Iporin as a new protein, which specifically interacts with the activated, GTP-bound conformation of rab1. Interestingly, Iporin also showed an interaction with the rab1 homolog from yeast Ypt1p, indicating that formation of this protein/protein complex is conserved during evolution.
Northern blot analysis of Iporin mRNA revealed an ubiquitous expression pattern (Figure [4](#F4){ref-type="fig"}). Highest amounts of Iporin mRNA were observed in brain and testis, suggesting that Iporin might have a distinctive role in these tissues. Neurogenesis as well as spermatogenesis are processes that result in the production of highly polarized cells and we speculate that Iporin might act as a regulator in these cellular systems \[[@B41]\].
Applying a set of deletion mutants and chimeras of rab1, we demonstrated that the Iporin binding domain is localized within the N-terminal half of rab1 and that the rab1 N-terminal hypervariable region is essential for the interaction (Figure [1](#F1){ref-type="fig"}). In contrast to PRA1, another known rab1 effector protein, our results show that the prenylation site of the rab1 GTPase is not necessary for the binding to Iporin \[[@B42]\]. A well characterized interaction between a rab GTPase and an effector molecule is the rab3a/rabphillin-3a interacting complex \[[@B43]\]. The specific interaction of rab3a with rabphillin-3a is mediated through a rab3a \"pocket\" contributed by three rab complementary-determining regions (RabCDR), which are localized at the N- and C-terminus. The RabCDRs are proposed to form regions of variable sequences among the rab family within a structural conserved framework and act as a structural element for protein/protein interactions \[[@B43]\]. This model was supported by the work of Seabra and co-workers. They defined a model of conserved regions within the rab family (RabF) and regions only conserved among the rab isoforms (RabSF) \[[@B31]\]. Projected on a three dimensional model of rab3a, the previously described RabCDR regions are represented by the RabSF1, 3 and 4 regions. They form one interaction surface and the RabSF2 region forms another surface, almost on the opposite side \[[@B31]\]. The data presented here about the interaction of the rab1b Q67R deletion mutants and the permanently active rab1b/rab6a chimeras with Iporin and Iporin Δ847, respectively, revealed that the N-terminal half of rab1b, including RabSF1 and RabSF2, is essential for the interaction with Iporin (Figure [1B](#F1){ref-type="fig"}). The C-terminal elements, comprising RabSF3 and RabSF4, do not play a role in the formation of the complex and may be necessary for structural reasons only. Surprisingly, this is in contrast to the results received from the rab3a/rabphillin-3a complex. Our data suggest that Iporin is able to interact with parts of both interaction surfaces. Thus, these results support the theory of a multi-protein interaction interface, where molecules can interact with different parts of both surfaces.
To date, four proteins (p115, GM130, golgin-84 and MICAL) have been identified which interact specifically with the activated conformation of rab1 \[[@B14]-[@B17],[@B25],[@B27],[@B28]\]. GM130 is a cis-Golgi-localized coiled-coil protein, targeted to membranes via the peripheral membrane protein GRASP65 \[[@B44]\]. Additionally, GM130 interacts with activated rab1, rab2, rab33b, the vesicle tethering-protein p115 and as we recently observed with Iporin \[[@B15],[@B45]\]. p115 has been identified to play a pivotal role in docking of COPI vesicles to the Golgi membranes by recruiting GM130 and Giantin in a long tethering complex \[[@B21],[@B22]\]. Giantin is a transmembrane protein, located on COPI vesicles and shares similarities with the recently identified golgin-84 and CASP \[[@B16],[@B46],[@B47]\]. It is worth to note, that rab1 was also shown to recruit p115 during budding to the ER-compartment to program COPII vesicles for fusion \[[@B14]\]. Another aspect, which has been discussed to be required for tethering events at the ER-Golgi stage are large multisubunit complexes like TRAPPI (transport protein particle). TRAPPI has been identified in yeast to function as tethering factor with GEF activity (guanosine exchange factor) for Ypt1p \[[@B48],[@B49]\]. The importance of this complex for mammalian transport steps between ER and Golgi remains to be clarified, but the increasing number of rab1 interacting molecules raises the theory of a huge protein complex with rab1 in a central position. Another possibility might be that activated rab1 only mediate interactions between other proteins, e.g. Iporin and GM130, in a time or place dependent manner.
Iporin displays some features that are quite different from the known rab1 interacting molecules. First GM130 as well as golgin-84 contain coiled-coil regions that are essential structural elements for the interaction with rab1 \[[@B15]-[@B17],[@B28],[@B50]\]. However, the rab1b binding domain of p115 has not been characterized so far \[[@B14],[@B51]\]. Iporin contains no coiled-coil regions, but interacts via a RUN domain with rab1 (see below). Second the known rab1 effector proteins are membrane-associated proteins, which are involved in tethering processes between donor and acceptor membranes at ER-Golgi compartments \[[@B14],[@B17],[@B22],[@B46]\]. Iporin displays a distribution (Figure [5Ab--c, 5B](#F5){ref-type="fig"}), which is also different from known proteins belonging to the tethering complex at Golgi membranes. Similar to MICAL, the localization of Iporin is not restricted to the ER-Golgi compartment leading to the conclusion that, in contrast to GM130 or GRASP65, Iporin is not a member of the Golgi matrix protein family \[[@B15],[@B24],[@B27]\]. The cytosol/membrane fractionation of HeLa cell extracts revealed a mainly membranous association (P100) of transiently transfected and endogenous Iporin, which is an evidence for a high and stable membranous association through its function (Figure [5Ab--c](#F5){ref-type="fig"}). At this step, Iporin might recruit further proteins to form a complex needed for a regulative purpose.
Interestingly, the deletion mutants Iporin ΔN847 and Iporin ΔC853 show a different distribution in the S100 fraction. Iporin ΔC853 exhibits a cytosolic accumulation, which is a hint for a much weaker accessibility to membranes (Figure [5Ab](#F5){ref-type="fig"}). This accumulation seems to be due to the missing carboxyterminal part of the protein and not the result of overexpression effects (Figure [5Ab](#F5){ref-type="fig"}). To test the influence of overexpressed permanently active rab1b, we transiently transfected HeLa cells with HA-tagged rab1bQ67L and compared transfected to non-transfected cells. The obtained data showed no significant effect on Iporin membrane association (Figure [5Ac](#F5){ref-type="fig"}). One could argue that the membrane association is not necessarily coupled to the interaction with rab1. Like GM130, which is linked via the GRASP65 protein to membranes, another protein, not yet identified, might be responsible for the attachment of Iporin to membranes. In addition, the overexpression of HA-rab1bQ67L had also no effect on the GM130 distribution, which interaction with rab1 is well characterized. Another reason could be, that for the interaction with rab1 only a limited amount of Iporin is sufficient and that the HA-rab1b Q67L compete with the GTP-bound endogenous rab1b.
As mentioned before, Iporin contains a RUN and a SH3 domain as well as proline- and glutamic acid-rich regions. These motifs have been described to function as targets for protein/protein interactions \[[@B35],[@B52],[@B53]\]. The RUN domain was previously shown to be part of the rap2 binding region and is supposed to mediate the interaction between the small GTPase rap2 and the effector protein RPIP8 \[[@B35]\]. In contrast, the RUN domain of Rabip4, an effector protein of the small GTPase rab4, has no influence on the interaction, but was supposed to be responsible for an association with a filamentous network \[[@B36]\]. We were able to map the rab1 interacting domain to a RUN domain-containing region and we suggest a rab1-specific interacting function for the RUN domain of Iporin (Figure [3](#F3){ref-type="fig"}).
Using the phage display method we identified PPΨX~2~M as a new SH3 binding motif, which is different from the originally known PXXP motif \[[@B52]\]. The Iporin SH3 domain together with the polyproline regions may act as regulative targets for not yet identified interacting proteins. Such a link was recently described for the small GTPase rab5. RN-tre, a rab5-specific GTPase-activating protein (GAP) interacts with Grb2. This association is mediated via the SH3 domains of Grb2 and the proline-rich regions of RN-tre \[[@B54]\].
Our recent discovery of an additional not yet identified interaction partner of Iporin, the GM130, is a very interesting piece of the \"Golgi puzzle\" (Figure [6](#F6){ref-type="fig"}). As has been shown in this and previous manuscripts, both proteins, the Iporin and GM130 interact with rab1 \[[@B15],[@B17]\]. As mentioned above, GM130 is important for vesicle docking processes at the *cis*-Golgi, where it interacts with p115. GM130 itself is recruited to ER-derived vesicles and vesicular-tubular clusters by the rab1 GTPase \[[@B50]\]. One could speculate, that Iporin, while interacting with rab1b, specifies the recruitment of the incoming vesicle to the *cis*-Golgi by displaying a high affinity to GM130 by finally binding to this protein. The presented data showed that the binding site of Iporin is different from other GM130 interacting proteins. p115 binds to the N-terminus, rab1b to the coiled-coiled region 3 and GRASP65 to the C-terminus of GM130 \[[@B15],[@B18],[@B44]\]. Iporin binds to the coiled-coiled region 6 which means that they could bind in several ways e.g. at the same time, one after another, two at the same time, etc.
The multi-domain structure of Iporin and the cellular distribution suggest that this protein might act as an adapter or scaffold protein, which links GTPases to certain intracellular signal transduction pathways. Noteworthy, Iporin displays several features different from known rab1 effector proteins. Our data about the Iporin-rab1-GM130 association demonstrate once more, that the increasing number of protein/protein interactions at different stages of ER to Golgi transport implies a more complex regulated network as originally thought.
Methods
=======
Yeast two-hybrid assay
----------------------
For the yeast two-hybrid screen we cloned human rab1b Q67R cDNA into the pAS2-1 bait vector and transformed Y190 as recommended by the Matchmaker II™ manual (BD Biosciences Clontech, Heidelberg, Germany). The bait yeast strain was then transformed with a human placenta cDNA library (BD Biosciences Clontech, Heidelberg, Germany) as prey. In order to test the specificity of the interaction between rab1b Q67R and putative binding partners, Y190 yeast cells were co-transformed with various bait and prey constructs and incubated on selection plates at 30°C for 5--7 days as described by manufacturer\'s instructions (BD Biosciences Clontech, Heidelberg, Germany). A protein/protein interaction was confirmed by β-galactosidase filter lift assays using X-gal as substrate. The cDNAs of selected yeast clones were purified, transformed into *E. coli*strain DH5α and sequenced.
Cloning of the full-length Iporin sequence and generation of deletion mutants
-----------------------------------------------------------------------------
To clone the full-length Iporin sequence, we amplified the cDNA encoding the N-terminal region of Iporin (which is represented by the Iporin ΔC853 deletion mutant) using the KIAA 0375 cDNA (generous gift from Takahiro Nagase, Kazusa Institute, Japan) as template. We generated a primer with a 5\' *Nde*I overhang containing the predicted start codon and a primer with a 3\' *Eco*RI overhang beyond a single endogenous *Hind*III restriction site, which was also found in the original prey clone. The PCR product, representing the N-terminal part, was digested with *Nde*I/*Hind*III and was inserted into the eukaryotic expression vector pSV-HA \[[@B55]\]. In a second step, the initially isolated prey cDNA encoding the C-terminal part of Iporin, was digested with *Hind*III and was inserted into the pSV-HA-N-terminal vector. To generate deletion mutants, fragments of Iporin cDNA were amplified with specific primers containing 5\' *Nde*I and 3\' *Eco*RI sites and were cloned into the pAS2-1 or pGADT7 vectors. Generation of rab1b/rab6a chimeras was described previously \[[@B29]\]. All resulting constructs were verified by sequencing. Details are available from A. Barnekow.
Northern blot analysis
----------------------
Multiple tissue Northern Blot (OriGene, Rockville, USA) of poly (A^+^) RNA from various human tissues was hybridized with ^32^P-labeled Iporin ΔN847 cDNA using the RediPrime Nick Translation Kit (Amersham Biosciences, Freiburg, Germany) according to the manufacturer\'s instructions. The X-ray film was exposed for 17 h at -80°C with an amplifying screen.
Antibodies
----------
The polyclonal antibody Aip519 against Iporin was raised in a rabbit immunized with GST-Iporin aa 848--991 (Eurogentech, Seraing, Belgium). The antibody was affinity purified using the antigen coupled to NHS-Sepharose (Amersham Biosciences, Freiburg, Germany). The mouse monoclonal antibody 1E7 against rab1b has been described earlier \[[@B55]\]. Mouse monoclonal antibodies against Calnexin (anti-Calnexin), GM130 (anti-GM130), p115 (anti-p115) and the HA-tag (16B12) were purchased from BD Biosciences (Heidelberg, Germany) and BAbCO (Berkely, USA), respectively. Secondary antibodies coupled to peroxidase were purchased from Amersham Biosciences (Freiburg, Germany).
Expression and purification of recombinant proteins
---------------------------------------------------
GST-Iporin ΔN847ΔC992 fusion protein, used for immunization and GST-Iporin ΔN1430 (encoding the SH3 domain) used for the phage display method (see below) were expressed in *E. coli*strain BL21 induced with 1 mM IPTG for 3 h at 30°C. The pelleted bacteria were washed, resuspended in PBS containing protease inhibitors (complete™, EDTA-free, Roche Diagnostics, Mannheim, Germany) and sonicated 10x for 10 sec. Triton X-100 was added to 1% final concentration and the lysate was incubated on ice for 30 min. After centrifugation at 16000 × g the cleared supernatant was incubated with glutathione-Sepharose beads for 30 min at 4°C. The beads were washed twice with chilled PBS and the fusion protein was eluted with 50 mM Tris-HCl, pH 8.0, containing 10 mM reduced glutathione. Samples were dialyzed against PBS overnight. The recombinant proteins used for GST-pulldown experiments were treated as described above without the elution and dialyzation step. Cleared supernatants were stored at -80°C until use.
In vitro binding assay
----------------------
GST-pulldown assays were performed using bacterially expressed GST fusion proteins containing the deletion mutants Iporin ΔN847ΔC1239, ΔN847ΔC992 and ΔN1430. Equal amounts of the GST fusion proteins or GST were immobilized on 10 μl packed glutathione-Sepharose beads and incubated overnight at 4°C. After the beads were washed three times with LB buffer (10 mM Tris pH7.4, 150 mM NaCl, 1 mM MgCl~2~, 1 mM CaCl~2~, 0.2% Triton X-100, complete™ EDTA-free, Roche Diagnostics, Mannheim, Germany), they were incubated with cytosolic extract from BHK cells, transiently transfected with pSV-HA-rab1b Q67R using the Ca-phosphate/DNA precipitation method \[[@B56]\]. To prepare the cytosolic extract, BHK or HeLa cells were washed three times with chilled PBS and scraped in a volume of 300 μl LB buffer. After homogenization by 15 passes through a 25 gauge needle, lysates were centrifuged at 14000 × g, 4°C for 1 h. The beads containing tubes were incubated for 3 h at 4°C on a rocking platform, supplemented with 50 μl goat serum and were adjusted with LB to a total volume of 300 μl. Beads were recovered by centrifugation, washed three times with 500 μl LB. Proteins were eluted by boiling the beads in SDS sample buffer analyzed by SDS/PAGE and Western blotting. The GM130-pulldown was performed as above with untransfected HeLa cell extracts and without the supplementation of goat serum.
Cell lines
----------
BHK and HeLa cells were grown in Dulbecco\'s modified Eagle\'s medium (DMEM) supplemented with 10% fetal calf serum and 1% glutamine at 37°C in a humidified 5% CO~2~incubator.
Cytosol/membrane fractionation
------------------------------
HeLa cells were grown on 10 cm tissue culture plates and transiently transfected with different pSV-HA-Iporin constructs or pSV-HA-rab1bQ67L using the Ca-phosphate/DNA precipitation method. After 20 hrs, cells were washed three times with chilled PBS, freezed at -80°C, thawed and scraped into 300 μl CMF buffer (250 mM sucrose, 10 mM Hepes/KOH pH 7.4, 1 mM EDTA, complete™ EDTA-free, Roche Diagnostics, Mannheim, Germany). Cells were homogenized by 15 passes through a 25 gauge needle and centrifuged at 1000 × g, 4°C for 2 min. The resulting PNS (postnuclear supernatant) was underlayered with 10 μl of 50% sucrose and centrifuged at 14000 × g and 4°C for 15 min. The supernatant was centrifuged for a second time at 100000 × g and 4°C for 1 h, which provided the S100 supernatant and the P100 pellet fraction. The P100 pellet was washed with 200 μl of CMF and centrifuged once more. Samples were analyzed by SDS/PAGE and Western blotting.
Immunofluorescence analysis
---------------------------
HeLa cells were grown at 40--60% confluency on coverslips and transiently transfected using the Polyfect transfection reagent (Qiagen, Hilden, Germany) according to manufacturer\'s instructions. After 24 hrs, the cells were washed with chilled PBS and fixed for 5 min. in 100% methanol at -20°C. The fixed cells were washed in PBS and blocked for 20 min. in PBS supplemented with 10% (v/v) goat serum. Monoclonal anti-HA tag antibodies (16B12) were diluted in PBS containing 2% (v/v) goat serum and incubated with the cells for 30 min. at RT. After washing with PBS, Alexa Fluor 488-conjugated goat anti-mouse IgGs were applied for 30 min. The cells were washed again with PBS, rinsed in water and mounted in Mowiol 4--88 (Calbiochem, Bad Soden, Germany).
Phage display
-------------
The fUSE5/15-mer M13 phage displayed random peptide library was screened with purified GST-Iporin ΔN1430 fusion protein containing the SH3 domain of Iporin. The used library consists of 2 × 10^8^primary clones and contains a foreign 15-mer oligopeptide on all 5 copies of pIII. The method was performed as described previously \[[@B57]\]. Briefly, 2 μg of purified GST-Iporin ΔN1430 fusion protein was immobilized on the surface of a 96 well plate and incubated with \~10^13^plaque forming units of the phage displayed random peptide library. The bound phages were washed, eluted, amplified and incubated again with newly immobilized GST-Iporin ΔN1430 fusion protein for a second and third round to purify the bound phages. The isolated phages were tested with GST as negative and GST-Iporin ΔN1430 fusion protein as positive controls by the Detection Module Recombinant Phage Antibody System (Amersham Biosciences, Freiburg, Germany) according to the manufacturer\'s instructions. The DNA from positive clones was isolated and sequenced.
Abbreviations
=============
Iporin, interacting protein of rab1; ER, endoplasmatic reticulum; Golgi, Golgi apparatus; GST, glutathione S-transferase; DMEM, Dulbecco\'s modified Eagle\'s medium; SH3, Src homology 3; RUN, RPIP8/UNC-14/NESCA; RabF1-5, rab family region 1--5; RabSF1-4, rab subfamily region 1--4; aa, amino acid; NESCA, new molecule containing SH3 at the carboxy-terminus; RPIP8, Rap2-interacting protein 8; Rabip4, Rab4 interacting protein; Rufy1, RUN and FYVE domain containing protein; SNARE, soluble n-ethylmaleimide-sensitive factor attachment protein receptor; SM, Sec1/Munc18-like proteins, MICAL, a Molecule Interacting with CasL; GM130, Golgi matrix protein with 130 kDa.
Authors\' contributions
=======================
MB: screening, yeast transformations, cloning of Iporin, construction of mutants, GST- pulldown, figure preparation, manuscript preparation
JF: construction of deletion mutant, co-transformation
JK: northern blot analysis, manuscript preparation
EO: GST-pulldown, cytosol/membrane fractionations
TM: phage display
MK: GM130 experiments
TW: conception
AB: conception, design, manuscript preparation, research funds collection
Acknowledgements
================
We thank Drs. I. G. Macara, S. R. Pfeffer and D. Gallwitz for control bait plasmids, the Kazusa DNA Research Institute (Japan) for providing the human KIAA0375 cDNA (Accession No. AB002373.2) and Dr. G. P. Smith for the fUSE5/15-mer M13 phage library. Further we thank M. Koester and M. Rosing for the construction of chimeras and Dr. M. Kail for critically reading the manuscript. This work was supported by the DFG and an FCI grant to A.B.
|
PubMed Central
|
2024-06-05T03:55:55.187646
|
2005-3-29
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079803/",
"journal": "BMC Cell Biol. 2005 Mar 29; 6:15",
"authors": [
{
"first": "Michael",
"last": "Bayer"
},
{
"first": "Julia",
"last": "Fischer"
},
{
"first": "Joachim",
"last": "Kremerskothen"
},
{
"first": "Edith",
"last": "Ossendorf"
},
{
"first": "Theodoros",
"last": "Matanis"
},
{
"first": "Magdalena",
"last": "Konczal"
},
{
"first": "Thomas",
"last": "Weide"
},
{
"first": "Angelika",
"last": "Barnekow"
}
]
}
|
PMC1079804
|
Background
==========
In a living eukaryotic cell, the nuclear envelope is a critical barrier between the nucleus and the cytoplasm, which prevents macromolecules from freely exchanging between the two compartments. The eukaryotic cells have developed multiple pathways designed to transport various types of macromolecules through specialized channels that are incorporated within the nuclear envelope. There are four major components involved in nucleocytoplasmic transport, including cargos, receptors, nuclear pore complex (NPC), and Ran \[[@B1]-[@B3]\]. Cargos are proteins, RNAs or ribonucleoprotein (RNP) particles that contain either nuclear localization signals destined for import into the nucleus from the cytoplasm or nuclear export signals destined for export from the nucleus to the cytoplasm \[[@B4]-[@B6]\].
NXF ([n]{.underline}uclear e[x]{.underline}port [f]{.underline}actor) family members (NXF 1 to 6) play essential roles in RNA exporting out of the nucleus \[[@B7],[@B8]\] through NPC. The structure of the founding member, NXF1 (also called TAP), has modular organization consisting of four major domains: a noncanonical RNP-type RNA-binding domain (RBD), four leucine-rich repeats (LRRs), an internal domain sharing significant sequence homology with the nuclear transport factor 2 (the NTF2-like domain) and a C-terminal ubiquitin associated (UBA)-like domain. The RBD domain and LRRs are responsible for recognition and binding of RNAs through RNP and CTE ([c]{.underline}onstitutive [t]{.underline}ransport [e]{.underline}lement-a cis-acting RNA sequence). The UBA-like domain interacts with multiple nuleoporins of NPC. The NTF2-like domain heterodimerizes with another protein, NXT1 \[[@B9],[@B10]\].
NXT1 was initially identified from human based on its sequence homology to NTF2 (\~26%) \[[@B11],[@B12]\] and association with the RNA exporting factor NXF1/TAP \[[@B13]\]. Earlier studies suggested that NXT1 was necessary for Crm1-mediated nuclear export of proteins \[[@B14]\], in which NXT1 directly binds to export receptor Crm1 and RanGTP (a form of Ran binding GTP), and efficiently stimulates the release of cargos in the cytoplasm after translocation through NPC \[[@B15]\]. However, recent experiments demonstrated that NXT1 appears to exclusively participate, independently of Ran and Crm1, in TAP-mediated nuclear export of RNAs. This finding has been widely confirmed in yeasts, *Drosophila*, *C. elegans*and vertebrates \[[@B7],[@B9],[@B10],[@B16]-[@B18]\]. It was shown that formation of a NXF1/NXT heterodimer via NTF2-like domain is required for the nuclear export of RNAs in a wide range of eukaryotic organisms \[[@B10],[@B19],[@B20]\]. Recently, another member, NXT2 (also called p15-2), was identified to be able to substitute NXT1 in forming NXF1/NXT heterodimers \[[@B7]\]. NXT2 has two alternative splicing variants of transcripts encoding p15-2a and p15-2b proteins that differ in the first exon \[[@B7]\]. To date, the function of NXT2 variants has not been elucidated.
We previously generated cloned zebrafish from long-term cultured fibroblast cells infected with a retrovirus expressing GFP reporter gene \[[@B21]\]. We bred the GFP transgene to homozygosity and found that one of the cloned fish lines bore a recessive embryonic lethal mutation, suggesting that viral insertion is the cause of the cardiac defect. More interestingly, the mutant embryos have reduced circulation, cardiac edema and aberrant heart valve formation after three days of development. Sectioning of homozygous mutant embryos revealed that the ventricular wall is thinner and the space between the myocardium and endocardium is enlarged. Molecular analysis of these mutants revealed that *NXT2*was disrupted by the viral insertion. Whole-mount RNA *in situ*hybridization detected *NXT2*expression in the heart as well as other tissues. RT-PCR analysis suggested that *NXT2*has multiple alternative mRNA splicing forms and intriguingly, one of them is significantly reduced in mutant embryos. Morpholino knockdown of this specific splicing form reproduced the mutant phenotypes. Taken together, our data demonstrate that NXT2 has a critical role in the morphogenetic integrity of the embryonic heart in zebrafish, and provide the first functional study of NXT2 in a vertebrate animal species.
Results
=======
Identification and characterization of a cloned zebrafish exhibiting heart defects
----------------------------------------------------------------------------------
Cloned zebrafish were obtained by nuclear transfer \[[@B21]\] from long-term cultured cells that were infected with retrovirus \[[@B22]\] containing a GFP reporter gene driven by the *Xenopus ef1 alpha*promoter. It has been shown that retroviral integration can cause mutations in zebrafish \[[@B23]\]. We therefore bred all of the cloned fish lines to homozygous for GFP to screen for abnormality. One of these lines exhibited specific heart defects. The mutant phenotype co-segregated with proviral GFP expression (n \> 2000), suggesting that the heart defects were caused by disruption of a functional gene by retroviral insertion. The mutant embryos were first distinguishable from their wild-type siblings by slightly dilated heart, slow heart rate, circulation reduction and mild pericardial edema after 3 days of development (Fig [1A, B, D](#F1){ref-type="fig"} and [1G](#F1){ref-type="fig"}). At this stage, cardiac function assays showed that the atrium and ventricle both contracted rhythmically in a normal atrio-venticular sequence. But, the heartbeat rates of mutants (134.3 ± 11.1 per minute, n = 72) were statistically significantly (P \< 0.01) slower than that of their wild-type siblings (154.7 ± 8.7 per minute, n = 42) and the significant difference remained at 4 days post fertilization (dpf) (mutant 134 ± 12.8 per minute, n = 37 *vs*wild type 169 ± 4.7 per minute, n = 23; P \< 0.01). Furthermore, while the wild-type embryos (Fig [1E](#F1){ref-type="fig"}) develop a narrow heart with forward blood flow through the embryo, both cardiac chambers are dilated in the enlarged pericardial sac (Fig [1C](#F1){ref-type="fig"} and [1H](#F1){ref-type="fig"}) and blood cells regurgitated between ventricle and atrium, resulting in either pooling the blood cells in two heart chambers or peripheral veins of mutant embryos at 4--5 dpf (Fig [1F, H](#F1){ref-type="fig"} and [1I](#F1){ref-type="fig"}). By 5 dpf, the mutant hearts are stretched to a long tubular structure (Fig [1F](#F1){ref-type="fig"}). Pericardial edema became prominent, circulation was completely terminated and severe edema appeared in the whole abdominal cavity, leading to necrosis in fin, muscle, skin and other tissues (data not shown).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Characterization of NXT2 mutant phenotypes. At 3 dpf, NXT2 mutants show pericardial edema (compare B and G to wild-type embryos A and D). By 4 dpf, both atrium and ventricle of NXT2 mutants show significant dilation (compare H to wild-type embryo E) with pericardial sac enlargement (C and H). At 5 dpf, NXT2 mutant heart appears to be lacking atrioventricular boundary (F, arrow) and shows blood accumulation in both cardiac chambers in enlarged heart sac (F) and /or peripheral circulation (I, tail vessel) whereas the wild type heart with a normal atrioventricular boundary (arrow) resides in a narrowed cardiac sac (inset in F). Sections of wild-type (J) and NXT2 mutant (K) hearts at 4 dpf show that ventricular wall of NXT2 mutant embryos is thinner and a space separating myocardium and endocardium in atrium is visible. In addition, we often observed mutant embryos lacking valve structure (arrows). RNA whole mount *in situ*hybridization analyses of myocardial differentiation in wild-type (L, M, N, R, T) and NXT2 mutant (O, P, Q, U, W) at 3 dpf. All ventral view, anterior to the top except S and V, which are lateral view, anterior to the left. Solid red line marks the site of the boundary between the ventricle (v) and the atrium (a). NXT2 mutants show normal expression of cardiac chamber-specific markers *amhc*(L and O), *vmhc*(M and P), *cmlc2*(N and Q). At 3 dpf, *versican*(R and U) and *bmp4*(T and W) are restricted to the atrioventricular boundary in wild-type embryos (black arrows), but, are diffusely expressed in both cardiac chambers in NXT2 mutants. *Versican*expression pattern in otoliths is not changed in mutant embryos (S and V, red arrows: heart). vv: veinous vessels; a: atrium; v: ventricle; av: atrioventricular boundary; ot: otoliths. wt: wild type; mutant: *zNXT2*homozygous mutant embryos; dpf: days post fertilization.
:::
:::
The reduced circulation and regurgitation defects may be the results of abnormalities in cardiac contractibility, differentiation of cardiomyocytes or the formation of cardiac valve. To further investigate these possibilities, we analyzed histological sections of wild type and mutant embryos. Normally, by 2 dpf, the ventricular myocardial cells thicken and become the primary pumping force of the heart. We found that in 4 dpf mutant embryos, the ventricular wall remains thin and the space between myocardium and endocardium was diminished in the ventricle but enlarged in the atrium compared to that of wild type embryos (Fig [1J](#F1){ref-type="fig"} and [1K](#F1){ref-type="fig"}). In addition, we frequently observed mutant samples lacking cardiac valve structure. In 5 dpf mutant embryos, the ventricular wall remained thinner and cardiac chambers were severely dilated (data not shown). These data suggested that myocardial structures were disturbed in mutant embryos during early development.
Various heart chamber-specific gene expressions demarcate vertebrate heart compartments and play an important role in regulating critical steps of chamber formation. To understand whether the identified mutation affects formation of specific chambers and morphogenetic movements of dynamic cardiomyocycte, expression of chamber-specific genes including *Vmhc, Amhc*and *Cmlc2*was examined. As shown in Fig [1L--Q](#F1){ref-type="fig"}, no difference was detected in expressions of *Amhc*(L and O), *Vmhc*(M and P) and *Cmlc2*(N and Q) between wild-type (L, M and N) and mutant (O, P and Q) embryos at 3 dpf (same results were obtained at 4 and 5 dpf, data not shown), suggesting that the mutated gene is not required for the initial chamber-specific differentiation of cardiomyocytes, in which *Amhc, Cmlc2*and *Vmhc*play important roles \[[@B24]-[@B27]\].
Epithelial-to-mesenchymal transition (EMT) plays a critical role in atrio-ventricular valve formation. In the zebrafish, this process begins at 33 hpf, when the myocardial cells at the atrio-ventricular boundary induce the adjacent endocardial cells to undergo EMT\[[@B28],[@B29]\]. *Bmp4*and *versican (cspg2)*are genes involved in cardiac valve formation. Coincidently, their cardiac expressions become restricted to the myocardial cells at the valve forming region at the same time as the formation of the prevalvular cushions, suggesting that they are thereby markers for the differentiation of the myocardial cells in the valve forming region and misexpression of these genes may indicate defects in valve formation\[[@B27],[@B28]\]. As shown in Figure [1](#F1){ref-type="fig"} (R-W), expression of *versican*and *bmp4*in mutant embryos (Fig [1U, 1V](#F1){ref-type="fig"}, and [1W](#F1){ref-type="fig"}) was broadly up-regulated and expanded throughout the two heart chambers at 3 dpf at which stage their expression should be restricted to the atrioventricular boundary as in wild-type embryos (Fig [1R, S](#F1){ref-type="fig"} and [1T](#F1){ref-type="fig"}). Meanwhile, no difference of *versican*expression (Fig [1S](#F1){ref-type="fig"} and [1V](#F1){ref-type="fig"}) was detected in the otoliths of developing ears in wild-type and mutant embryos \[[@B30],[@B31]\], suggesting that the abnormal *versican*expression pattern is due to heart defects observed in these mutant embryos other than general developmental delay.
NXT2 is responsible for heart defects
-------------------------------------
Of the 2000 analyzed mutated embryos, the heart defect always co-segregate with the proviral GFP expression, suggesting that the viral insertion causes the observed heart phenotypes. By analyzing the genomic DNA by Southern blot, we found that there is a single viral insertion in this fish line (Fig [2A](#F2){ref-type="fig"} and [2B](#F2){ref-type="fig"}). A 1071 bp genomic sequence flanking the viral insertion was isolated by inverse PCR and used to search EST databases, leading to the identification of a candidate gene encoding a 143 amino acid protein that shares high homology with that of NXT1 and/or NXT2 from human, mouse, rat, *Drosophila*and *C. elegans*. By comparing genomic structures of *NXT1*and *NXT2*genes from these species, we noticed that *NXT1*was encoded by a single exon whereas *NXT2*gene contains multiple exons, which is similar to our candidate gene (Fig [2C](#F2){ref-type="fig"}). Based on similarities in protein sequences (Fig [2D](#F2){ref-type="fig"}) and genomic organizations, we concluded that our candidate gene represents a homolog of *NXT2*rather than *NXT1*and named it *zNXT2*.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Identification of *NXT2*as a candidate gene responsible for the heart defect. Southern blot analysis shows that a 4.3 kb (digested by Hpa I; A, lane 2) and a 3.8 kb (digested by Dra I; A, lane 3) are detectable using a specific viral DNA probe (B). Lanes 1 and 4 (A) are positive and negative control, respectively. DNA fragments corresponding to 3.8 kb / Dra I and 4.3 kb / Hpa I restriction were gel-purified and used for inverse PCR to isolate junction genomic sequences with primers described in (B) and Methods. This resulted in identification of NXT2 gene, which consists of five exons (C) and encodes 143 putative amino acids (D). The provirus is inserted within the first intron (red arrow) upstream of the putative initiation codon ATG in exon 2 (C). Alignments of protein sequences from various organisms revealed that zebrafish NXT2 shares 73.9%, 72.5%, 71.6%, 70.2%, 70.9%, 68.8%, 43.6% and 34.6% amino acid homology with that of rat NXT2, human NXT2, human NXT1, rat NXT1, mouse NXT1, *Xenopus*NXT1, *Drosophila*NXT1 and *C. elegans*NXT1, respectively (D). Identical residues are shown in red boxes. Zebrafish *NXT2*is ubiquitously expressed at 2 dpf as shown by RNA whole mount *in situ*hybridization (E, arrow points to heart). E: lateral view, anterior to the left. zNXT2: zebrafish NXT2; hNXT1: human NXT1; hNXT2: human NXT2; mNXT1: mouse NXT1; rNXT1: rat NXT1; rNXT2: rat NXT2; cNXT1: *C. elegans*NXT1; dNXT1: *Drosophila*NXT1 and xNXT1: *Xenopus*NXT1.
:::
:::
To determine the expression pattern of *zNXT2*, we carried out RNA whole mount *in situ*hybridization with its sense and antisense RNA probes. It was found that *zNXT2*was ubiquitously expressed during early embryogenesis but clearly detectable in heart at 2 dpf (Fig [2E](#F2){ref-type="fig"}).
zNXT2a is the alternatively spliced transcript required for heart function
--------------------------------------------------------------------------
Search for zNXT2 homologues revealed that one copy of *NXT*gene exists in *Drosophila, C. elegans*and mouse whereas two copies are present in rat (rNXT1 and rNXT2) and human (hNXT1 and hNXT2) genomes, respectively (Fig [3A](#F3){ref-type="fig"}). In humans, two alternative splicing variants of *NXT2*(*NXT2a*and *NXT2b*) were identified in EST database, which were further confirmed by cloning and sequencing the corresponding cDNAs \[[@B7]\]. In order to determine if *zNXT2*also has multiple alternative RNA transcripts, 5\' RACE analysis was performed \[[@B32]\] and the PCR products were sequenced. From 19 sequences obtained, 6 represented the original *zNXT2*encoding 143 amino acid protein as mentioned above (now named as *zNXT2a*), 2 lacked the third exon, and 11 did not have the first two exons and part of third exon. The last variant, now named *zNXT2b*, appeared to be transcribed from the third exon and started at the putative codon GTG (Fig [3A](#F3){ref-type="fig"}). The same observation has also been made in humans in that the second splicing variant *NXT2b*starts from the putative initiation codon GTG \[[@B7]\]. The putative zNXT2b protein contains most of the functional domains of zNXT2a, since they share the last two identical exons.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Alternative splice variants of the zebrafish *NXT2*. **A:**Solid black boxes represent exons, lines represent introns. Initiation codons and stop codons are indicated in capitalized letters. Solid blue box represents an exon of human NXT2 splice variant NXT2b, with GTG as initiation codon. The zebrafish NXT2 splice variant NXT2b is indicated below zebrafish NXT2 with GTG as initial codon. Primers NTf/NTr and NXTf/NXTrm (horizontal arrows) were used for PCR amplification to identify splicing variants of zNXT2. chr.: chromosome or linkage group. **B:**RNA expression profile of zebrafish *NXT2*splice variants by RT-PCR. Equal amounts (2 μg) of total RNA, isolated from wild-type embryos (lanes 1, 4, 7), NXT2 mutant embryos (lanes 3, 6, 9) and their heterozygous siblings (lanes 2, 5, 8), were used for RT-PCR analyses. Lanes 1--3 refer to PCR products of endogenous eflα gene as positive control, lanes 4--6 represent products of primers NXTf/NXTrm, and lanes 7--9 indicate PCR products of primers NTf/NTr. RNA transcription level of *zNXT2*, represented by NTf/NTr in lane 9 (indicated by an asterisk), is remarkably reduced in mutant embryos, compared with that of their heterozygous and wild-type sibling embryos in lanes 8 and 7. No significant difference is detected among different genotypic embryos by NXTf/NXTrm in lanes 4, 5 and 6. -/-: *zNXT2*homozygous mutant; +/-: heterozygous sibling;+/+: wild type sibling; ef1 alpha, NXT2a and NXT2b: transcripts detected by appropriate primers.
:::
:::
To determine which transcript is responsible for the mutant phenotype we compared *zNXT2a*and *zNXT2b*mRNA expression level in mutant and wild type embryos. Two sets of primers, NTf/NTr for the first and the fourth exon to detect *zNXT2a*and NXTf/NXTmr for the fourth and the fifth exon to detect both *zNXT2a*and *zNXT2b*, were designed for RT-PCR analysis (Fig [3](#F3){ref-type="fig"}). As shown in Fig [3B](#F3){ref-type="fig"}, *zNXT2a*expression level was dramatically reduced in homozygous mutant embryos (indicated by an asterisk) compared to their wild-type or heterozygous siblings whereas no difference of mRNA levels was observed when both *zNXT2a*and *zNXT2b*were detected using NXTf/NXTmr primers. Taken together, these data suggested that loss of *zNXT2a*transcripts might be responsible for this mutant phenotype.
Blocking zNXT2a activity by morphlino phenocopies the mutant heart defects
--------------------------------------------------------------------------
Morpholino technology is a powerful tool for blocking translation of specific mRNA transcripts and has been extensively used in zebrafish \[[@B33]-[@B35]\]. To further verify that zNXT2a was responsible for the observed heart defects, an antisense morpholino oligo specific to this transcript was injected in wild type embryos at one to four-cell stage. Analysis of more than 500 embryos from multiple injected clutches showed that approximately 80% of injected embryos developed the same heart defects as that in the virus-induced mutant embryos (Fig [4A--H](#F4){ref-type="fig"}), including pericardial edema, slightly dilated heart, slow heart rate and reduced circulation. Mock-injections have no effect on heart development in control embryos (data not shown).
::: {#F4 .fig}
Figure 4
::: {.caption}
######
Morpholino antisense knockdown of *zNXT2*in transgenic zebrafish embryos expressing cmlc2-GFP and flk-RFP. zNXT2 knockdown causes pericardial edema (compare A, C, E, G with B, D, F, H), abnormal relative positions of two chambers and chamber dilation (compare insets in C-H showing CMLC2 promoter driven GFP expression patterns in myocardium of atriums and ventricles). Observation of endothelial cells at the atrioventricular boundary in living *flk-1*- RFP transgenic embryos allowing visualization of cardiac valve formation (I and J). Clustering of endothelial cells was clearly visible in the cardiac valve region in wild-type embryos (I, arrow) whereas injection of *zNXT2*morpholino caused a failure of clustering of endocardial cells at the atrioventricular boundary (J). All embryos are lateral view, anterior to the left. a: atrium; v: ventricle; OT: outflow tract; A, C, E, G: uninjected wild-type embryos; B, D, F, H: morpholino-injected embryos. wt: wild type; dpf: days post fertilization; MO: zNXT2 morpholino.
:::
:::
The ability to use morpholinos to reproduce *zNXT2a*mutant phenotypes allowed us to confirm cardiac defects observed in living mutant embryos (Fig [1](#F1){ref-type="fig"}) through using live transgenic zebrafish embryos expressing GFP in specific cell lineages. First, cardiac myosin light chain 2 (CMLC2) promoter driving GFP transgenic zebrafish \[[@B26]\] embryos were used to determine development of two heart chambers. As shown in Fig [4](#F4){ref-type="fig"}, heart chambers of morpholino injected embryos at 3 dpf exhibited pericardial edema and dilated heart compared with that of wild-type embryos (Fig [4A, B, C](#F4){ref-type="fig"} and [4D](#F4){ref-type="fig"}). In 4--5 dpf morpholino injected embryos, cardiac chambers became abnormally stretched and relative positions of two chambers were abnormal (Fig [4F](#F4){ref-type="fig"} and [4H](#F4){ref-type="fig"}) whereas wild-type embryos had heart chambers in a narrowed pericardial sac (Fig [4E](#F4){ref-type="fig"} and [4G](#F4){ref-type="fig"}). This finding is consistent with the previous results described based on viral insertional mutated embryos (Fig [1](#F1){ref-type="fig"}).
We further examined endocardial morphology at the atrioventricular valve region using receptor tyrosine kinase *flk-1*promoter driving RFP transgenic zebrafish embryos. Previously, tie-2 GFP transgenic zebrafish has been used to address this issue. Similar to *tie-2*, *flk-1*also plays a crucial and unique role in endothelial cell differentiation and vascular morphogenesis, including vessels and heart endocardium \[[@B36],[@B37]\]. In uninjected wild-type embryos, endocardial cells cluster at the atrioventrivular boundary and form a ring shape structure at the valve forming region (Fig [4I](#F4){ref-type="fig"}). However, more than 70% of the *zNXT2a*morpholino injected *flk-1*-RFP transgenic embryos (n = 129) failed to form this cluster (Fig [4J](#F4){ref-type="fig"}), suggesting that *zNXT2*is required for this early endocardial morphogenetic event. Together with RNA whole mount in situ data of *bmp4*and *versican*, these results indicate that cardiomyocyte differentiation and thickening of the endocardium at the atrioventricular boundary in the mutant embryos were significantly impaired, resulting in abnormal valve formation.
Discussion
==========
Ventricular myocardial thickening and atrioventricular valve are impaired by disruption of zNXT2a
-------------------------------------------------------------------------------------------------
In zebrafish, between 24 and 48 hpf, the linear heart tube gradually bends at the boundary of two chambers to create an S-shaped loop. Meanwhile, clustering of endocardial cells as a ring structure at the atrioventrocular boundary by 33--37 hpf initiates the first step of the formation of cardiac valve \[[@B27],[@B28],[@B31]\]. While the heart is pumping, differentiation and morphogenesis continue, which includes remodeling of the cardiac valve to completion, finally building on a linear heart tube to produce an adult heart.
Sectioning of *zNXT2a*homozygous mutant embryos revealed that the ventricular wall is thinner as early as 3 dpf (data not shown, but see Fig [1](#F1){ref-type="fig"} for 4 dpf data). Decreased cell proliferation and/or increased apoptosis may cause such a defect. Since the thinner myocardial wall appears very early during development we prefer a model that the under-developed ventricular wall is due to a deficiency in myocardial cell proliferation. Alternatively, this could result from abnormally tight intercalation of ventricular cells with the same numbers of cells in the mutant embryos as previously described \[[@B38]\]. In regard to the atrioventricular valve defect, it has been reported that endothelial shear stress (blood flow) plays a critical role in endocardial cushion by using special surgical experiments in zebrafish embryos \[[@B39]\]. In contrast, other researchers \[[@B40]\] argued that reduction in myocardial function is primarily responsible for the defect in atrioventricular cushion development, which was observed in *sih*(no early heartbeat) and *cfk*(cardiac dilation and no early blood flow) mutant embryos. In *zNXT2*mutant embryos, the earliest observable defects are reduced heart rate, dilated cardiac chambers and a thin ventricular myocardial wall. Heart morphology and chamber specification ascertained by expression of *Amhc, Vmhc*and *Cmlc2*are all normal at this stage in mutant embryos (Fig [1](#F1){ref-type="fig"} and [4](#F4){ref-type="fig"}, data not shown by 2 dpf). Later on from 3--4 dpf, the atrioventricular valve was aberrant and the relative position of atrium and ventricle was perturbed, as revealed by sectioning analysis, expression of *versican*and *bmp4*(Fig [1](#F1){ref-type="fig"}) and morpholino knockdown in CMLC2-GFP and flk-GFP transgenic embryos (Fig [4](#F4){ref-type="fig"}), respectively. Additionally, injection of a higher dose (approximately 10 ng per embryo) of *zNXT2a*morpholino into embryos causes defective endocardial cushion formation as early as 2 dpf, as illustrated by a failure of endocardial clustering at atrioventricular boundary (data not shown). This indicates that valve malformation resulted by interruption of *zNXT2a*might be either a primary defect or a defect due to disturbed function of chambers at later stage. In addition, our data implies that the mutant cardiac valve malformation is either an independent event or caused by lacking of myocardial cell proliferation or differentiation. The later situation will be consistent with the observation by Bartman et al (2004). Nonetheless, our findings establish that zNXT2 has a critical role in driving cardiac valve formation to final completion.
zNXT2a is the mRNA transcript variant responsible for heart defects
-------------------------------------------------------------------
As in humans, *zNXT2*has multiple alternative splicing mRNA transcripts with two distinct start sites (ATG and GTG) (Fig [3A](#F3){ref-type="fig"}). It is possible that different *zNXT2*splicing variants are differentially expressed in different tissues. Moreover, *zNXT2b*starting from GTG may represent a major transcript because of its higher copy numbers in the amplified cDNA pools (11 out of 19 sequenced samples) whereas *zNXT2a*starting from ATG might be enriched in heart. The biological significance of such a variation in *NXT2*transcription has not been addressed. Our studies demonstrated that the alternatively spliced transcripts might have specific functions in different tissues during embryonic development and organogenesis. Specifically, *zNXT2a*appears to be required for heart development in zebrafish. Viral insertion in the intron immediately upstream of the exon containing ATG caused specific knockdown of *zNXT2a*transcription in mutant embryos whereas other transcripts such as *zNXT2b*appear normal. Injection of a morpholino against *zNXT2a*can phenocopy heart defects, confirming that zNXT2a protein is the main factor required for zebrafish heart development. Furthermore, when the morpholino dose was increased, more injected embryos developed pericardial edema one day earlier than the insertional mutants of *zNXT2*(data not shown). This suggests that our *zNXT2a*insertional mutant may be a hypomorphic allele due to insertion of the viral vector into a non-coding region of the locus. With morpholino data it is clear that *zNXT2a*is directly implicated in heart development that requires its sufficient expression activity.
Nuclear transport factors are implicated in tissue-specific biological functions
--------------------------------------------------------------------------------
Nuclear protein or RNA transporting has previously been regarded as a general house keeping function for all biological systems. However, increasing evidence suggest that this can be a highly regulated process for specific biological functions with specific RNA targets. In *Drosophila*, studies demonstrated that homolog of NTF2 (DNTF2) is required for nuclear translocation of the Rel proteins Dorsal, Dif and Relish into the nucleus. These three Rel proteins are activated in response to immune challenges and imported into the nucleus upon microbial challenges, which further induce and regulate transcriptions of anti-microbial peptides attacin, cecropin, defensin, metchnikowin, diptericin and drosomycin. Disruption of DNTF2 leads to impairment of nuclear import of the Rel proteins and thus immune response is impaired in NTF mutants compared with wild type \[[@B41],[@B42]\]. NXF5, a member of the nuclear export factor NXF family, binds to RNAs as well as to NXT members for RNA nuclear export. In a male patient with a syndromic form of mental retardation, a nonfunctional NXF5 was identified, which is split by the breakpoint situated in the 5\' UTR between exons 1 and 2 of *NXF5*. This suggests that interruption of the gene NXF5 might result in the disease phenotypes associated with syndromic mental retardation including short stature, general muscle wasting, and facial dysmorphism by deranging the export or transport of specific mRNAs \[[@B43],[@B44]\].
Formation of an NXT1 (or NXT2) and TAP (NXF1) heterodimer is critically required for mRNA export from the nucleus via TAP/NXT1 pathway \[[@B9],[@B10],[@B13]\]. Earlier experiments demonstrated that NXT1 is essential in Crm1-mediated nuclear export of proteins, in which NXT1 directly binds to export receptor Crm1 and RanGTP, and improves the release of cargos into the cytoplasm after translocation through the NPC \[[@B15]\]. Very recently, it has been reported that CRM1 is involved in export of specific subsets of cellular RNAs, which include *tra-2*mRNA participating in sex cell fate determination in *C. elegans*and *IFN-a1*mRNA generated upon viral infection to human cells \[[@B45],[@B46]\]. However, little is known about how their cofactor NXT1 (or NXT2) plays a role in a biological pathway. Our finding provides the first evidence that an alternative splicing form of nuclear transport factor NXT2 is specifically implicated in heart development of zebrafish. For future studies, identification of specific RNA targets and elucidation of the molecular mechanism underlying NXT2-mediated exporting of these targets, either mRNA or micro RNA, in heart should be the focus. In addition, this study might provide the basis for some inheritable cardiac valve diseases in human. It should be a reasonable step to determine genetic linkage relationship between a NXT2 locus and defined cardiac diseases.
Conclusion
==========
We demonstrated that disruption of *NXT2*gene, a member of NXT (nuclear export factor) family, is responsible for heart-specific defects in a mutant zebrafish generated by a retroviral insertion introduced into the genome by animal cloning technology. Our study suggests that NXT2 has critical roles required for myocardial cell differentiation and formation of cardiac valve at the atrioventricular boundary. Since NXT2 is highly conserved between fish and man it is conceivable that it may also play a role in mammalian heart development and might be involved in cardiac diseases.
Methods
=======
Mutant identification and characterization
------------------------------------------
We initially obtained 14 cloned zebrafish lines by nuclear transfer \[[@B21]\]. After breeding each line to homozygosity for the retroviral insertion that was introduced into the zebrafish donor nuclei as a traceable GFP marker, we identified one line with embryonic heart defects that co-segregate with the GFP expression of the viral construct.
Adult zebrafish and embryos were maintained and staged as previously described \[[@B47]\]. Homozygous mutant embryos were imaged using an Axiocam imaging system (Zeiss). For histological analysis, fixed embryos were dehydrated, embedded in plastic (JB-4, Polysciences Inc.), sectioned at 8 μm and stained with Toludine Blue.
Identification of NXT2 as the candidate gene
--------------------------------------------
Two rounds of inverse PCR were performed to isolate genomic DNA fragments immediately flanking the viral insertion based on the retroviral sequence \[[@B48]\]. For the first inverse PCR, genomic DNA from heterozygous mutant fish was restricted with Dra I. This generated an approximately 3.8 kb proviral-genome-containing genomic DNA fragment as revealed by Southern blot hybridization with a specific viral probe (Fig [2A](#F2){ref-type="fig"} and [2B](#F2){ref-type="fig"}). Using Geneclean III kit (Bio 101, Vista, California), a pool of genomic DNA containing this fragment was purified from the gel, then ligated into circular form and finally used as inverse PCR templates. Two sets of primers specific to viral -sequences were designed as follows:
HT1:5\'-AATCCCGGACGAGCCCCCAAATGAAAGA-3\' and HT3: 5\'-ATAGAGTACGAGCCATAGTTAAAATAAA-3\' (outer primers);
HP2: 5\'-TTTCTTTGTTCCTGACCTTGATC-3\' and 1 HP: 5\'-AACCCCTCACTCGGCGCGCCAG-3\' (inner primers). Templates were amplified for two rounds. The final PCR products were subcloned into the pCR2.1 vector (Invitrogen, Life Technologies) for sequencing. The obtained sequence was used for designing new primers for additional inverse PCR. Similarly, heterozygous genomic DNA was also digested with Hpa I, which generated a 4.3 kb fragment containing retroviral genome and a flanking sequence for PCR amplifications. Again, two rounds of amplification were carried out for the second inverse PCR, in which a set of primers (HP2: 5\'-TTTCTTTGTTCCTGACCTTGATC-3\'; 2 HP: 5\'- TCGTGGTCTCGCTGTTCCTTGG-3\') against the retroviral LTR region was used to enrich for DNA templates. Then the second round of PCR amplification was performed from the enriched templates with another set of primers HP2 and F2 (5\'-GTGCATGGATAAGAAAAGACGGGTAACG-3\', which is against the zebrafish genomic sequence flanking the proviral insertional locus) (Fig [2A](#F2){ref-type="fig"} and [2B](#F2){ref-type="fig"}). A total of 1071 bp genomic sequence was obtained and used to search for EST and genomic databases of zebrafish.
Sequence analysis
-----------------
To identify the mRNA coding sequence, the 1071 bp genomic flanking sequence was used to BLAST search against the zebrafish EST database. An EST (BI325953) was identified that shares identical sequence to the inverse PCR product and contains an open reading frame of 143 amino acids. Highly similar protein sequences from different species including human, mouse, rat, *Drosophila*and *C. elegans*were retrieved from cDNA databases by performing protein sequence similarity searches using the deduced zebrafish amino acid sequence. Multiple sequence alignments were constructed by using the Jotun Hein Algorithm Method (DNA Star Software). To predict the genomic structures of *NXT*homologues in various organisms, the BLASTN program for genomic sequences was employed based on cDNA and /or EST sequences.
Morpholino knockdown
--------------------
A morpholino was designed against ATG region of zebrafish *NXT2*(5\'-TCCACCGTTAAAGACATGACTGGTC-3\', Gene Tools, LLC, Corrallis, OR). It was injected into one-cell to four-cell stage embryos at 3 μg/μl (2 nl per embryo) in 1 X Danieau solution as described by other researchers \[[@B35]\]. Different stages of injected embryos were imaged under a regular bright filter on a Zeiss microscope. Mock-injections using a morpholino targeting a pancreas-specific gene \[[@B49]\] or a morpholino with a random sequence at 4--6 ng per embryo was performed as controls.
Identification of alternative splicing variants of zebrafish NXT2
-----------------------------------------------------------------
To identify alternative splicing transcripts of NXT2, FirstChoice RLM-RACE Kit (Catalog \#1700, Ambion) was employed to specifically amplify the 5\' intact region of *zNXT2*mRNA through its first strand cDNA using two pairs of 5\' adaptor primers (outer primer: 5\'-GCTGATGGCGATGAATGAACACTG-3\'; inner primer: 5\'-CGCGGATCCGAACACTGCGTTTGCTGGCTTTGATG-3\') and 3\' specific primers (NXTmr -- outer primer: 5\'- AAGCTCAGTTTGCCCAGTCT-3\'; NTr -- inner primer: 5\'-TGGCAGTCAAGAGTTTGCAC-3\') against the internal exons of the gene. RLM-RACE strategy is designed to amplify cDNA only from full-length, capped mRNA, usually producing a single band for a specific gene after PCR \[[@B32]\]. Total RNA isolated from 3-day wild-type embryos was used for 5\' RACE experiments.
Detection of mRNA transcripts by RT-PCR
---------------------------------------
To compare relative expression levels of *NXT2*transcripts, total RNA samples were isolated from homozygous mutant embryos (-/-), their wild-type (+/+) and GFP positive normal (+/-) sibling embryos, respectively, which were sorted out based on GFP expression and heart phenotypes. For measuring mRNA levels, equal amounts of total RNA isolated from three types of embryos were subjected to first strand cDNA synthesis by reverse transcription under the same condition using M-MLV Reverse Transcriptase (Gibco, Invitrogen). And then, equal amounts of cDNA product were used for PCR amplification of targeted cDNA. Amplification of another cDNA, *ef1 alpha*, was used as a control. The following specific primers against *zNXT2*coding sequences were designed across introns for this purpose: NTf, 5\'-GACGGATGCTGGATTTGTTT-3\'and NTr, 5\'-TGGCAGTCAAGAGTTTGCAC-3\'; NXTf, 5\'-GAAACGCAGTCACTGGTCAA-3\' and NXTrm, 5\'-AAGCTCAGTTTGCCCAGTCT-3\'. *ef1 alpha*primers were zEF1a-1: 5\'-TCACCCTGGGAGTGAAACAGC-3\' and zEF1a-2: 5\'-ACTTGCAGGCGATGTGAGCAG-3\'. Different cycle numbers (20, 25, and 29) of quantitative PCR amplification were employed to determine a threshold that produced detectable products on gel using the following program: 94°C, 3 minutes; 94°C 30 seconds, 60°C 40 seconds, 72°C 1.5 minutes; 72°C 10 minutes.
Whole-mount RNA in situ hybridization
-------------------------------------
Sense and antisense digoxigenin-labeled RNA probes were generated from cDNA clones of the zebrafish *NXT2, amhc, vmhc, versican, cmlc2*and *bmp4*genes using a DIG/Genius 4 RNA Labeling kit (Boehringer Mannheim). Whole-mount RNA *in situ*hybridization conditions were as described by Jowett \[[@B50]\]. Images were obtained using an AxioCam digital camera (Zeiss) on Stemi SV11 Apo (Zeiss) dissection microscope.
Authors\' contributions
=======================
HH contributed to all aspects of the experimental data and drafting the manuscript. BZ performed some RNA in situ hybridization experiments. PAH carried out sectioning of mutants. JC provided discussions and edited the manuscript. SL designed the study, drafted and finalized the manuscript. All authors read and approved the final manuscript.
Acknowledgements
================
We thank Xiaodong Shu for helpful discussion and are grateful to Huidong Pan for taking care of our fish. This work was supported by grants from NIH to SL (R01 DK65638) and to Chen (R01HD41367).
|
PubMed Central
|
2024-06-05T03:55:55.192028
|
2005-3-24
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079804/",
"journal": "BMC Dev Biol. 2005 Mar 24; 5:7",
"authors": [
{
"first": "Haigen",
"last": "Huang"
},
{
"first": "Bo",
"last": "Zhang"
},
{
"first": "Parvana A",
"last": "Hartenstein"
},
{
"first": "Jau-nian",
"last": "Chen"
},
{
"first": "Shuo",
"last": "Lin"
}
]
}
|
PMC1079805
|
Background
==========
Acne vulgaris is a common, chronic inflammatory disease of pilosebaceous follicles. Several factors have been suggested to influence acne including diet, menstruation, sweating, UV radiation, stress, and occupation \[[@B1]\]. Although some studies have shown that cigarette smoking aggravate acne \[[@B2],[@B3]\], others did not confirm this association \[[@B4]\], or even showed a protective effect \[[@B5]\]. Therefore this study was undertaken to evaluate the relationship between acne vulgaris and cigarette smoking.
Methods
=======
Three hundred and fifty patients with acne vulgaris and a similar number of patients with skin diseases other than acne, aged 15 -- 40 years, attending in a skin clinic in Tehran, Iran were asked to complete a questionnaire on smoking habits anonymously in the waiting room. Acne was diagnosed on a clinical basis by the presence of either of comedones, papules, pustules, nodules or cysts but those with only acne scars and no active lesions were not included.
Current smoking was defined as smoking at least one cigarette in a week. All data were analyzed with the SPSS for windows (release 11.5).
Results
=======
Two hundred and ninety-three patients with acne and 301 patients with other skin diseases completed the questionnaires (response rates of 83.7% and 86.0%, respectively). Twelve acne patients (4.1%) and 27 controls (9.0%) were current smokers (odds ratio: 0.43, 95% CI: 0.22 -- 0.87, p \< 0.05). But after adjustment for sex, this difference was not significant (odds ratio: 0.61, 95% CI: 0.30--1.26, p \> 0.05, Mantel-Haenszel test). Assuming a 2-sided significance level of 0.05, this study had a power of 77.8% to detect an odds ratio of 2.
The mean age of the patients in the acne group was 22.08 years (SD = 4.60) and in the control group was 26.09 (SD = 6.52) years (t test, p \> 0.05). There were 9 smokers among 64 males with acne and 20 smokers among 116 without it (odds ratio: 0.79, 95% CI: 0.33--1.84, p \> 0.05). Also there were 3 smokers among 229 females with acne and 7 smokers among 185 without it (odds ratio: 0.34, 95% CI: 0.09--1.32, p \> 0.05). Also the amount and duration of smoking were not significantly different between two groups (table [1](#T1){ref-type="table"}, Fisher\'s exact test: p \> 0.05).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
The amount and duration of smoking in acne and control patients
:::
Smoking habit Acne patients (12) Control patients (27)
--------------- -------------------------- ----------------------- ----
Amount 1 cigarette per week 2 5
1--6 cigarettes per week 2 9
\> 6 cigarettes per week 8 12
Not mentioned 0 1
Duration \< 3 months 1 2
3 months -- 3 years 6 7
\> 3 years 5 16
Not mentioned 0 2
:::
Discussion
==========
We could not find any association between acne and smoking in this study. Although patients with acne were less likely to smoke, but this association was not present after adjustment for sex. The patients included in this study had a wide age range (i.e., 15 to 40 years). As a consequence we could not separate factors associated with acne onset from changes in personal habits occurred during the course of the disease.
Mills for the first time studied the association between acne and smoking \[[@B5]\]. He studied 156 patients with acne through a questionnaire on smoking habits and found that 19.7% of 96 male and 12.1% of female patients were smokers, which was significantly less than national statistics (34.5% and 32.7%, respectively). So he suggested that some component of cigarette, possibly nicotine, has an anti-inflammatory action on acne. We compared the acne patients with patients with other skin diseases attending the same clinic in the same time period, and believe that they are more suitable as a control group that national statistics. Also Mills studied patients with severe acne under treatment with isotretinoin, whereas this study included patients with all forms of acne.
Jemec et al \[[@B4]\] studied the prevalence of acne and the smoking habits of a random sample of 186 15- to 22-year-old subjects and found that 40.7% of men and 23.8% of women had clinical acne but smoking was not significantly associated with acne (odds ratio: 0.54, 95% confidence interval: 0.17--1.78). Although we used a different approach, a case-control design, reached a similar conclusion.
On the other hand, Schafer et al \[[@B2]\] examined 896 persons, aged 1--87 years, in a cross-sectional study and found the overall prevalence of acne was 26.8%. Acne prevalence was significantly higher in active smokers as compared with non-smokers (40.8% vs. 25.2%, odds ratio: 2.04, 95% confidence interval: 1.40--2.99). This study has a wide age range and if we consider the patients with the age range of our study (15--40 years), no association between acne and smoking could be found (54/102 smokers as compared with 93/184 non-smokers had acne, odds ratio:1.10, 95% confidence interval: 0.68--1.79, p \> 0.05).
Smoking has been associated with several adverse effects on the skin \[[@B6]\]. But protective effects of smoking against certain inflammatory diseases, such as ulcerative colitis \[[@B7]\], acne rosacea \[[@B8]\], pemphigus vulgaris \[[@B9]\], and recurrent aphthous stomatitis \[[@B10]\] have also been reported. Available data do not support any association between acne vulgaris and smoking.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
AF participated in the design and conduct of the study and preparation of the manuscript.
RS participated in the design and statistical analysis of the study.
SMD participated in the design and conduct of the study.
MNK participated in the design and conduct of the study.
All authors read and approved the manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-5945/5/2/prepub>
|
PubMed Central
|
2024-06-05T03:55:55.195151
|
2005-3-24
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079805/",
"journal": "BMC Dermatol. 2005 Mar 24; 5:2",
"authors": [
{
"first": "Alireza",
"last": "Firooz"
},
{
"first": "Reza",
"last": "Sarhangnejad"
},
{
"first": "Seyyed Massoud",
"last": "Davoudi"
},
{
"first": "Mansour",
"last": "Nassiri-Kashani"
}
]
}
|
PMC1079806
|
Background
==========
There is growing evidence that long term exposure to the types of unfavorable working conditions that have been observed among some VDTs users might have serious health consequences \[[@B1]\].
Dermatological manifestations, especially on the face are one of these health outcomes \[[@B2]\].
The most common manifestations among these patients are nonspecific erythema, acne rosacea, seborrheic dermatitis, pruritus, burning sensation, and dry skin \[[@B3]\].
The prevalence of these manifestations among VDTs users ranges from 8--10% in a series of descriptive studies to 13.5% in other reports \[[@B4]\].
Radiation emissions from VDTs such as x-ray, ultraviolet, infrared are within acceptable levels and there is neither a connection between these radiations and health consequences, nor with any dermatological manifestations \[[@B4],[@B5]\].
As a whole, the exact cause of these facial manifestations of VDTs users is not clear, but physical conditions of the workplace such as dryness, occupational stress \[[@B6]\], electrostatic fields \[[@B7]\], and to a lesser extent, electromagnetic fields of VDTs can play a role \[[@B8],[@B9]\].
The relationship between working with VDTs and dermatological manifestations has not previously been investigated. Accordingly, this study was designed.
Methods
=======
An analytical cross-sectional study was conducted by using office workers of an organization in Tehran.
Age, gender and weekly hours of work with VDTs were considered as independent variables and dermatological manifestations on the face including erythema, acne rosacea, scaling, pruritus and burning sensation as dependent variables.
600 office workers were selected randomly from the workers with approximately same environmental conditions (temperature, humidity, light, etc.) and were divided into two groups according to VDT exposure in the past year.
The exposure-positive group consisted of those with 7 hours or more weekly exposure to VDTs in the workplace or home and the exposure-negative group with less than 7 hours weekly exposure.
The workers filled in a questionnaire and had a physical examination.
Non specific erythema, acne rosacea, scaling and seborrheic dermatitis were detected during examination and were considered as positive dermatologic findings. Pruritus and burning sensation which got worse or were produced by working with VDTs and alleviated after leaving the workplace were considered as acceptable positive dermatological findings.
Results
=======
251 workers were females and the remainders 349 were males.
Age and sex distribution of the workers is demonstrated in Table [(1)](#T1){ref-type="table"}. The average age was 44.5 years.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Frequency distribution of office workers with relation to age and gender.
:::
**Age(year)** **Male** **Female** **Sum**
--------------- ---------- ------------ ---------
25--29 15 10 25
30--34 48 41 89
35--39 64 57 121
40--44 95 56 151
45--49 99 66 165
50--54 25 20 45
55--59 3 1 4
Sum 349 251 600
:::
294 workers had seven or more weekly hours of exposure and 306 had less than seven hours of weekly exposure in the past year.
In the exposure-positive group and exposure-negative groups 128 and 123 workers were females and 165 and 183 workers were males respectively.
The average weekly exposure in the exposure- positive group was 14 hours.
In the exposure-positive group 27 workers (16 female and 11 male) and in the exposure- negative groups 5 (3 female and 2 male) had dermatological manifestations respectively, as depicted in Table [(2)](#T2){ref-type="table"}.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Frequency of dermatologic manifestations in workers with more than 7 hours weekly exposure and less than 7 hours weekly exposure
:::
**Dermatologic manifestation** **Frequency in exposure negative group** **Frequency in exposure positive group**
-------------------------------- ------------------------------------------ ------------------------------------------
Nonspecific erythema 1 5
Rosacea acnea 1 2
Erythema,,Scaling 1 3
Itching, Burning 2 17
Sum 5 27
:::
Frequency of the exposure-positive workers according to their weekly hours of exposure and the frequency of dermatologic manifestations is shown in Table [(3)](#T3){ref-type="table"}.
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Frequency distribution of exposure positive group with weekly work hours and dermatologic manifestation
:::
**Exposure time per week** **People with dermatologic manifestations** **Total people with exposure** **Percent %**
---------------------------- --------------------------------------------- -------------------------------- ---------------
7--11 4 104 3.85
12--16 11 126 8.6
17--21 4 32 12.5
22--26 6 24 25
27--31 2 8 25
(Chi-square, linear by linear association: statistic = 12.735, df = 1, p-value \< 0.005)
:::
Discussion
==========
Statistical analysis of the confounding factors (age and sex) was performed between the two groups and no statistically significant difference was observed.
Use of chi-square test led to a P.value of less than 0.05, indicating statistical differences between exposure-positive and exposure-negative groups for the above mentioned dermatological manifestations.
On the other hand, Table [(3)](#T3){ref-type="table"} clearly indicates that the frequency of dermatological manifestations of the face tend to increase with increasing weekly hours of work with VDTs (chi-square, linear by linear association: statistic = 12.735, df = 1, P.value \< 0.005).
In a study conducted by Stenberg B. et al., psychosocial conditions and exposure to electromagnetic fields or conditions associated with such factors were related to an increased occurrence of skin symptoms.
The results also indicated that personal factors such as atopic dermatitis and physical exposure factors influencing indoor air quality, such as paper exposure and cleaning frequency were related to an increased prevalence of symptoms. The results suggest that skin symptoms reported by VDTs users have a multi-factorial causation \[[@B2]\]. According to other reports, mainly from Norway and Sweden, video display terminal work is suspected of causing skin rashes.
Three different studies, have tried to elucidate the question, and the results point to a possible relationship between VDT work and aggravation of some common skin diseases such as rosacea, seborrheic and atopic dermatitis, and acne. Whether this depends on physical, chemical, or psychological factors is still unknown \[[@B10]\].
Conclusion
==========
According to other studies and our own study, we can propose a relationship between dermatological manifestations on the face and exposure to VDTs and the probability of the occurrence of these manifestations increase with increasing exposure time.
Based on our findings and those of others, we recommend that in workers with long time exposures to VDTs who display dermatological manifestations of the face, occupational history of working with VDTs, weekly hours of exposure, and effects of exposure on their symptoms should be considered.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
AO participated in the design of the study and performed the data collection.
MP conceived of the study, and participated in its design and coordination.
SA drafted the manuscript and coordination.
RE performed the statistical analysis.
IM and MM drafted the manuscript.
All authors read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-5945/5/3/prepub>
|
PubMed Central
|
2024-06-05T03:55:55.195820
|
2005-4-3
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079806/",
"journal": "BMC Dermatol. 2005 Apr 3; 5:3",
"authors": [
{
"first": "Omid",
"last": "Aminian"
},
{
"first": "Parvin",
"last": "Mansoori"
},
{
"first": "Akbar",
"last": "Sharifian"
},
{
"first": "Ehsan",
"last": "Rafeemanesh"
},
{
"first": "Maria",
"last": "Mazaheri"
},
{
"first": "Mohammadreza",
"last": "Iraniha"
}
]
}
|
PMC1079807
|
Background
==========
Studies of the evolutionary history of protein interaction network (PIN) data have produced an almost bewildering range of (partially) contradictory results \[[@B1]-[@B6],[@B8]-[@B12]\]. While PIN data is notoriously prone to false positive and negative results \[[@B5],[@B13]\], reasons for disagreements are probably more diverse than just the quality of the interaction data. Failure to account for protein abundance -- as measured by gene expression levels, or by proxy, the codon-adaptation index -- has been criticized \[[@B3]\]; the choice of species for comparative analysis will also affect any evolutionary inferences as shown by Hahn *et al.*\[[@B12]\]. This may either be due to loss of power (*e.g.*fewer reliably identified orthologues between more distantly related species) or to differences in underlying PINs in distantly related species. Below, for example, we will show that results obtained from a comparison between the two hemiascomycetes *Saccharomyces cerevisiae*and *Candida albicans*differ considerably from those obtained using a distant *S. cerevisiae*-- *Caenorhabditis elegans*comparison. Finally, it has recently been shown that many studies may have suffered from the fact that present network data, and this is in particular true for PINs, are random samples from much larger networks. Unless these subnets are adequate representations of the overall network, their structural properties (such as node connectivity) may differ quite substantially from that of the nodes in the global network. This is, for example, the case for so-called scale-free network models \[[@B14]\].
Moreover, many studies have ignored the underlying network structure \[[@B15]\] in the statistical analysis. The network, however, introduces dependencies between connected proteins which should not be ignored. Fraser *et al.*\[[@B2]\] for example find that (i) there is a negative correlation between a protein\'s evolutionary rate and its connectivity *k*(the number of its interactions), (ii) connected proteins have positively correlated evolutionary rates, and (iii) connected proteins do not have correlated connectivities. All three statements cannot, of course, be strongly true simultaneously. Here we observe only relatively weak -- though statistically significant -- correlations between connectivity and evolutionary rate. We will argue that in a regression framework \[[@B16]\] some of these quantities contain very little information indeed about the corresponding properties of their interaction partners. Furthermore, we will demonstrate that when analyzing network data the network structure must be included into the analysis from the outset. Here we will first perform an evolutionary analysis of the yeast and nematode PIN data available in the DIP database \[[@B7]\], a hand-curated dataset combining information from a wide range of sources, followed by a comparison of the two datasets. When making comparisons between yeast species and between nematode species, we use only a single PIN dataset -- for *S. cerevisiae*and *C. elegans*, respectively -- and take comfort from the observation of Hahn *et al.*\[[@B12]\] who find that evolutionary analysis involving closely related reference taxa produces consistent results. Previously, topological comparisons of biological network data from different species have been made \[[@B17]\] but here we focus on shared evolutionary characteristics of PINs in the two species. We would expect at least some level of similarity of biological networks between species; but the more distantly related two organisms are, the more changes can have accumulated in their respective molecular networks. Thus, the depth of the phylogeny can affect the evolutionary analysis of PINs; it is, for example, unlikely that PINs have been conserved over large evolutionary time-scales.
Results
=======
Evolutionary analysis of the *S. cerevisiae*PIN
-----------------------------------------------
For the evolutionary analysis of the yeast PIN we use a panel of related yeast species: *Saccharomyces mikatae*, *Saccharomyces bayanus*, *Saccharomyces casteliii*, *Saccharomyces kluyveri*, *C. albicans*and *Schizosaccharomyces pombe*(see Methods section); the evolutionary relationship between these species is shown in figure [1](#F1){ref-type="fig"}. We thus focus on relatively recent evolutionary change which allows us to study the effects of the network structure on the rate of evolution more directly than *e.g.*distant comparisons of *S. cerevisiae*and *C. elegans*, which may, after all, have different PINs.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Phylogeny of the organisms used in the study**. The evolutionary relationship of the organisms used in this study. The last common ancestor of the ascomycetes in this phylogeny has been estimated to have lived approximately 330 million years ago. For the nematodes only two annotated genomes were available: their last common ancestor is believed to have lived approximately 100 million years ago.
:::
:::
Connectivity, expression and evolutionary rates in the *S. cerevisiae*PIN
-------------------------------------------------------------------------
For most protein sequences we have not been able to identify orthologues in all yeast species used in this analysis. We therefore defined the averaged relative evolutionary rate *R*(see Eqn. (1) in the *Methods*section) which allows us to make comparisons for 4124 out of the 4773 yeast genes for which we have interaction data.
In figure [2](#F2){ref-type="fig"} we show the dependence between inferred evolutionary rates and connectivities and expression levels, respectively. Our comparative analysis found statistically significant, though small, negative correlation, measured by Kendall\'s *τ*, between estimated evolutionary rates and a protein\'s number of interactions. In table [1](#T1){ref-type="table"} and figure [3](#F3){ref-type="fig"} we observe that comparisons with all species support this notion We furthermore estimated approximate confidence intervals for *τ*from 1000 bootstrap replicates \[[@B18]\] (shown in table [1](#T1){ref-type="table"}).
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Dependence of evolutionary rate in ascomycetes on the number of protein interactions and expression level**. The averaged relative rate *R*decreases with increasing number of interaction partners (*τ*≈ -0.06) and the expression level (*τ*≈ -0.23). The 95% bootstrap intervals for Kendall\'s *τ*values obtained from the six species comparisons are always negative (see table 1). The linear regression curves (red) appear concave on the log-transformed *x*-axis.
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Evolutionary analysis of *S. cerevisiae*Correlations between evolutionary rate, number of connections and expression level of proteins and the confidence intervals for Kendall\'s *τ*statistic obtained for the different ascomycete species. Values of *τ*that have associated p-values \< 0.01 are highlighted in bold. *X*1 denotes correlation with evolutionary rate obtained from a pairwise sequence comparison between *S. cerevisiae*and species *X*; *X*2 differs from *X*1 only in that the evolutionary rate was obtained using a maximum likelihood estimate. *M*denotes a rate obtained with respect to *S. mikatae*, *B*to *S. bayanus*, *C*to *S. castellii*, *K*to *S. kluyveri*, *A*to *C. albicans*, and *P*to *S. pombe*.
:::
Species comparison
-------------- -------------------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- -----------
M1 M2 B1 B2 C1 C2 K1 K2 A1 A2 P1 P2
Connectivity **-0.13** **-0.16** **-0.13** **-0.14** **-0.11** **-0.12** **-0.12** **-0.14** **-0.08** **-0.08** **-0.10** **-0.10**
2.5-% -0.17 -0.19 -0.16 -0.17 -0.13 -0.14 -0.15 -0.17 -0.11 -0.11 -0.14 -0.13
97.5-% -0.11 -0.13 -0.11 -0.12 -0.08 -0.09 -0.08 -0.10 -0.06 -0.05 -0.07 -0.07
Expression **-0.25** **-0.30** **-0.26** **-0.29** **-0.29** **-0.32** **-0.28** **-0.32** **-0.28** **-0.28** **-0.30** **-0.28**
2.5-% -0 28 -0 33 -0 28 -0 30 -0 31 -0 34 -0 32 -0 35 -0 31 -0 30 -0 33 -0 31
97.5-% -0.22 -0.27 -0.24 -0.27 -0.27 -0.30 -0.25 -0.29 -0.26 -0.25 -0.27 -0.25
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Correlation and partial correlation between evolutionary rate, number of interactions and expression level**. Kendall\'s rank correlation (blue) and partial rank correlation coefficients (red) between *R*and the number of interactions (correcting the partial *τ*for expression level) and expression (correcting for the number of interactions).
:::
:::
Observed negative correlations between estimated evolutionary rates and the expression level -- which have been reported previously by Pal *et al.*\[[@B19]\] -- are more pronounced. Equally, *k*, the number of a protein\'s interactions, and expression levels are also correlated (*τ*= 0.09). There has been considerable controversy as to whether the effect of a protein\'s connectivity can be studied independently of expression levels (see *e.g.*\[[@B3],[@B4]\]). The observed values of *τ*suggest that expression levels are better predictors of the evolutionary rate than are connectivities. Calculating partial rank correlation coefficients, *τ*~*p*~, provides further evidence for this: correcting for expression reduces the correlation between the evolutionary rate *R*(or any of the individual rates) and the number of interactions, as is apparent from figure [3](#F3){ref-type="fig"}. As the phylogenetic distance between species increases, the negative partial correlation between evolutionary rate and connectivity decreases compared to the uncorrected rank correlation measure *τ*.
In the supplementary tables S1-S3 \[see [Additional file 1](#S1){ref-type="supplementary-material"}\] we show the evolutionary rates for the different functional categories, processes and cellular compartments (taken from *Gene Ontology*(GO) \[[@B20]\]. Interestingly, once the effects of expression and protein function on the estimated evolutionary rate are taken into account the dependence of the latter on connectivity in a generalized linear regression model \[[@B16]\] (where we log-transformed the expression level to obtain an approximately normal distribution) is considerably reduced. This can be assessed formally using the Akaike information criterion (AIC) \[[@B21]\] on the sub-models where one of the terms has been dropped (see methods). For the full model we obtain AIC = -407.4. Dropping expression from the model results in AIC = -196.9, indicating that a substantial amount of information about the evolutionary rate is contained in the expression levels. Dropping the other terms individually while retaining the rest results in: AIC = -392.9 if the connectivity is dropped from the statistical model, and AIC = -352.6 (process), -250.1 (function) and -392.7 (compartment). We thus have the following order of statistically inferred impact on the evolutionary rate (with a slight abuse of the notation): expression\>function\>process\>connectivity≈compartment. Using the rates obtained from comparisons with the individual species results in the same ordering.
Evolution of interacting proteins in *S. cerevisiae*
----------------------------------------------------
So far we have treated nodes/proteins as independent (using only their connectivities in the analysis) but we will now consider the extent to which interactions introduce dependencies into the data. It is intuitively plausible that interacting proteins have similar evolutionary rates, and this has indeed been reported by Fraser *et al.*\[[@B2],[@B22]\] and studied by others, too, *e.g.*\[[@B3],[@B12]\]. Just like them we find that evolutionary rate decreases with connectivity; we also observe that the connectivities of interacting proteins are anti-correlated in yeast (*τ*≈ -0.03 with *p*\< 10^-8^). This is well explained from the statistical theory of networks \[[@B14],[@B23]\], as well as structural analyses of PIN data, where it is found that highly connected proteins form hubs which connect sparsely connected proteins.
Taken together this would mean that the evolutionary rates of connected proteins should also be anti-correlated. This is, however, not the case when we look at the yeast PIN, where we find that evolutionary rates of interacting proteins are positively correlated as measured by Kendall\'s *τ*. The correlations we observe are only relatively weak (even though they are significant) *τ*≈ 0.05 -- 0.10 with *p*\< 10^-8^. In figure [4](#F4){ref-type="fig"} we show the distribution of the *τ*rank correlation under the correct network Null model (see methods) for rates, expression levels and connectivities of interacting proteins. The observed value always lies outside the distribution of the expected values. Also shown in the figure are the probabilities that two interacting proteins have identical GO-classifications for function, process and cellular compartment, respectively. Again the observed probabilities lie outside the distribution under the Null model.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Statistical dependencies of interacting proteins in *S. cerevisiae***. Bootstrap distributions of Kendall\'s *τ*between evolutionary rates, expression levels and numbers of interactions and probabilities that protein function and the processes and cellular compartments by which proteins are classified are identical for a pair of interacting proteins. The grey histograms show the distribution of the statistics obtained from 1000 bootstrap replicates and the red vertical lines indicate the observed value. The bootstrap procedure was constrained such that each sample reproduced the degree distribution of the observed PIN.
:::
:::
Correlation, even partial correlation, may, however, be an inadequate statistical measure if the data is structured (as in a network); one should then rather focus on the power of a factor such as expression level or connectivity to predict evolutionary rates. We assess this formally through the use of statistical regression models which describe the evolutionary rate of one protein as a function of the rate of its interacting partner, as well as of its expression level, number of interactions, function, process and cell compartment. The AIC, which for the full model is AIC = -2397.6, allows us to order the factors by the information they contain about a protein\'s evolutionary rate. The order (and the respective AIC value on dropping the factor from the model) is as follows: Expression (AIC = -1399.6), function (AIC = -1445.9), process (AIC = -1956.6), cellular compartment (AIC = -2226.6), connectivity (AIC = -2316.8), and the rate of one of its interaction partners (AIC = -2397.0). Note that, measured by the AIC, the evolutionary rate of an interaction partner provides virtually no additional information about a protein\'s own evolutionary rate, once the protein\'s own expression level, function and process have been taken into account.
Thus, in summary, we observe that the evolutionary rate of yeast proteins is inversely related both to their connectivity in the PIN and to their expression levels, with expression levels having a greater impact on a protein\'s evolutionary rate than connectivities. Finally, while there is statistically significant correlation between the rates of interacting proteins, the rate of one interaction partner carries very little information about the rate of the other protein if other factors are taken into account.
Evolutionary analysis of the *C. elegans*PIN
--------------------------------------------
In the evolutionary analysis of *C. elegans*we use *C. briggsae*, the only other congeneric nematode for which high quality whole-genome data is available. Since nematodes are multicellular, care has to be taken when analysing the effects of gene expression on evolutionary rate, as expression levels will vary considerably between tissues and, indeed, between different stages of the nematode life cycle. Because codon usage bias as a selective response increasing translational efficiency should be driven by the overall expression level of a protein integrated over both tissue and time, the codon-adaptation index (CAI; see Methods and \[[@B24]\]) can serve as a meaningful averaged quantity reflecting overall integrated expression levels better than a direct measurement of mRNA expression level data obtained from any single tissue type.
Connectivity, expression and evolutionary rates in the *C. elegans*PIN
----------------------------------------------------------------------
The correlation of evolutionary rate and connectivity is somewhat reduced compared to *S. cerevisiae*with a point estimate of *τ*= -0.05 with a 95% bootstrap CI of \[-0.097, -0.017\]. Anti-correlation between the CAI measure of expression and evolutionary rates is again much more pronounced with *τ*≈ -0.30 and approximate bootstrap CIs of \[-0.333, -0.264\]. The resulting scatter plots of rate vs. connectivity and rate vs. CAI are shown in figure [5](#F5){ref-type="fig"}.
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Dependence of evolutionary rate in nematodes on the number of protein interactions and CAI**. The estimated evolutionary rate decreases with increasing number of interaction partners (*τ*≈ -0.03) and the expression level (*τ*≈ -0.30). We have again transformed the *x*-axis in the scatterplot of rate vs. connectivity which leads to the concave shape of the regression line (red).
:::
:::
Partial correlation coefficients again show that the influence of expression is greater than that of connectivity: *τ*~*p*~≈ -0.03 for the partial correlation measure between rate and connectivity, while *τ*~*p*~≈ -0.30 if the correlation between expression (CAI) and rate is corrected for connectivity. This is confirmed by performing an ANOVA \[[@B25]\] on the regression between rate, CAI and connectivity where no significant correlation can be found between rate and connectivity (*p*≈ 0.62). Generalized linear regression modelling shows that measured by the AIC a model in which the rate depends only on the CAI but not on the connectivity (AIC = -660.5) is more powerful than a model in which the rate depends on both connectivity and CAI (AIC = -618.4). In the absence of extensive GO data we find that the CAI is the only statistically significant predictor for a protein\'s evolutionary rate.
Evolution of interacting proteins in *C. elegans*
-------------------------------------------------
Comparing properties of interacting proteins we again find a negative correlation between their respective connectivities (*τ*= -0.07) and a weaker positive correlation between their evolutionary rates (*τ*= 0.03).
The corresponding 95% bootstrap CI for *τ*does, however, include 0 and negative values; thus there is no statistical basis for concluding that evolutionary rates of interacting proteins are correlated in *C. elegans*even if we consider only the rank correlation measure. In figure [6](#F6){ref-type="fig"} the distribution of *τ*under the correct Null model (see methods) confirms this result as the observed correlation between the evolutionary rates of interacting proteins falls into the 95% confidence interval obtained from the Null model. Expression levels are, however, significantly correlated and connectivities remain significantly anti-correlated. Regression models, equivalent to those performed for yeast, confirm the negligible information a protein\'s evolutionary rate contains about the evolutionary rate of an interacting protein.
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**Statistical dependencies of interacting proteins in *C. elegans***. Bootstrap distributions of Kendall\'s *τ*between evolutionary rates, expression levels and numbers of interactions. The grey histograms show the distribution of *τ*obtained in 1000 bootstrap replicates and the red lines indicate the observed value. The bootstrap procedure was constrained such that each sample reproduced the degree distribution of the observed PIN.
:::
:::
In summary, for *C. elegans*we find that expression, even if measured indirectly through the CAI, is a better predictor about a protein\'s evolutionary rate than connectivity and GO classifications. The evolutionary rates of connected proteins do not appear to be correlated.
Comparing the PINs of *S. cerevisiae*and *C. elegans*
-----------------------------------------------------
It is instructive to compare the PINs of the two model organisms, yeast and worm, directly. We have therefore used our earlier approach of identifying and analysing orthologues to the yeast and nematode PIN data. While we are, of course, aware that this may be problematic given the two or three billion years of evolutionary history separating the two organisms, it should serve as a useful illustration of the amount of information one model-organism is likely to provide about another (including, of course, humans). Using this approach we found a total of 524 pairs of orthologues. These we aligned and from the alignments we estimated evolutionary rates. For all of these proteins we have PIN data and for most we also have information about their expression levels in the two species. The results are summarized in tables [2](#T2){ref-type="table"} and [3](#T3){ref-type="table"}. Although they essentially agree with the earlier results, they do suggest that the choice of species used for inferring the evolutionary rate can influence the analysis. For example, the partial correlation between interaction and evolutionary rate (calculated directly from the *S. cerevisiae*-- *C. elegans*amino acid sequence comparison) accounting for expression is much less reduced compared with the simple correlation coefficient (*τ*~*p*~= -0.20 in *S. cerevisiae*, and *τ*~*p*~= -0.10 in *C. elegans*) than when evolutionary rates are calculated using more closely related target species. Over long evolutionary distances it appears as if connectivity and expression level act almost independently. However, the more reliable comparisons of the previous section suggest that this is not the case.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Correlations obtained from a direct comparison of *S. cerevisiae*with *C. elegans*Orthologues in the *S. cerevisiae*and *C. elegans*PINs where identified by reciprocal BLAST searches and evolutionary rates, estimated previously (see table 1), were analysed for correlation between evolutionary rate, the number of interactions and expression levels. We also performed an analysis with evolutionary rates estimated directly from the distant *S. cerevisiae*and *C. elegans*comparison.
:::
**Comparison** **Evolutionary Rate obtained from closely related species** **Evolutionary Rate obtained from *S. cerevisiae*- *C. elegans***
--------------------- ------------------------------------------------------------- ------------------------------------------------------------------- ----------- -----------
Nr. of Interactions **-0.11** **-0.13** **-0.20** **-0.10**
2.5-percentile -0.20 -0.23 -0.26 -0.19
97.5-percentile -0.02 -0.04 -0.14 -0.03
Expression **-0.33** **-0.44** **-0.25** **-0.42**
2.5-percentile -0.41 -0.50 -0.32 -0.47
97.5-percentile -0.24 -0.36 -0.19 -0.37
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Correlations between orthologous proteins in the *S. cerevisiae*and *C. elegans*PINs Observed rank correlations (measured by Kendall\'s *τ*) for evolutionary rates (measured with respect to *S.mikatae*and *C. briggsae*, respectively), connectivity and protein expression level (estimated by mRNA expression level in *S. cerevisiae*and CAI in *C. elegans*).
:::
**Quantity** **Observed***τ* **95% CI**
------------------- ----------------- -------------
Evolutionary Rate 0.24 0.12--0.35
Connectivity 0.07 0.001--0.14
Expression 0.32 0.26--0.39
:::
Comparing properties of orthologous proteins we find that their expression levels (using the CAI as a proxy in *C. elegans*) show the strongest correlation while their respective PIN connectivities show the lowest value for Kendall\'s *τ*statistic. This may be due to the noise in the PIN data or the incomplete nature of present PIN data sets. We expect that the relatively small proportion of *C. elegans*proteins included in the DIP database will also lead to an inaccurate representation of the *C. elegans*PIN.
Discussion
==========
There are considerable differences between the various published studies \[[@B2]-[@B4],[@B12]\], both in terms of protein interaction data and phylogenetic comparisons. We therefore focus on closely related species for both *S. cerevisiae*and *C. elegans*in the evolutionary analysis, since we probably have to assume that the underlying PIN is relatively more conserved over short evolutionary distances. While we found some evidence that highly connected proteins evolve more slowly than sparsely connected proteins, (i) the negative correlation between rate and expression level is more pronounced, (ii) in *S. cerevisiae*and *C. elegans*connectivity turns out to be a worse statistical predictor of the evolutionary rate than expression. For *S. cerevisiae*we also find that protein function and the principal biological process a protein is involved in have a greater impact on the evolutionary rate of a protein than its connectivity.
We believe that the importance of expression over connectivity in determining the evolutionary rate may be due to three factors. First, highly abundant genes are perhaps more visible to purifying selection \[[@B26]\], which might tend to lower the rate at which they evolve. Second, and more importantly, highly expressed genes, which are under selection for translational efficiency, use only a small subset of the cognate codons for a particular amino acid (this, incidentally, is exploited in the construction of the CAI), and because this subset is often the same even in phylogenetically remote organisms -- for example, for those amino acids encoded by nnU and nnC (*e.g.*phenylalanine or cysteine), nnC is almost universally preferred -- the silent substitution rate is reduced. Third, the replacement substitution rate in highly expressed proteins should also be reduced for a similar reason to selection for translational efficiency at silent sites: selection for more cheaply synthesised amino acids at replacement sites \[[@B27]\]. This can be shown to lead to the avoidance of amino acids which are metabolically expensive to synthesise at functionally-unconstrained sites in highly expressed proteins, which reduces the set of acceptable amino acids at such sites and thereby lowers the replacement substitution rate compared with that at functionally-unconstrained sites in low expression proteins \[[@B28]\].
We have also applied an improved resampling procedure to the analysis of correlation between rates and expression levels of connected proteins. In our analysis we treated properties of the network as a confounding variable and in addition to studying correlations we also show how informative properties of one protein are about properties of its interaction partners. We find that the correct procedure broadens the resampling confidence intervals but that expression levels of interacting proteins remain considerably closer than would be expected by chance. Conversely we found no evidence of a correlation in the evolutionary rate of interacting proteins in *C. elegans*and only extremely weak evidence in *S. cerevisiae*. Our results also suggest, that the evolution of interacting proteins is not as tightly correlated as some researchers have proposed. This level of disagreement may be caused by uncertainties in the data or the fact that subnets sampled from larger networks inaccurately reflect the properties of the true network \[[@B29]\].
Conclusion
==========
We believe that the effects of the network structure on the evolution of proteins, and vice versa, is much more subtle than has previously been suggested. In the present dataset expression levels appear to have shaped a protein\'s evolutionary rate more than its connectivity. If we are happy to accept present PIN data with the necessary caution, then this observation is consistent with a scenario where expression levels are more conserved between species than are details of the interaction network. Nevertheless, we believe that it is important to consider the PIN and a protein\'s connectivity explicitly and from the outset in any statistical analysis as the underlying network appears to act as a confounding factor.
Methods
=======
Data
----
### Protein interaction data
The names and sequences for proteins with known interactions in Saccharomyces cerevisiae and Caenorhabditis elegans were retrieved from the Database of Interacting Proteins (DIP) on the 5th of July \[[@B7]\]. The database mainly contains information extracted from the research literature, but recently the database was enriched with information obtained by analysing structures of protein complexes deposited in PDB \[[@B30]\]. We have data for 4773 yeast proteins (comprising 15461 interactions) and 2386 nematode proteins (with 7221 interactions). While there have been recent attempts at quantifying levels of confidence in given protein interactions these generally lead to substantial decreases in sample size. For this reason we have therefore chosen to take the PIN data as it is deposited in the hand-curated DIP database (we have also performed analyses with such restricted subsets which agree with the results presented here).
### Protein sequence data
In addition to *S. cerevisiae*sequences downloaded from the DIP, publicly available protein sequences of six other yeast species were investigated: *Saccharomyces pombe*\[[@B31]\], *Candida albicans*\[[@B32]\], *Saccharomyces mikatae, Saccharomyces bayanus, Saccharomyces kluyveri and Saccharomyces castellii*\[[@B33]\]. Genomic protein sequences for only one other Caenorhabditis species apart from *C. elegans*are publicly available, *C. briggsae*\[[@B34]\]. All sequence files were converted to searchable indexed databases; these are available from the authors.
### Expression data
*S. cerevisiae*expression data came from Cho et al. \[[@B35]\] who characterised all mRNA transcript levels during the cell cycle of *S. cerevisiae.*mRNA levels were measured at 17 time points at 10 min intervals, covering nearly two full cell cycles \[[@B36]\]. The mean of these 17 numbers was taken to produce one general time-averaged expression level for each protein.
*C. elegans*is a multicellular organism in which different cells have different functions. This means that different proteins have different expression levels in different cells, which are present in different numbers, so taking a simple mean of the expression levels in a single cell type would be meaningless. In addition, *C. elegans*has a complex life-cycle, with different proteins being expressed in different stages of that cycle. Thus, an alternative way of calculating a single expression level for each protein had to be used. It has long been known that highly expressed genes tend to use only a limited number of codons thus displaying high codon bias. Sharp and Li \[[@B24]\] devised a measure for assessing the degree of deviation from a preferred pattern of usage estimated from the clustering of codon usage across proteins, which they called the Codon Adaptation Index (CAI). We adopt this measure as our expression level proxy.
Methods
-------
### Phylogenetic analysis
The close relationship between the species considered here, apart from the distant comparison between *S. cerevisiae*and *C. elegans*, makes identification of orthologues relatively straightforward. Orthologous protein sequences were detected by reciprocal BLAST searches in the standard way. Multiple alignments of inferred orthologues were obtained using ClustalW.
Evolutionary rates were obtained using PAML (Phylogenetic Analysis by Maximum Likelihood) \[[@B37]\]. We used both the observed fraction of amino acid differences, referred to by *M*1, *B*1,\... (where the letters refer to different species, see footnote to Table [1](#T1){ref-type="table"}), and the distance related measure calculated from the trees inferred by PAML, referred to by *M*2, *B*2, \.... Both rates are highly correlated *τ*≈ 0.9. In order to estimate the latter rate the phylogeny had to be reconstructed. Inferred phylogenies were assessed for their agreement with the commonly accepted family tree of yeast species (see figure [1](#F1){ref-type="fig"}; we found excellent agreement among the inferred trees assessed using the *clann*software package \[[@B38]\]) and the widely accepted phylogenies for the yeast and nematode species, which are shown in figure [1](#F1){ref-type="fig"}. For further analyses we chose to use the maximum likelihood rate.
### Statistical analysis
In order to be able to compare evolutionary rates for as many proteins as possible we defined the averaged relative evolutionary rate of each protein *i*via
where *v*~*i*~is the number of comparisons from which an evolutionary rate can be estimated.
We generally found that analysis of the dependence of *R*= {*R*~1~, *R*~2~, ..*R*~*n*~} on the number of interactions *etc.*behaved similarly to analysis of the species specific rates. We used ANOVA \[[@B25]\] and partial correlation coefficients to study the impact the different factors had on the evolutionary rates. All analyses were performed using the R statistical environment and the NetZ package (available from the authors). In order to investigate the relative influence of the various factors (number of interactions, expression levels, GO-data \[[@B20]\]) we used linear and generalized linear regression modelling (implemented in R). The Akaike information criterion (AIC) \[[@B16],[@B21]\] was used to distinguish among the different nested submodels of the full model. The model which has the smallest AIC (defined as 2(-1 k(*θ*) + 2*v*) where 1 k(*θ*) is the log-likelihood of a -- potentially vector-valued -- parameter *θ*, and *v*is the number of parameters) is the model which offers the best (in an information sense \[[@B21]\]) description of the data. The full model included the number of interactions, expression levels (or CAI in the case of *C. elegans*) and GO-data as explanatory variables. When comparing evolutionary rates of interacting proteins the evolutionary rate at the interacting protein was added as an explanatory variable. The AIC (and related approaches \[[@B21]\]) aims to identify a statistical model that offers the most efficient description of the data (in an information theoretic sense) from a set of trial models.
We explicitly incorporated the network structure into the statistical analysis. This is necessary if there is reason to believe that properties of the network may determine aspects of the evolutionary history, for example when we want to test if the evolutionary rates of interacting proteins are correlated. Here we use resampling or bootstrap procedures \[[@B18]\] to determine if properties (*e.g.*expression levels, evolutionary rates, connectivities) of interacting proteins are more similar than would be expected to occur by chance. If instead we had paired proteins completely at random we would potentially have masked confounding effects due to the network (for example if expression depends strongly on a protein\'s network properties, *i.e.*connectivity). In our network-aware resampling procedure we therefore pick each protein with a probability that is proportional to its number of interaction partners. Each bootstrap replicate is thus also a sample with the correct nodal properties and (statistically) the same degree distribution as the true network. In the structural analysis of networks the need to account for network properties in the construction of the correct Null model has long been realized \[[@B14],[@B17]\] but this is, to our knowledge, the first time that such a topologically correct Null model has been applied to the evolutionary analysis of network data. As an illustration figure [7](#F7){ref-type="fig"} shows the bootstrap distribution of correlation coefficients of expression levels in yeast for the correct Null model and for the model where proteins are paired completely at random. Ignoring the correlation of the data introduced by the underlying network structure reduces the bootstrap confidence intervals considerably (we find that the two-sided 95% CIs are reduced by approximately 20% compared to the network aware bootstrap replicate). This mirrors the effects observed in population and evolutionary genetics where the underlying genealogy/phylogeny increases the CIs compared to the case of truly independent observations.
::: {#F7 .fig}
Figure 7
::: {.caption}
######
**Confidence intervals calculated with and without including the network structure**. Distribution of Kendall\'s *τ*measuring the expected correlation between the expression levels of interacting proteins in *S. cerevisiae*. The grey distribution has been calculated under the correct empirical Null distribution, where pairs of interacting proteins are chosen such that the degree distribution of the re-sampled protein network is the same as that of the true network. The red bars indicate the distribution obtained under the conventional (and inadequate) Null model where pairs of proteins are chosen entirely at random. Including the network structure into the bootstrap procedure leads to a broader distribution.
:::
:::
Routines used to perform the statistical analysis of the network data are collected in the NetZ package which can be obtained from the corresponding author.
Authors\' contributions
=======================
IA collected the data, performed the phylogenetic analysis, helped with the statistical analysis and with writing the manuscript. JS calculated the codon adaptation index, helped with the evolutionary and statistical analysis and writing the paper. JA, DH and SB helped with data retrieval, the evolutionary analysis of the PIN data, and writing the manuscript. MPHS devised the study, performed the statistical analysis and wrote the manuscript.
Supplementary Material
======================
::: {.caption}
###### Additional File 1
We have used GO classifications \[[@B20]\] to evaluate the extent to which function (table S1), biological process (table S2), or celluar compartment (table S3) of a protein may influence the evolutionary rate of proteins in *S. cerevisiae*.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
IA thanks the Wellcome Trust for a PhD studentship in Bioinformatics. MPHS gratefully acknowledges a Wellcome Trust Research fellowship. This research was also funded by the Royal Society and EMBO. We thank Eric de Silva, Piers Ingram, Bob May and Carsten Wiuf for helpful discussions.
|
PubMed Central
|
2024-06-05T03:55:55.196752
|
2005-3-18
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079807/",
"journal": "BMC Evol Biol. 2005 Mar 18; 5:23",
"authors": [
{
"first": "Ino",
"last": "Agrafioti"
},
{
"first": "Jonathan",
"last": "Swire"
},
{
"first": "James",
"last": "Abbott"
},
{
"first": "Derek",
"last": "Huntley"
},
{
"first": "Sarah",
"last": "Butcher"
},
{
"first": "Michael PH",
"last": "Stumpf"
}
]
}
|
PMC1079808
|
Background
==========
Many complex systems can best be analyzed as networks where the basic building blocks of the system are represented as nodes and their interactions as links: Recent studies of systems as disparate as the network of scientific co-authorships, sexual contacts and the World-Wide-Web have revealed unexpected similarities, suggesting that their structure and growth is ruled by a set of generic organizing principles \[[@B1],[@B2]\]. A variety of biological systems, like food webs and the various biochemical interactions between genes, proteins and metabolites, have been found to exhibit similar large-scale traits \[[@B3]-[@B6]\]. The most prominent is the scale-free property of the connectivity distribution. When combined with a modular structure, the resulting network consists of a hierarchy of interwoven clusters \[[@B7]-[@B10]\].
Protein crystallography reveals that the fundamental unit of protein structure is the domain. Independent of neighboring sequences, this region of a polypeptide chain folds into a distinct structure and mediates biological functionality \[[@B11]\]. Comparing domain architectures of proteins in multicellular organisms evidence emerged that preexisting domain architectures have predominantly been supplemented with single domains at their terminal sites \[[@B12],[@B13]\].
Functional links between proteins have also been detected by analyzing the fusion patterns of protein domains. Two separate proteins A and B in one organism may be expressed as a fusion protein in other species. A protein sequence containing both A and B is termed a Rosetta Stone sequence. However, this framework only applies in a minority of cases \[[@B14]\].
The structure of the protein domain network, where individual domains are nodes and their co-occurrences in a protein are interpreted as links, also displays a scale-free structure \[[@B15]-[@B17]\]. Domains that are involved in cell-cell interactions, signal transduction and cell differentiation have been found to accumulate links, reflecting increasing complexity of the organisms specific evolutionary development in going from bacteria to eukaryotes.
In a recent study \[[@B18]\], we classified yeast proteins as being either *globally*or *locally*central according to the number and density of links in their network neighborhoods. In particular, we applied an iterative decomposition method (see Methods) that systematically uncovered core networks with nodes having degrees of at least *k*. In nesting through the different cores, we gradually defined highly connected proteins in the innermost cores as *globally*central while we call proteins that have been placed in cores on the periphery *locally*central. This categorization allowed us to demonstrate that *globally*central proteins participate in a substantial number of complexes while simultaneously displaying a high level of evolutionary conservation. Here, we apply this core decomposition method to study the properties of the protein domain co-occurrence networks of the eukaryotes *S. cerevisiae*, *C. elegans*, *D. melanogaster*, *M. musculus*and *H. sapiens*, allowing us to classify the various domains as either globally or locally central. In going from the single celled Yeast to the considered highly evolved multi cellular organisms we find that the number of globally central domains increases with the organisms level of evolutionary development. Also the overlap network which consists only of the nodes and links shared by *all*the organisms specific cores reveals those domain fusions that have been preserved through evolution. Comparing the co-occurrence networks to the physical protein domain interaction network of *S. cerevisiae*\[[@B17],[@B19]\] we find that links that appear in the innermost cores of the co-occurrence network of higher eukaryotes strongly coincide with physical interactions. The co-occurrences of domains that make them end up in the innermost cores of the co-occurrence networks might represent evolutionary patterns that serve as a putative proteome backbone. Since we find the driving force behind fusion events not to be a high frequency of interactions between a given protein domain pair but a large number of individual interaction partners, we conclude that links appearing in the innermost cores of the co-occurring networks are the result of underlying important domain interactions.
Results
=======
Statistics of domain networks
-----------------------------
Table [1](#T1){ref-type="table"} summarizes basic statistics of the domain co-occurrence networks of *H. sapiens*, *M. musculus*, *C. elegans*, *D. melanogaster*and *S. cerevisiae*. All the domain co-occurrence networks have a major component containing the vast majority of the nodes, co-existing with many small, connected components. Both the average degree ⟨*k*⟩ and the clustering coefficients ⟨*C*⟩ of the networks gradually increase with elevated level of the organisms development. Determining the number of domains *N*proteins in the proteomes of *H. sapiens*, *M. musculus*, *C. elegans*, *D. melanogaster*and *S. cerevisiae*contain, we observe the presence of power-laws in frequency distributions thus obtained, *P*(*N*) \~ *N*^-*δ*^(Fig. [1b](#F1){ref-type="fig"}). This result confirms that the vast majority of proteins contains only a single domain \[[@B20]\], while a minority accumulates more domains. In Fig. [1c](#F1){ref-type="fig"}, we find that *N*shows a positive power-law correlation from the mean number of co-occurring domains -- the degree -- ⟨*k*⟩ \~ *N*^*ε*^, suggesting that on average frequently occurring domains are combined with an increasing number of changing partners. All co-occurrence networks display a scale-free degree distribution \[[@B15]\] (Fig. [1d](#F1){ref-type="fig"}), as exemplified by the presence of power-laws, *P*(*k*) \~ *k*^-*θ*^. Similarly, we find a power-law dependence of the clustering coefficient from the degree exemplified by a generalized Zipf-law ⟨*C*(*k*)⟩ = *α*(*β*+ *k*)^-*γ*^, indicating the network\'s inherent modularity (Fig. [1e](#F1){ref-type="fig"}). As summarized in Table [1](#T1){ref-type="table"}, the dependence of the power-law exponents in Figs. [1b--e](#F1){ref-type="fig"} well mirrors the level of an organisms evolutionary development. In particular, the higher the level of the organisms development, the smaller the values we find for *δ*(Fig. [1b](#F1){ref-type="fig"}), *θ*(Fig. [1d](#F1){ref-type="fig"}) and *γ*(Fig. [1e](#F1){ref-type="fig"}). In turn, we find the opposite for *ε*(Fig. [1c](#F1){ref-type="fig"}). Obviously, the increase in complexity of the organisms development coincides with an elevated networks heterogeneity. Since we find similar numbers of domains in the higher eukaryotes (Table [1](#T1){ref-type="table"}), we assume that an elevated degree of development presumably is increasingly obtained by frequent domain combinations.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Statistics of domain networks The domain networks of *S. cerevisiae*, *C. elegans*, *D. melanogaster*, *M. musculus*and *H. sapiens*all contain a major component (*mc*) which incorporates the majority of the domains (), coexisting with many small, connected components (*N*~*cc*~). The number of edges () of the main component, the mean node degree ⟨*k*⟩ and the mean clustering coefficient ⟨*C*⟩ gradually increase with the organisms level of development. Referring to Fig. 1, the frequency distributions of the number *N*of domains per protein (Fig. 1b) follows a power-law *P*(*N*) \~ *N*^-*δ*^. Similarly, we find a power-law in the distributions of the occurrence of domains *N*and their mean degree ⟨*k*⟩ in the organism specific co-occurrence networks *k*\~ *N*^*ε*^(Fig. 1c). The degree distributions of the organisms specific co-occurrence networks can be approximated by a power-law *P*(*k*) \~ *k*^-*θ*^(Fig. 1d). Similarly, we approximated the degree dependence of the clustering coefficient by a generalized Zipf-law ⟨*C*(*k*)⟩ = *α*(*β*+ *k*)^-*γ*^(Fig. 1e).
:::
organism *N*~*cc*~ ⟨*k*⟩ ⟨*C*⟩ *δ* *ε* *θ* *α* *β* *γ*
------------------- ----------- ----- ------- ------- ------- ----- ----- ----- --------- ------ -----
*H. sapiens* 172 733 4,048 5.52 0.42 2.1 0.5 1.4 61.6 17.9 1.5
*M. musculus* 173 668 3,566 5.34 0.43 2.5 0.5 1.5 67.5 17.1 1.5
*D. melanogaster* 172 506 2,274 4.49 0.39 2.3 0.4 1.6 1,407.8 34.7 2.7
*C. elegans* 167 495 2,120 4.28 0.34 2.4 0.4 1.7 3,886.7 30.9 2.5
*S. cerevisiae* 177 175 516 2.95 0.32 2.6 0.3 2.0 39.4 11.3 1.5
:::
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Basic statistics of domain occurrence networks**. **(a)**All domains co-occurring in a single protein are represented as a fully connected unweighted clique in the network. **(b)**Determining the number of domains each protein contains in *H. sapiens*, *M. musculus*, *D. melanogaster*, *C. elegans*, and *S. cerevisiae*, we observe power-laws *P*(*N*) \~ *N*^-*δ*^in frequency distributions thus obtained (see Table 1 for detailed values). This inhomogeneity in domain architectures suggests that the vast majority of proteins in all organisms considered contains only one domain. **(c)**Counting the occurrence of each domain in the proteomes of the organisms under consideration, we find a positive power-law dependence from the mean number of co-occurring domains -- the degree -- ⟨*k*⟩ \~ *N*^*ε*^, suggesting that on average frequent occurrence of a domain coincide with the participation in various domain architectures (see Table 1 for detailed values). **(d)**. The domain networks of *H. sapiens*, *M. musculus*, *D. melanogaster*, *C. elegans*, and *S. cerevisiae*display scale-free behavior, a network feature which is characterized by the power-law in the degree distribution *P*(*k*) \~ *k*^-*θ*^\[15\] (see Table 1 for detailed values). **(e)**The network\'s inherent modularity is indicated by the presence of a power-law dependence between the clustering coefficient and the degree as a generalized Zipf-law ⟨*C*(*k*)⟩ = *α*(*β*+ *k*)^-*γ*^(see Table 1 for detailed values). With respect to (b,c,d,e), we observe that the organisms specific distributions differ by their individual power-law exponents, indicating their levels of evolutionary development.
:::
:::
Cores of domain networks
------------------------
Due to the size of the domain co-occurrence networks considered we find different numbers of *k*-cores. While the networks of *H. sapiens*and *M. musculus*are decomposed into 8 nested *k*-cores, where *k*= 1, \..., 8 we find 6 *k*-cores in *D. melanogaster*and *C. elegans*(*k*= 1, \..., 6). There are only 4 in *S. cerevisiae*(*k*= 1, \..., 4). The placement of a node in a certain core allows an assessment of its meaning for the topology. A hub -- a highly connected node -- that is only a member of the peripheral *k*-cores is defined as locally central, while nodes (not necessarily the biggest hubs of the whole network) being members of the innermost cores are globally central (Fig. [2f](#F2){ref-type="fig"}). In Table [2](#T2){ref-type="table"}, we compiled lists of the highest connected domains in each core layer, being the set of nodes two consecutive cores do not have in common. Notably, the innermost *k-core*is not populated by the largest hubs, indicating that a high degree alone does not necessarily imply a central placement in the network. In fact, we observe that with even a considerably low degree domains can be placed in the innermost cores.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Five most connected domains in the innermost core-layers using their degree *k*in the full domain networks of *S. cerevisiae*, *C. elegans*, *D. melanogaster*, *M. musculus*and *H. sapiens*. Notably, the innermost layer *l*(equivalent to the innermost core) is not populated by the largest hubs, indicating that a high degree alone does not necessarily imply a central placement in the network. In fact, we observe that with even a considerably low degree *k*domains are placed in the inner cores, a domain specific feature we call *globally*central. We also observe highly connected domains that only make it to the outer cores, indicating a *locally*central position in the respective networks.
:::
*S. cerevisiae* *C. elegans* *D. melanogaster* *M. musculus* *H. sapiens*
----------------- -------------- ------------------- --------------- -------------- ----- ------------ ----- ----- -------------- ----- ----- -------------- ----- -----
domain *l* *k* domain *l* *k* domain *l* *k* domain *l* *k* domain *l* *k*
SH3 4 16 pkinase 6 44 pkinase 6 50 pkinase 8 64 pkinase 8 74
pkinase 4 13 PH 6 38 SH3 6 41 EGF 8 60 EGF 8 64
C2 4 11 SH3 6 33 PH 6 40 PH 8 49 PH 8 59
DAG PE-bind 4 4 efhand 6 27 ank 6 38 SH3 8 49 SH3 8 52
pkinase C 4 4 PDZ 6 26 PDZ 6 35 ig 8 44 ig 8 52
PH 3 14 EGF 5 38 zf-C2H2 5 24 fn31 7 31 zf-C3HC4 7 51
helicase C 3 13 ank 5 36 zf-C3HC4 5 23 Idl-recept a 7 26 WD-40 7 39
zf-C3H4 3 10 zf-C3HC4 5 24 UBA 5 23 tsp 1 7 25 tsp 1 7 27
UBA 3 10 ig 5 24 helicase C 5 22 zf-CCHC 7 24 UBA 7 26
myb DNA-bind 3 10 fn3 5 20 efhand 5 21 WW 7 21 vwc 7 21
WD40 2 11 F-box 4 19 WD-40 4 18 zf-C3HC4 6 48 zf-C2H2 6 31
AAA 2 10 zf-C2H2 4 18 rrm 4 13 WD40 6 31 PHD 6 27
ank 2 8 WD-40 4 17 TPR 4 11 zf-C2H2 6 28 helicase C 6 24
UCH-2 2 6 rrm 4 15 tsp 1 4 10 helicase C 6 22 rrm 6 23
HATPase c 2 6 PHD 4 15 zf-CCHC 4 10 TPR 6 19 TPR 6 23
clathrin 1 3 UBA 3 10 ubiquitin 3 10 PHD 5 25 BRCT 5 18
ENTH 1 2 homeobox 3 10 AAA 3 9 bromodomain 5 19 PWWP 5 14
zf-CCCH 1 2 Kunitz BPTI 3 10 heme 1 3 7 BRCT 5 17 DEAD 5 14
exo-endo-phos 1 2 metallophos 3 9 GTP-EFTU 3 7 SET 5 13 zf-RanBP 5 13
SAP 1 2 Dna J 3 7 kinesin 3 6 DEAD 5 13 myb-DNA bind 5 13
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Cores of the domain co-occurrence networks**. The *k*-core of a graph is defined as the largest subgraph where every node has at least *k*links. For each choice of *k*, we determine the *k*-cores by iteratively pruning all nodes with degree lower than *k*and their incident links. In the schematic representation, the 1-core consists of all the nodes while the 3-core only contains the nodes on orange background. Panels a-e show the 2 innermost *k*-cores (red: 1-core and yellow: 2-core) of the domain networks mapped for the proteomes of **(a)***S. cerevisiae*, **(b)***C. elegans*, **(c)***D. melanogaster*, **(d)***M. musculus*and **(e)***H. sapiens*. **(f)***Local*vs. *global*centrality. Interpreted as its importance a node is related to its degree and network neighborhood. A hub that is only a member of the outer *k*-cores is defined as locally central (top-left), while nodes (not necessarily the biggest hubs) being-members of the innermost cores are globally central (top-right).
:::
:::
In Fig. [2a--e](#F2){ref-type="fig"} we show the two innermost cores of the protein domain networks of *S. cerevisiae*, *C. elegans*, *D. melanogaster*, *M. musculus*and *H. sapiens*, respectively. We observe that the increase in the organisms complexity is not only reflected by higher numbers of cores, but the innermost cores differs in size. We find, that *H. sapiens*has 43 domains in the innermost core, while we find 30 in *M. musculus*and *D. melanogaster*, 14 in *C. elegans*and 5 in *S. cerevisiae*. Reflecting the increasing evolutionary development of the underlying organism, the cores are enriched with domains predominantly associated with functions such as cell-cell contact and signal transduction. This observation agrees well with the known evolutionary development from the single cellular *S. cerevisiae*to the multicellular higher eukaryotes. In particular, the innermost core of *S. cerevisiae*(panel a) consists of the following clique of interconnected domains: (i) pkinase and (ii) pkinaseC which are signal transduction domains, (iii) DAG-PE-bind, a domain which binds Diacyl-glycerol, activating the family of the previously mentioned kinases, (iv) HR1 which is involved in binding the small G-protein rho, and (v) C2, a Ca^2+^-dependent membrane-targeting module found in many cellular proteins that are involved in both signal transduction and membrane trafficking.
Nesting through the innermost cores of the more evolved organisms, we find that the initial small innermost core of Yeast is enriched with clusters of densely connected domains (Fig. [2b--e](#F2){ref-type="fig"}). Obviously, this expansion of the innermost core is mostly caused by domains which are involved in signal transduction and cell-cell contacts as well as cell development, suggesting that the demand to maintain a multicellular organism is the driving force for fusing protein domains. In particular, we observe that domains providing these functions, such as PDZ, efhand, SH2 and SH3 to name a few (Fig. [2b--e](#F2){ref-type="fig"}, Table [2](#T2){ref-type="table"}), increasingly populate the innermost cores. However, the affiliation of one domain to a certain core is not inevitably constant. In fact, we observe that pkinaseC which appears in the innermost core of the Yeast domain network occurs in the 2-core of the network of the multicellular organisms. The affiliation of HR1 to the innermost core of Yeast appears to be the effect of a single protein architecture which contains numerous domains, since this particular domain does not appear in the inner cores of the other organisms. Since the majority of Yeast proteins only has one to two domains, being teamed up with a reasonably large number of domains in a given protein is beneficial for a domain to make it to the innermost cores. In multicellular organisms, the number of proteins with a large number of domains increases, inevitably resulting in an inflation of the inner cores. Consequently, the evolutionary significance of a domain is well reflected in its ability to remain present in the inner cores of different organisms.
Domain interaction network
--------------------------
Information about protein domain interactions as of the InterDom database \[[@B21]\] constitute an undirected network of Yeast protein domain interactions. In contrast to domain co-occurrence networks, each link has a weight which reflects the frequency of the corresponding interactions relative to a random background distribution \[[@B21]\]. The degree distribution of the domain interaction network (Fig. [3a](#F3){ref-type="fig"}) is well fitted by a generalized Zipf-law *P*(*k*) \~ *α*(*β*+ *k*)^-*γ*^, suggesting that a few ubiquitous domains (hubs) dominate the web of domain interactions. The network\'s inherent modularity is expressed by the power-law form of the degree dependent clustering coefficient ⟨*C*(*k*)⟩ \~ *k*^-*δ*^(Fig. [3a](#F3){ref-type="fig"}, inset). In order to combine the impact of topology and weights, Barrat et al. \[[@B22]\] introduced a series of measures that allowed a more significant assessment of the impact weights have on the networks statistical properties. In a weighted representation of a domain interaction network the strength of a domain is the sum of the weights *w*~*ij*~carried on each link,  (Note, that the strength *s*~*i*~of a node *i*is the degree *k*~*i*~if we consider a network where all weights are 1). In the inset of Fig. [3b](#F3){ref-type="fig"}, we observe that the strength of the average interaction weakly decreases with increasing degree *k*. Assessing the distribution of weights, we define the average strength per link by . This measure allows us to observe a decreasing trend of *s*(*k*) with *k*(Fig. [3b](#F3){ref-type="fig"}) as a power-law, suggesting that domains with many interaction partners only occasionally interact with each partner.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Statistics of the domain interaction network of Yeast**. The domain interaction network has an average node connectivity of ⟨*k*⟩ = 16.9 along with a reasonably high degree of clustering ⟨*C*⟩ = 0.34. **(a)**The degree distribution of the domain interaction network displays a power-law, following the generalized Zipf-law *P*(*k*) = *α*(*β*+ *k*)^-*γ*^where *α*= 3,406.4, *β*= 67.4 and *γ*= 2.3. The network\'s inherent modularity is suggested by the presence of a power-law dependence in the average clustering coefficient ⟨*C*⟩ \~ *k*^-*β*^(inset), where *β*= 0.5. **(b)**The average strength *s*~*i*~of each interaction domain *i*displays a power-law (*s*~*i*~(*k*~*i*~) \~ ) over four decades. Obviously, this is an effect of a domains level of interaction, since we only recover a weak decrease of the strength *s*~*i*~toward higher degree *k*~*i*~, *s*~*i*~\~ *k*^-0.1^.
:::
:::
How is then the domain interaction network related to the domain co-occurrence network? In each core of the domain co-occurrence networks, we calculated the fraction of links present in the Yeast domain interaction network. Fig. [4a](#F4){ref-type="fig"} shows these frequencies of links for the eukaryotes *S. cerevisiae*, *C. elegans*, *D. melangaster*, *M. musculus*and *H. sapiens*, all displaying a decreasing trend that a fusion is accompanied by the physical interactions of domains when going from the innermost to the outermost core. Calculating the mean strength of domains based on the links that are present in the different cores by superimposing the respective weights *w*~*ij*~from the domain interaction network, we observe an ascending trend when nesting outwards toward the periphery of the domain co-occurrence networks (Fig. [4b](#F4){ref-type="fig"}). We interpret the observations that (i) domains which appear in the inner cores likely physically interact with (ii) a low average strength as follows: Domains with numerous interaction partners have an elevated chance of being fused in a higher eukaryote, while domains which interact frequently are less likely to be fused. Indeed, the innermost cores display the lowest average strengths, confirming that the driving force behind the fusion of a domain pair is not their frequent interaction, but rather the engagement of the two domains in a multitude of interactions with other domains.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Driving force behind fusion proteins**. **(a)**Nesting toward the innermost core for the eukaryotes *S. cerevisiae*, *C. elegans*, *D. melanogaster*, *M. musculus*and *H. sapiens*we find that the co-occurrence links increasingly coincide with links in the Yeast protein domain interaction network. **(b)**The interaction strength *s*of domains is the sum of the interaction weights of all links a domain is involved in the corresponding cores of the respective co-occurrence networks. Averaging over the size of the corresponding cores, the average interaction strength decreases toward the innermost cores.
:::
:::
Overlap of domain network cores
-------------------------------
Many of the domains appear ubiquitous to the innermost eukaryotic cores of the co-occurrence network (see e.g. Fig. [2](#F2){ref-type="fig"}). As already mentioned, the evolutionary and functional significance of a domain is indicated by its presence in the innermost cores of many organisms. Similarly, the conservation of links represents an evolutionary and functional signal. So far, it is unclear if *links*between these globally central domains have been preserved in all the eukaryotes. Fig. [5b--d](#F5){ref-type="fig"} shows the domain links simultaneously present in the four central cores of *S. cerevisiae*, *C. elegans*, *D. melanogaster*, *M. musculus*and *H. sapiens*. Note that all the links in Fig. [5](#F5){ref-type="fig"} have been preserved during evolution, suggesting the existence of a deeper reason why these domains seemingly always appear together in proteins. To be more specific: The central core of this domain-overlap network consists of a triangle set up by C2, pkinase and DAG-PE-bind (Fig. [5b](#F5){ref-type="fig"}, red nodes). Nesting outward through the cores, we find a further accumulation of triangles, all established by such prominent domains as PH, SH3 and RasGEF (Fig. [5b](#F5){ref-type="fig"}, yellow and red). In the subsequent overlap of cores (Fig [5c](#F5){ref-type="fig"} and [5d](#F5){ref-type="fig"}) we observe the presence of the important signaling domains zf-CCCH, zf-C2H2 and zf-C3HC4 from the zinc-finger family. Note that *all*of the (co-occurrence) links in the overlap networks correspond to physical interactions between the domains, *i.e*. the fraction of links in the overlap cores that are present in the interaction network is 1.0.
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Overlap of domain co-occurrence networks**. **(a)**We define the overlap of two networks as the edges, and their concomitant nodes, common to both networks. **(b)**The overlap of the four innermost *k*-cores of the co-occurrence domain graphs of *S. cerevisiae*, *C. elegans*, *D. melanogaster*, *M. musculus*and *H. sapiens*only shows a small number of conserved edges (red: 1-core, yellow: 2-core, green: 3-core, blue: 4-core). The overlap of the 1-cores consists of a fully connected kernel populated by signaling domains. Nesting outward in the overlap of the 2, 3, 4-cores (**(b),(c)**), domains that are responsible for signal transduction such as zinc-fingers and cell-cell contacts are dominating.
:::
:::
Discussion & conclusions
========================
Although the PFAM database provides comprehensive domain information, it covers only a part of the considered proteomes. Similarly, the determination of putative domain interactions depends on the quality and completeness of the underlying sets of protein interactions. Yet, the heterogeneity of scale-free networks indicates that the general characteristics of domain co-occurrence and interaction networks are independent of the webs actual size \[[@B17]\]. In particular, such networks are governed by the presence of highly connected hubs and cohesive areas, factors that not only influence their integrity but also the determination of *k*-cores. Since biological networks have been found to be stable upon random perturbations, we expect that the addition of new data will not dramatically impact our findings. The idea of analyzing the protein domain co-occurrence network as a sequence of nested cores and comparing the overlap between the central cores of eukaryotic organisms with increasing level of evolutionary development, gives new and fundamental insights into the qualitative arrangement and evolutionary utilization of the proteome. The evolutionary trend toward multicellularity requires proteomes capable of new and additional complex cellular processes such as signal transduction or cell-cell contacts. On a node based level, this trend toward higher complexity is reflected by an considerable heightened connectivity of domains that support such functions in multicellular organisms \[[@B15]\]. Turning our attention to a link-based level, panels in Fig. [2](#F2){ref-type="fig"} suggest an analogous result. The steadily increasing size of the innermost *k*-cores allows us to observe that the demand of maintaining complex cellular process does not only impact the level of single domains, but also operates on a combinatorial level of domain arrangements. Nevertheless, many protein families involved in inter- and intracellular signaling pathways, apoptosis \[[@B23]\], development, and immune and neural functions \[[@B12],[@B13]\], are indeed augmented in *H. sapiens*relative to *D. melanogaster*and *C. elegans*. Although human phenotypic complexity by far exceeds that of *D. melanogaster*and *C. elegans*, proteome dimensions remain surprisingly similar, allowing us to conclude that increased functional complexity is not simply a matter of proteome size but strongly underlines the role of innovations on the level of domain (re-) arrangements.
In fact, a significant portion of the protein architecture is found to be homolog in *H. sapiens*and *D. melanogaster*while substantial innovation in the creation of new protein architectures also has been detected \[[@B12]\]. The expansion of selected domain families and the accompanying evolution of complex domain architectures by joining presumably pre-existing domains coincides with the increase in the organisms level of evolutionary development. In particular, changes in the domain architectures are the consequence of a cellular mechanism commonly known as \'domain shuffling\', appearing in different disguises \[[@B20]\]. In simple cases of creating a new domain architecture, domains are simply inserted in already preexisting domain arrangements, a mechanism known as domain insertion while domain duplication refers to the internal duplication of at least one domain in a gene. Comparing domain architectures of proteins in multicellular organisms evidence emerged that preexisting domain architectures have been supplemented with single domains at their terminal sites, another mechanism that is known as domain accretion \[[@B13]\]. Our results do not favor one mechanism over the other. Yet, the panels in Fig. [2](#F2){ref-type="fig"} support the assumption that domain (re-)arrangements massively helped to evolve complex proteomes that are capable to maintain complex cellular processes, that have not been possible with the extension of single protein domain families alone. In the same way, network patterns we obtained from the comparison of cores which appear in all organisms under consideration will not tell us which mechanism predominantly gave rise to their emergence. Yet, we see that such small sized network patterns (see Fig. [5](#F5){ref-type="fig"}) presumably represent a repository of domain combinations around which the individual proteomes unfolded. These patterns predominately contain domains that play dominant roles in proteins which are essential for the inner workings of a multicellular organism, presumably serving as a possible backbone for the evolution of proteins mainly involved in signal transduction and cell-cell contacts.
The decomposition of the domain co-occurrence networks into *k*-cores allows us to uncover those sets of domains that are embedded in densely connected areas of the networks. The high connectivity as well as the nature of the partners those domains appear with indicate a central topological and functional role in the proteome of the considered organisms. Nesting toward the innermost cores the significance of these links is supported by the observation that pairs of co-occurring domains increasingly are present as physical interactions in Yeast. Utilizing the combined information of the co-occurrence network and the physical interaction network, we also find that domains tend to interact infrequently if they have many different interaction partners. In contrast, we observe that domains interact increasingly frequently once they have a small number of partners. Although we considered domain fusions on an indirect and qualitative basis this series of observations suggests that the driving force behind domain fusion events is not frequent interactions. In fact, it seems that the *number*of interactors, the connectivity, of the domains mainly influences a domains propensity to fuse with other interactors. The trend to spatially organize otherwise randomly diffusing domains might help to organize the flow of information in cells. Concluding, we find that domain fusion is a tool to superannuate the random diffusive interaction of a domain pair by embedding them in an architecture which ensures their interactions that would be difficult by random diffusion in a cell alone.
Methods
=======
Network representation
----------------------
An undirected unweighted network of *n*nodes is conveniently represented as an symmetric *n*× *n*adjacency matrix *A*= (*a*~*ij*~), where *a*~*ij*~= 1 if there exists an edge between nodes *i*and *j*and *a*~*ij*~= 0 otherwise. In a weighted network, the adjacency matrix reads as *A*= (*a*~*ij*~*w*~*ij*~), where *w*~*ij*~represents the weight of edge *ij*. Consistently, the degree being the number of neighbors a node *i*has is . As a generalization of a nodes degree the strength *s*~*i*~of a node i is defined as \[[@B22]\].
Proteome databases and domain co-occurrence network
---------------------------------------------------
The Integr8 database \[[@B24],[@B25]\] provides comprehensive statistical and comparative analyzes of the proteomes of fully sequenced organisms. Every predicted protein is annotated with the domains it contains, utilizing the combined efforts of different domain sequence sources. For our analysis, we focused on the domain data retrieved from the PFAM database, a reliable collection of multiple sequence alignments of protein families and profile hidden Markov models \[[@B26]\]. We construct the protein domain networks by considering all PFAM domains (or nodes) that are co-occurring in a protein to be a fully connected clique of undirected links (see Fig. la). In Integr8 we find 19,061 proteins that have a PFAM annotation in *H.sapiens*, as well as 18, 953 of *M. musculus*, 9, 785 of *D. melanogaster*, 12, 587 of *C. elegans*and 3, 791 of *S. cerevisae*. Although domain combinations *ij*potentially occur repeatedly in a proteome, we assign weight *w*~*ij*~= 1 to every link between domains *i*and *j*. Following this procedure, we generated domain networks for the proteomes of *H. sapiens*, *M. musculus*, *C. elegans*, *D. melanogaster*and *S. cerevisiae*.
Domain interaction data
-----------------------
The Interdom database \[[@B21],[@B27],[@B28]\] provides computationally derived putative domain interactions of Yeast. Based on PFAM domain information \[[@B26]\] for each set of protein interactions including pairwise protein interactions, protein complexes and Rosetta Stone sequences the presence of potential domain interactions is determined. The occurrence of a domain interaction in each protein interaction set is evaluated by comparing the observed frequencies to a random background model. A score thus obtained reflects the abundance of a particular pair-wise domain interaction, allowing the assessment of the reliability and the significance of the considered domain interaction. Considering these scores as weights *w*~*ij*~of interactions between protein domains *ij*, we generate an undirected network of 3, 353 domains that are embedded in 28, 339 weighted interactions.
Network degree distribution
---------------------------
The simplest way to characterize a network is by the degree *k*(or connectivity) of the nodes, reflecting the number of neighbors each node has. Accordingly, we define the average degree of a network as ⟨*k*⟩ = (1/*N*) *k*~*i*~, where *N*is the total number of nodes. Recent studies of biological networks have produced compelling evidence that the network degree distribution -- the probability that a node has *k*neighbors -- is scale-free with the functional form *P*(*k*) \~ *k*^-*γ*^\[[@B1],[@B9]\]. An important feature of the power-law distribution is the presence of a minority of nodes, carrying a vast number of connections, called \'hubs\'.
These hubs exhibit an increased propensity to be simultaneously lethal and conserved through evolution \[[@B17],[@B29],[@B30]\], thus playing a crucial role for the integrity of protein interaction networks.
Network clustering
------------------
Another important feature of biological networks is their tendency to exhibit cohesive areas: The clustering coefficient \[[@B31]\] of a node *i*measures the actual number of triangles that node *i*is a member of, relative to the possible number of triangles. Formally, it is defined as
where *n*~*i*~denotes the number of triangles. Accordingly, we define the average clustering coefficient as ⟨*C*⟩ = (1/*N*) *C*~*i*~. The clustering coefficient of a network also carries information about its modular nature, since ⟨*C*⟩ \~ 1 necessitates the presence of tightly interconnected clusters of nodes. Note that the network has a hierarchical architecture when ⟨*C*(*k*)⟩ \~ *k*^-*α*^, allowing the existence of discernible, yet topologically overlapping, functional modules. Apparently, networks with this structure are observed in most types of biological systems where a small subset of hubs play the important role of linking, and hence bridging, the various network modules \[[@B9],[@B32]\].
*k*-cores
---------
The *k*-core of a graph is defined as the largest subgraph for which every node has at least *k*links (Fig. [2](#F2){ref-type="fig"}): For each choice of *k*, we determine the *k*-cores by recursively pruning all nodes with degree lower than *k*and their incident links. In particular, we applied the following recursive algorithm: (1) sort nodes according to their present degree, and (2) remove the nodes with degree lower than *k*\[[@B18],[@B33],[@B34]\]. The layer *l*~*k*~of two consecutive cores *k*, *k*+ 1 is defined as the set of nodes that both cores do not have in common, i.e. nodes which only occur in the larger core. Since the innermost core does not have a successive core, we define the innermost core to be equivalent to a layer.
|
PubMed Central
|
2024-06-05T03:55:55.200633
|
2005-3-23
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079808/",
"journal": "BMC Evol Biol. 2005 Mar 23; 5:24",
"authors": [
{
"first": "Stefan",
"last": "Wuchty"
},
{
"first": "Eivind",
"last": "Almaas"
}
]
}
|
PMC1079809
|
Background
==========
During the last few years a surge in information on mitochondrial DNA control region sequence has been observed along with information from coding region substitutions in diverse populations of the world to understand their genetic diversity, structuring and origins \[[@B1]-[@B11]\]. More recently, mt DNA sequence data has been used to explore the peopling of Asia and to comprehend various population demographic parameters \[[@B12]-[@B20]\]. From a human genetics perspective India assumes importance in Asia because (1) the extensive diversity of populations residing in the sub-continent, which are biologically and culturally differentiated among themselves, (2) a clear distinction exists between non-tribal and tribal populations, autochthonous to the sub-continent \[[@B15],[@B21]\], lastly and more importantly this expanse is believed to be one of the early regions of settlement of modern humans \[[@B20]\]. Of the four major matrilines identified till date L, M, N and R, about 60% of Indians trace their maternal roots in Indian specific branches of haplogroup M that is reported to have emerged from the African haplogroup L3. With the exception of the M1 lineage that is confined to Ethiopia \[[@B22]\], all the other branches of this macrohaplogroup including M\*, C, D, G, E and Z haplogroups are observed in Asia \[[@B12],[@B14],[@B23],[@B24]\]. The lineages M2, M3, M4, M5, M6, M18 and M25 are exclusive to India, with M2 reported to be the most ancient lineage in the sub-continent with an age estimation of 60,000 yrs-75,000 yrs. Furthermore, the frequencies of these clades among the different geographic, linguistic phyla and social strata have been investigated in detail, yet the fundamental question regarding origin of this super-haplogroup still remains unanswered \[[@B15],[@B20]\]. While some authors have suggested a southwest Asian origin of M macrohaplogroup, followed by a back migration to Africa \[[@B15]\], others support its African ancestry \[[@B25]\]. One major drawback in arriving to a conclusion is the limitation of control region sequences in providing reliable estimate of phylogeny owing to homoplasy and recurrent mutations \[[@B23],[@B26]\].
Complete mitochondrial genome sequencing has gained importance in resolving phylogenies and understanding human evolution. Extensive genome sequencing studies have been carried out in different lineages of L, N and M major- haplogroups across different global populations. Though the phylogenies of East Asian counter parts of M lineages: M7, M8a, M8C, M8Z, M9, E, D, G have been resolved in detail, but till date no similar studies have been attempted on the sub-lineages of the Indian M haplogroup \[[@B9],[@B27]-[@B33]\]. The complete mt DNA sequence information from Indian M lineages will not only help answer questions regarding the origin and antiquity of this haplogroup, and resolving the phylogeny to finer branches but would also be highly relevant in understanding various parameters of population genetics, mitochondrial disorders and disease diagnosis and to some extent in forensic work \[\[[@B28]\] and references therein\].
The present study was undertaken to construct an unambiguous phylogeny for the M macrohaplogroup and to estimate its antiquity. Mitochondrial genomes were initially classified on the basis of their HVSI and coding region motifs, followed by complete sequencing of twenty-three samples representing different M matrilineals.
Results
=======
We have found seven group defining basal substitutions and described fourteen lineages in detail from complete mt DNA genome sequence information, which would be helpful in further resolving some of the Indian M lineages. The M trunk differs from revised Cambridge reference sequence (rCRS) with substitutions at A73G, A263G, T489C, A750G, A2706G, A1438G, A4769G, C7028T, A8701G, A8860G, T9540C, A10398G, C10400T, T10873C, G11719A, C12705T, C14766T, T14783C, G15043A, G15301A, A15326G and C16223T. The coding region mutation sites analyzed in the present study were different from those observed in the sister M1 lineage found in Ethiopia. The M phylogenetic tree constructed on whole mt sequence information of twenty four samples belonging to different M lineages and their sub-types including M1, M2, M2a, M2b, M30, M30a, M18, M\*, M3, M4, M5, M6, and M25 is provided in Figure [1](#F1){ref-type="fig"}.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Phylogenetic tree of M macrohaplogroup based on complete mt DNA genome sequences. Numbers along links refer to substitutions at nucleotide positions with respect to rCRS. Suffixes are transversions. Grey colour represents potential new lineage. Asterisk denotes unclassified lineage. The M haplogroup differs from the rCRS at sites: 73, 263, 489, 750, 2706, 1438, 4769, 7028, 8701, 8860, 9540,10398, 10400, 10873, 11719, 12705, 14766, 14783, 15043, 15301, 15326, 16223.
:::
:::
M2 lineage
----------
The complete sequencing of five mt DNA genomes belonging to M2 and its sub-lineages, M2a and M2b, indicated that coding region mutations T477C, T1780C, A8502G were associated with HVSI motifs C16223T and G16319A, which formed the root of M2 lineages. In case of M2a, we report a novel basal substitution at site T9758C in addition to previously reported transitions at G5252A and A8396G. Screening of T9758C site in twenty seven Indian individuals possessing M2a specific HVSI and coding region motifs, clearly establishes this mutation as a marker of this sub-lineage. The M2b sub-lineage having HVSI motif G16274A and T16357C, in addition to M2 defining mutations sites did not share any coding region substitutions with M2a. Overall, M2 lineage presents the maximum number of coding region mutation sites than any other analyzed lineage in this study.
M30 lineage
-----------
A new lineage M30 was recognized in the M macrohaplogroup, comprising of seven mt genomes, six of whose HVSI motifs did not correspond to any of the earlier established M lineages and one that was recently identified as M18 lineage (represented as shaded region in Fig [1](#F1){ref-type="fig"}). Since lineages of M have already been catalogued from M1 to M25, this potential new lineage is designated as M30 to avoid any ambiguity in classification of M macrohaplogroup. This branch arises from the main M trunk with transition at site G12007A. Finer resolution of this lineage was achieved by further clustering four complete sequences with mutation at two sites, T195A and G15431A into a sub-lineage designated as M30a. A total of eighteen Indian individuals belonged to M30a lineage. Three mt DNA genomes further branched out from M30 lineage, possessing only G12007A substitution. Interestingly, the newly described M18 (C16223, A16318T) matriline is one of the branches that directly emerged from M30 lineage. Sequence analysis of ten M18 mt DNA genomes showed the presence of G12007A transition. Eighteen Indian individuals were identified from our mtDNA database as possessing a HVSI motif (C16223T, A16300G) similar to the Saora sample, which arose as an offshoot of M30 lineage. All eighteen individuals tested positive for G12007A transition, suggesting that it might be acceptable to place M18 under M30 lineage.
M5 lineage
----------
The basal motif T12477C, G16129A and C16223T describes M5 lineage of majorhaplogroup M. Whole genome sequencing of three samples with similar HVSI motif, G16129A and C16223T, revealed that only one sample (I. B306) exhibited the T12477C mutation, and designated as M5a sub-lineage (Fig [1](#F1){ref-type="fig"}). This site was nevertheless absent in the other two samples, one of which had a similar HVSI motif as M5a mt DNA genome, while the other exhibited an additional site G16048A in its control region motif. Our study identifies a transversion at C10986A, shared by all the three samples, suggesting that these sequence types branched out from a common root. Analysis of C10986A substitution in seven Indian samples possessing HVSI motif, G16048A, G16129A and C16223T, confirms our finding that different branches emerged from M5 lineage. It is, however, important to note that G16129A might not be important in defining M5 and C10986A marker remains confined to this lineage until tested in a sufficiently large number of undefined M\*samples.
M25 lineage
-----------
The M25 lineage has been recently described by the presence of G15928A and T16304C. It differs from the M halogroup by five coding region substitutions and arises directly from the M trunk with no additional group defining motifs.
M31 lineage
-----------
The present analysis defines another new lineage that emerged from the main trunk of M macrohaplogroup, with a substitution at A5319G. Keeping in succession with the numbering of lineages, we designate this lineage as M31. This branch comprises of the previously well-defined M6 lineage along with M3 and M4 lineages.
The two M6 matrilines completely sequenced, harbor the characteristic group defining mutations at site T16231C, T16362C and C3539T. Our analysis deciphered another novel substitution at site A5301G in this lineage.
Whole sequencing of two M3 genomes (defined by T16126C transition) demonstrated a lack of similarity in their coding region substitutions. Since both T16126C and C16311T are fast mutating sites, we propose the non-existence of M3 and M4 lineages in the M phylogeny.
Three mt DNA genomes could not be differentiated in the study and were left designated as M\*. One of the M\* genomes was previously characterized as M3 lineage and differed in its coding region substitutions from the other M3, currently placed under M31 lineage. The matrilineal type possessing C16223T and T16325C as HVSI motif has been observed in relatively high frequency in Indian populations. Contrary to our expectation, full sequencing of this mt DNA (Ho69), did not exhibit the presence of G12007A mutation site that was observed in other unidentified M lineages analyzed in this study. Similar results were observed after complete sequencing of the HVSI motif type C16223T and C16251T.
The newly described phylogeny based on twenty-three Indian specific mt DNA sequence information radiates from the trunk of M macrohaplogroup with limited number of branches, differing in their basal motifs with one or more coding region substitutions. The branching pattern displayed by M2, M30, M31 and their sub-lineages mirrors the expansion model, indicating that most of the lineages described on basis of their control region sequence, occupy peripheral positions of the phylogeny. M2 is estimated to be the oldest lineage with an age of 49,686+/- 10,903 years before present (YBP). Furthermore, sub-lineage M2a differs from M2 by a minimum of six coding region substitutions, reiterating the antiquity of M2. The newly defined M30 and M31 lineages differ from the root of M by six to eight substitutions and hence, display similar expansion ages of approximately 33,042+/- 7,840 YBP.
Discussion
==========
Analysis of short stretches of mt DNA HVSI and HVSII region have significantly aided in clearly discriminating some of the M lineages. With the aim of understanding migration routes of diverse Indian people, more control region sequences are being generated without much support from coding region sites, resulting in an increasing number of conflicts within the classification of its lineages. We report a phylogenetic tree constructed from the whole genome sequencing of twenty-three Indian and one Ethiopian M lineage to resolve some of the anomalies occurring due to recurrent mutations in control region.
The control region sequences have exhibited the presence of an array of M lineages in India \[[@B12],[@B16],[@B20]\], despite which, complete mt DNA sequencing suggests that most of these lineages arose as limited offshoots of the main M trunk. The newly constructed M phylogeny displays difference in the branching patterns of lineages, with total number of substitution sites varying even within a lineage. Substantial variation in branch length within a lineage is indicative of the existence of further branching that could probably be delineated with generation of additional data. The M2 genome and its sub-lineages, more specifically M2a, have been well described in Indian population. Nevertheless complete sequencing of M2a demonstrated the presence of a novel site T9758C, which is characterized as a diagnostic marker for M2a sub-lineage, in addition to the previous reported, G5252A and A8396G transitions \[[@B15]\]. This finding reinforces the importance of sequencing a large number of individuals belonging to a lineage for describing a detailed phylogeny. The study was unable to trace any specific marker for M2b sub-lineage. In the absence of any lineage specific marker for M2b till date, it might be suggested that this sub-lineage is not a distinct clade of M2 lineage but is an M2 with additional HVSI motif (G16274A, T16357C). However, more M2b genomes need to be completely sequenced before reaching this conclusion and though HVSI sequences are not very reliable for constructing phylogenies, this cluster can well differentiate individuals with only one or both mutations and in turn resolve the phylogeny to its finer sub-lineages. M2 lineage is the oldest M lineage found in India with an estimated age of approximately 50,000 YBP, using only coding region motifs estimation, opposed to the expansion date of 60,000--75,000 yrs calculated from control region sequence information \[[@B15]\].
Although the G12007A substitution has been previously identified in other haplogroups, besides the M lineages \[[@B29]\], this study presents a novel lineage M30 that was differentiated to include mitochondrial genomes possessing G12007A substitution. The erection of M30a sub-lineage with its root at T195A and G15431A may help in further classifying M\* samples that have yet to be identified owing to the absence of any characteristic HVSI motif.
Mitochondrial genomes possessing the 16223, 16300 motif appear to be a promising new sub-lineage arising from M30. Additional complete mtDNA sequencing of similar sub-types may further help in precisely defining this branch. The M30 lineage was relatively younger than the M2 lineage with an expansion age of approximately 33,000 YBP, calculated on the basis of its coding region sequence information.
An important contribution of this study is placement of M18, M6 and previously defined M3 and M4 lineages in the M phylogeny. In the absence of a coding region marker for M18 lineage \[[@B20]\], G12007A substitution provides a stable root to M18 type, which is defined only on basis of the HVSI motif A16318T. The recognition of M31 lineage with an A5319G basal transition further reduces the number of branches arising from the trunk of M lineage. Since M6 is already well characterized, we propose that it remain as a sub-lineage of M31. However, it is essential that previously defined M3 and M4 lineages be completely removed from the phylogeny. Furthermore, it might be realized that the newly defined M4a lineage \[[@B20]\], might in fact be a sub-lineage or independent lineage by itself.
The M phylogenetic tree has largely aided in resolving the position of M5 lineage. Until recently, transition at G16129A along with basal motif of M, was used to characterize this lineage \[[@B34]\] and is currently described by the presence of coding region mutation at T12477C \[[@B20]\]. The phylogenetic tree constructed on the basis of complete mt DNA genome sequencing provides evidence to support our finding that at least two sub-lineages arise from M5 that share a transversion at site C10986A and may or may not possess T12477C transition. Presence of T12477C transition in only one of the two M5 mt DNA genomes sharing an identical HVSI motif, C16223T and G16129A, further substantiates the importance of coding region markers in precisely identifying mitochondrial phylogenies. Even though G16048A, HVSI motif has not been included under M5 owing to absence of T12477C, this study includes this motif under M5 lineage. However, prior to defining G16048A, G16129A and C16223T cluster, it is imperative that more samples representing this HVSI motif be completely sequenced. The age of M5 lineage is estimated to be 34,095+/- 6,425 YBP, indicating that M5 and its sister lineages M30 and M31 probably branched out from M haplogroup around the same time.
The newly defined M25 lineage did not share mutation sites with any other lineage and independently arose from M trunk with G15928A and T16304C substitutions.
The moderately high frequency of C16223T, T16325C HVSI motif types in Indian samples suggests that there might be a potential new lineage, which might be more accurately described once additional genomes possessing this motif are fully sequenced. In the absence of this sequence information, no attempt was made to classify this sequence type in to a lineage and hence, designated as M\*. The other M\* lineages bearing the control region motif, C16223T, T16126C and C16223T, C16251T, C16267T could not be resolved further for similar reasons.
Although, this study presents only a preliminary view of the M phylogeny, yet the emerging data may be highly useful in resolving the long-standing debate on Asian origin of M macrohaplogroup. Since M macrohaplogroup is derived from L3, which finds its roots in East Africa, it is believed that presence of M1 in Ethiopia further substantiates an African origin of M. A similar hypothesis had been drawn for the U6 lineage that is autochthonous to North Africa although U haplogroup displays its maximum diversity in Near East \[[@B35]\]. The authors of this study prepared an in-depth phylogeography of U6 to infer a back migration of this lineage from West Asia to North Africa. In the absence of a detailed M1 phylogeny, we have focused our attention on M2 to estimate the place of split of M from L3 as Africa or Asia. Interestingly, a single M2 genome differs in its coding region from the root of M at ten sites as compared to M1, which possess only four substitutions. Also, sub-lineages of M2, M5, M30 and M31 show long branch lengths, highlighting the deep roots of these lineages. Considering the antiquity of M2 and other East Asian specific M lineages \[[@B33]\], Ethiopian M1 lineage is by far a relatively newer branch. Our study on M1 and M2 mitochondrial genomes clearly established the Asian origin of M macrohaplogroup, followed by a back migration to Africa. We further suggest that as more M1 mt DNA genomes are sequenced, there is a possibility that this lineage might find its root in one of the peripheral branches of Asian M lineage.
Conclusion
==========
This study presents the, first ever, phylogenetic tree constructed for the M lineages predominant in India. The phylogenetic tree emerged from the present study consists of seven branches. A significant achievement of this endeavor is the recognition of M30 and M31 lineages, which encompass seven and three sub-lineages respectively. While the previously defined lineages, M18 and M6 are now found to be sub-lineages of M30 and M31; our study suggests the non-existence of M3 and M4 as independent lineages. Thus, the present analysis also determines the roots of few previously undefined mt DNA genomes, reinforcing the importance of screening lineage defining coding region substitutions before describing a lineage. Another significant conclusion emerged from this study is substantiating the Asian origin hypothesis of M macrohaplogroup. This study would provide baseline information for precisely describing the macrohaplogroup M, as more complete mt DNA sequence information is generated. Furthermore, it establishes the superfluousness of control region diversity reported in the M macrohaplogroup by suggesting that most of M lineages might in fact be derived from limited basal branches.
Methods
=======
Amplification and sequencing of HVSI and coding region motifs
-------------------------------------------------------------
Genomic DNA was extracted from whole blood by standard Phenol/chloroform method. Amplification and sequencing of control region were performed in 1258 samples (authors unpublished data) as described in our earlier studies \[[@B36],[@B37]\]. Samples were initially analyzed for substitutions at site 10397 and 10400 via RFLP protocol before characterizing them under M haplogroup and subsequently clustered into different lineages of M via sequencing of the required coding region fragments- T447C, T1780C and A8502G for M2, G5252A for M2a, T12477C for M5, C3539T for M6 and G15928A for M25 \[[@B15],[@B20]\].
Complete mtDNA sequencing
-------------------------
Socio-cultural affiliations of twenty-three individuals selected for complete sequencing are presented in Table [1](#T1){ref-type="table"}. These mt DNA genomes belonging to M\* (n-8), M2 (n-1), M2a (n-3), M2b (n-1), M3 (n-2), M4 (n-1), M5 (n-3), M6 (n-2), M18 (n-1) and M25 (n-1) sub-lineages of Indian macrohaplogroup M were completely sequenced in the current study. DNA amplification and sequencing was carried out using primers described elsewhere \[[@B38]\]. New group defining substitutions were re-sequenced, and their frequency was determined in other Indian samples with similar control region motifs. Each sample was completely sequenced twice to remove any ambiguous sites. Additionally, since several segments of the same mt DNA had to be screened, care was taken to avoid artificial recombination caused by potential crossovers.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Detailed population information of 23 mitochondrial genomes analyzed in current study.
:::
S.No. Sample ID Population Social status Ethnicity Geographical origin Linguistic Affiliation
------- ----------- ----------------- --------------- ------------ --------------------- ------------------------
1 Gow96 Gowda Caste Australoid Karnataka Dravid
2 Kur150 Kuruva Tribe Australoid Karnataka Dravid
3 Kur126 Kuruva Tribe Australoid Karnataka Dravid
4 Chr252 Christian Caste Australoid Karnataka Dravid
5 Bho134 Bhovi Caste Australoid Karnataka Dravid
6 Mus112 Muslim Caste Australoid Karnataka Dravid
7 Mus114 Muslim Caste Australoid Karnataka Dravid
8 IB306 Iyengar Brahmin Caste Caucasoid Karnataka Dravid
9 Lyn180 Lyngayat Caste Caucasoid Karnataka Dravid
10 Chen Chenchu Tribe Australoid Andhra Pradesh Dravid
11 Lam18 Lambadi Tribe Caucasoid Andhra Pradesh Indo-European
12 Lam8 Lambadi Tribe Caucasoid Andhra Pradesh Indo-European
13 N.Gond Naikpod Gond Tribe Australoid Andhra Pradesh Dravid
14 Kom4 Komati Caste Australoid Andhra Pradesh Dravid
15 B.Kur Kurmi Caste Caucasoid Bihar Indo-European
16 B.Ban Baniya Caste Caucasoid Bihar Indo-European
17 B.Yad Yadav Caste Caucasoid Bihar Indo-European
18 Raj90 Rajput Caste Caucasoid Bihar Indo-European
19 Katk Katkari Tribe Australoid Maharashtra Indo-European
20 Paw50 Pawara Tribe Australoid Maharashtra Indo-European
21 Sao Saora Tribe Australoid Orissa Austro-Asiatic
22 Mah6 Mahali Tribe Australoid West Bengal Austro-Asiatic
23 Ho69 Ho Tribe Australoid Jharkhand Austro-Asiatic
:::
Phylogenetic analysis
---------------------
The sequence information generated by whole mt genome sequencing of twenty Indian specific M lineages was used to construct a phylogenetic tree of macrohaplogroup M. Substitutions were reported with respect to the revised Cambridge Reference Sequence (rCRS) \[[@B39]\]. Only those mutation sites occurring in all sequenced genomes of a maternal lineage were placed at the root. The complete sequence of the Ethiopian M1 genome \[[@B25]\] was included in the M Phylogenetic tree to determine its relationship with its sister lineages.
Time estimate
-------------
Age estimates for the M lineages were computed using only coding region substitutions identified from the complete mt DNA genome sequencing. The mean number of mutations per site to the most recent common ancestor was estimated and converted to real time using a substitution rate of 1.26 × 10^-8^per site per year \[[@B40]\].
Authors\' contributions
=======================
RR, JB and HB carried out carried out extensive sequencing, RFLP experiments and also analyzed the genetic data. RR did phylogenetic analysis and drafted the manuscript. RT provided critical and valuable information during processing of data. VKK is responsible for conceiving and designing of the study and contributed significantly in interpretation of data and shaping of the manuscript. All authors read and approved the final manuscript.
Electronic reference
====================
Rajkumar R, Banerjee J, Gunturi HB, Trivedi R, Kashyap VK: Phylogeny of the M superhaplogroup inferred from complete mitochondrial genome sequence of Indian specific lineages. Genome Biology 2004, 6:P3 <http://genomebiology.com/2004/6/2/P3>
Acknowledgements
================
We are deeply indebted to Richa Ashma, Sonali Gaikwad, Sanghamitra Sahoo and Anamika Singh for allowing us to use the information of complete mt DNA genomes sequenced and analyzed by them. The present study would not have been possible without their contribution. A special thanks is extended to T Sitalaxhmi for reconfirming mutation sites. We express our appreciation to the blood donors analyzed in the present work. This work was supported by a research grant under the IX Five Year Plan to CFSL, Kolkata. RR and JB were assisted with DFS, and HBG with CSIR (New Delhi), research fellowships.
|
PubMed Central
|
2024-06-05T03:55:55.205126
|
2005-4-2
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079809/",
"journal": "BMC Evol Biol. 2005 Apr 2; 5:26",
"authors": [
{
"first": "Revathi",
"last": "Rajkumar"
},
{
"first": "Jheelam",
"last": "Banerjee"
},
{
"first": "Hima Bindu",
"last": "Gunturi"
},
{
"first": "R",
"last": "Trivedi"
},
{
"first": "VK",
"last": "Kashyap"
}
]
}
|
PMC1079810
|
Background
==========
Normally, aging results in the natural senescence of multiple organ systems including the kidney \[[@B1]\], the autonomic nervous system \[[@B2]\], and the heart \[[@B3]\]. While numerous mechanisms are involved in age related changes in the body, one important factor contributing to decreased function is a reduction in nitric oxide production (a potent vasodilator) in tissues \[[@B4]\]. In addition, there is a reduction in beta adrenergic receptor sensitivity associated with the aging process \[[@B5]\], reducing the ability of the sympathetic nervous system to respond to stress.
Diabetes, similar to the aging process, can be associated with autonomic nervous system impairment including endothelial dysfunction and impairment of autonomic neurons \[[@B6],[@B7]\]. This reduces function in both the sympathetic \[[@B8]\] and the parasympathetic nervous systems \[[@B9]\]. It can include loss of parasympathetic peripheral nerves and damage to autonomic ganglia, both of which are believed caused by vascular endothelial function \[[@B10],[@B11]\]. Thus when the body is placed under stress, almost 30% of all people with diabetes show autonomic impairment. When autonomic stressors are combined, such as orthostatic changes and heat exposure together, nearly all patients with diabetes showed severe autonomic impairment \[[@B7]\].
The common mechanism for many of these changes with aging and diabetes is the inability of the blood vessels to adequately dilate. This leaves a predominant vasoconstrictor tone, causing inadequate blood supply to the skin and various organs of the body \[[@B12],[@B13]\].
Endothelial cell function can be assessed by the change in blood flow after occlusion of the circulation \[[@B14]\]. Since the vasodilatation that results from local anoxia is entirely dependent on endothelial cell function and independent of the autonomic nervous system, this technique shows the effect of aging and diabetes only on vascular endothelial cells without the complication of involving other elements of the nervous system. The relationship between age, diabetes and vascular reactivity to occlusion should provide some evidence as to the interrelationships between these factors. This was one purpose of the present investigation.
Thiazolidinediones have been shown to increase blood flow in patients with type 2 diabetes by increasing nitric oxide production in vascular endothelial cells \[[@B28]\]. Further, administration of rosiglitazone, a member of this family (RSG), has been shown to increase resting forearm blood flows in both non diabetics and people with diabetes \[[@B15]\]. Thus, in the present investigation, a series of experiments was accomplished on patients with diabetes to assess the interrelationships between age, diabetes and RSG administration for 3 months to see if RSG will improve either the diabetes or age related loss in vascular reactivity to occlusion or both.
Methods
=======
Subjects
--------
The subjects with diabetes in this study were 8 men and 8 women, all with type 2 diabetes. The mean age, heights, BMI (Body Mass Index) and weights are shown in table [1](#T1){ref-type="table"} at the onset of the study. There were 15 on ACE inhibitors, 50% were on a statin, 3 were on beta- blockers and all but 1 were non smokers. Fifty percent were classified as having coronary artery disease and 50% had microalbuminuria. Sixty six percent had neuropathies, and 75% had retinopathies. Mean baseline values were: HbA1c: 8.8+/-2.4, Total cholesterol: 213+/-80, LDL: 125+/-80, HDL: 45+/-18, and Triglyceride: 275+/-274. Blood pressure was 132+/-38 systolic and 82+/-29 diastolic. Fourteen of the 16 were identified with hypertension.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
demographics of subjects with diabetes
:::
Number Age (years) height(cm) weight(kg) BMI HBA1C
------------ -------- ------------- ------------ ------------ ------ -------
men 8
mean 60.0 185.1 101.8 29.6 9.0
sd 19.9 12.5 24.2 5.7 2.8
women 8
mean 63.8 161.9 83.2 31.1 8.4
sd 11.7 6.3 27.3 7.9 1.5
group mean 16 61.2 178.0 96.1 30.0 8.8
sd 10.2 15.6 26.7 6.3 2.4
:::
Analysis of 3 month values revealed significantly lower HbA1c: 7.0 (P \< .05) and lower trends in cholesterol and blood pressure: total cholesterol 206, LDL: 117, HDL: 42.9, and Triglyceride: 280. Blood pressure was 126+/-34 systolic and 72+/-22 diastolic. The reduction in both systolic and diastolic blood pressure was significant.
All medications were kept constant throughout the study.
The younger volunteers represented a healthier cohort than the patients with diabetes. None of the subjects had metabolic syndrome, and none smoked. They had a lower BMI, weight, and were about 5 years younger than the subjects with type 2 diabetes. None took any type of medication. Their average resting blood pressure was 121+/-34 systolic and 75+/-28 diastolic. For the control group, the BMI in the younger subjects (less than 45 years) was 23.2+/-6.1 while the older subjects (\>45) averaged 27.9+/-5.3. Both younger and older groups of controls were all physically active faculty and staff at Loma Linda University, School of Allied Health.
Measurements
------------
Forearm blood flows -- Forearm blood flows were measured by Whitney strain gauge plethysmography. Whitney strain gauge plethysmography is a technique of measuring limb blood flow by volume plethysmography, a full description is given elsewhere \[[@B14],[@B16]\].
Procedures
----------
Muscle temperature varies in the forearm between 27 and 42 deg C. This is due to the fact that the forearm is a shell tissue and temperature varies to help gain or loose heat from the core of the body \[[@B17]\]. Limb tissue temperature varies with room temperature, clothing, body fat content, and the phase of the menstrual cycle \[[@B18]\]. Therefore, a thin person may have resting arm metabolism less than 20% of that of a person with a high body fat content due to the high Q10 of the tissues \[[@B19]\]. Therefore, to remove some of the variability in previous studies, a water bath was used to elevate all forearm temperatures to that of the core, 37 deg C. Subjects placed their arms in a bath heated to core body temperature of 37 deg C with the arm held dependant and the elbow at an angle of 90 degrees such that their arms were submerged to the belly of the biceps muscle. The bath was well stirred. After 15 minutes, resting arm flows were recorded. A 4 min period of arterial occlusion was then induced by a cuff under the axilla and inflated to 200 mmHg was then used. The cuff was then released and blood flow was measured for another 2 minutes. All experiments were repeated on controls and subjects with diabetes prior to administration, after 2 and 4 weeks, and 3 months on RSG. Measurements on the control subjects were repeated at 1 and 3 months.
Statistical analysis
--------------------
Statistical analysis involved the calculation of means, standard deviations, T tests, and analysis of variance (ANOVA). The level of significance was p \< 0.05. All data in \"results\" is expressed ± the standard deviation. Regression lines were calculated by the method of least squares.
Results
=======
The results of the experiments are shown in figures [1](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}, [3](#F3){ref-type="fig"}. Figure [1](#F1){ref-type="fig"} illustrates the blood flow in the control subjects and subjects with diabetes prior to administration of RSG after 4 min of vascular occlusion. Prior to occlusion, the average blood flow in the control subjects was 2.24 +/- 0.64 cc/100 cc tissue/min. After occlusion, for the control subjects, blood flow peaked at 26.8 +/- 4.86 cc/100 ml tissue/min for the first flow approximately 3 seconds after vascular occlusion; 2 minutes post occlusion, the flow was back to the initial resting flow. In contrast, for the subjects with diabetes, blood flow at rest was only 1.0 +/- 1.02 cc/100 ml tissue/min. After 4 minutes of occlusion, the peak flow only averaged 6.4 +/- 2.82 cc/100 ml tissue/min and after 1 minute post occlusion, flow was 1.5 +/- 1.42 cc/100 ml tissue/min. The total blood flow response to occlusion was significantly less in the subjects with diabetes (p \< 0.01). Since the duration and magnitude of the blood flow response to occlusion were different in the subjects with diabetes compared to control subjects, a separate calculation was made, the excess flow. This was accomplished by calculating the total blood flow above the normal resting flow observed in the 2 minutes following vascular occlusion. For the control subjects, the average excess flow was 56.58 +/- 12.57 cc/100 ml tissue of blood whereas for the subjects with diabetes the excess flow was 13.6 +/- 8.01 cc/100 cc tissue resulting from the occlusion, the data in the control group was significantly higher than for the group with diabetes (p \< 0.01).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
This figure shows the blood flow recorded for 2 minutes following the release of an arterial occlusion cuff on the brachial artery in control subjects (diamonds) and subjects with diabetes (squares). Illustrated here are the average results for all subjects in each group +/- standard deviation. Flows are expressed in cc/100 ml muscle per minute and the time scale on the bottom is the flow number. Flows are recorded every 12 seconds starting at 3 seconds post occlusion.
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
This figure shows the blood flow in the forearm of the subjects with diabetes at rest (cc/ 100 ml tissue/minute) before taking rosiglitazone (pre) and at two weeks (2 w), four weeks (4 w), and three months (3 m) taking 4 ml per day of Rosiglitazone. All data is shown with the appropriate standard deviation.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
This figure illustrates the excess flow above rest during a two minute period after the release of an occlusion cuff on the brachial artery of control subjects and subjects with diabetes. Individual data points are shown on control subjects and subjects with diabetes prior to initiation of a three month administration of rosiglitazone (pre rsg) and after 2 weeks (2 w), four weeks (4 w), and three months (3 m). Data is plotted in relationship to the age of the subject. The regression lines are calculated by the method of least squares. On the control subjects the regression line was flow = -0.518 age + 77.78. Prior to administration of rosiglitazone of the subjects with diabetes the regression equation was flow = -0.253 age + 31.01. After two weeks of administration of rosiglitazone the regression equation was blood flow = -0.348 age + 41.6. After 4 weeks of rosiglitazone the regression equation was blood flow = -0.243 age + 50.27. Finally after 3 months of rosiglitazone was post occlusion flow = -0.274 age + 64.94
:::
:::
After subjects were placed on 4 mg RSG for 3 months, resting blood flow increased. Resting flows in subjects with diabetes, when compared to the control subjects, were significantly less prior to administration of RSG (p \< 0.01) and after 2 weeks and 4 weeks this difference was preserved (p \< 0.05). However, after 3 months, while the resting flows were still numerically less in the subjects with diabetes, due to the large variance and small number of subjects, there was no statistical difference in resting blood flows between controls and subjects with diabetes (p \> 0.05). The total excess flows increased before and after 2 weeks, 4 weeks and after 3 month administration with RSG. Blood flows were 13.06+/-8.01, 19.01 +/-8.00, 35.2+/-13.4 and 42.08 +/-20.08 cc/100 cc tissue. Before and at 3 months, the flows, when compared to the controls, were significantly less (p \< 0.05).
Since aging has a known effect on resting and exercising blood flow, the data in figure [3](#F3){ref-type="fig"} has been plotted in relationship to age. For the control subjects, for example, as shown at the top of the figure, there was a linear decrease in post occlusion flows (as well as resting flows) associated with aging. The regression line through the data points in fig [3](#F3){ref-type="fig"} showed that for every year a subject aged, they lost 0.51 cc of blood flow per 100 g tissue per minute in other words, a loss of approximately 5 cc in post occlusive reactive hyperemia for every 10 years increase in age.
For the subjects with diabetes, prior to taking RSG, there was also an age related loss in excess flow as shown in the figure. The slopes of the lines between the non-diabetic and diabetic subjects prior to taking RSG were not statistically different from each other (p \< 0.01). Thus for subjects with diabetes, there was a similar reduction in post occlusive hyperemia with aging to a greater extent than in control subjects. After 2 and 4 weeks on RSG, there was still the same aging effect on the reduction of post occlusive flow however; difference between the groups was reduced. After 3 months there was no significant difference in flows in controls and patients with diabetes when related to age (p \> 0.05).
For the controls, the correlation between age and excess flow was -0.56, a significant correlation (p \< 0.01). With an R2 of 0.31, 30% of the change in flow after occlusion could be accounted for by age alone. In the subjects with diabetes, the correlation between age and flow was 0.71 with an R2 of .49 before RSG. Two weeks after RSG the correlation was 0.61, after 4 weeks was 0.57. After 3 months it was 0.45. Thus, diabetes seems to potentate ageing. Before RSG, 20% more of the variability in flows with age was observed. After 3 months on RSG, the contribution of ageing to blood flow was the same.
Discussion
==========
Both aging and diabetes are associated with a loss in function of the cardiovascular system. One commonly reported factor in both is a reduction in the ability of the peripheral vasculature to vasodialate \[[@B1],[@B2],[@B20]\]. Normally, with the arm at rest, there is always both tonic vasoconstrictor and vasodilator tone to the vascular endothelial cells. In the skin, for example, there is a slow tonic release of norepinephrine, a vasoconstrictor, and acetylcholine, a vasodilator \[[@B21]-[@B23]\]. If there is an increase in vasoconstrictor activity, then vasoconstriction predominates and if more vasodilator activity is present, then vasodilatation predominates \[[@B24]\]. Thus, even reflex changes in blood flow can be modulated up or down by central sympathetic constrictor or dilator activity \[[@B24]\].
In the present investigation, resting blood flow and the blood flow after vascular occlusion was reduced with aging and diabetes. These results agree with that shown by others \[[@B15],[@B25]\]. At first examination, it would appear that a similar mechanism might be involved in reducing vascular reactivity with age and diabetes. Certainly, in the present investigation, older subjects with diabetes had a greater reduction in resting flows and post ischemic flows than was seen in younger subjects. A simple loss in the ability to vasodilate would explain both the reduced post ischemic flows and the shorter duration of the flows. If vasodilatation were impaired, then blood flows would be less after ischemia and, when flows began to return to pre ischemic levels, the dominant vasoconstriction would shorten the hyperemia as observed here. But the hyperemia was not shortened with age for control subjects, only subjects with diabetes suggesting different mechanisms causing the reduction in blood flow between age and diabetes.
Part of the mechanism for reduced resting and post ischemic blood flow with both aging and diabetes has been linked to a reduction of the vascular endothelial cells ability to produce nitric oxide, a potent vasodilator substance \[[@B4]\]. This may be due to a defect in nitric oxide synthesis, decreased nitric oxide sensitivity or reduced availability of l-arginine, the precursor of nitric oxide. Nitric oxide is released by vascular endothelial cells in response to a variety of stimuli including sympathetic vasodilator nerves \[[@B4]\]. But here is where the similarity may end.
Data presented here make it unlikely that ageing and diabetes effect endothelial function by a similar mechanism. In the graph in figure [3](#F3){ref-type="fig"}, the relationship between age and cardiovascular reactivity (excess flows) was linear. For the subjects with diabetes, there was no statistical difference in the slopes but the intercept was very different; 77.78 cc for the control subjects and 31.01 cc for the subjects with diabetes. Simply looking at these 2 groups of subjects, it might be inferred that diabetes potentiated the loss in endothelial function with age. And yet, after 3 months on RSG, there was no statistical effect of diabetes alone on vascular endothelial function but ageing still had a pronounced effect. If a similar mechanism were involved, i.e. loss of NO in both, it seems too coincidental that the curves remained parallel relating age to endothelial function in subjects with diabetes and after 3 months on RSG; the data from the diabetic subjects would exactly match the controls.
It seems more likely that 2 separate mechanisms are involved. While there is a loss in nitric oxide production with age, it is not clear if this is the sole culprit in causing a loss of endothelial function with age. The studies on rosiglitazone also point to this possibility. Recent studies show at least 2 other mechanisms associated with ageing. First, there is a loss in beta-adrenergic receptor sensitivity with ageing due to structural changes in the receptor making it insensitive to catacholamines \[[@B5]\]. Since the response to occlusion is not mediated by sympathetic nerves, this probably plays little role here. But in a recent study \[[@B26]\], data was presented which shows that nitric oxide may not be the predominant factor in controlling skin circulation after occlusion. These investigators found that the relative NO-dependent portion of cutaneous active vasodilatation accounted for 23% of vasodilatation in young subjects. Other factors were responsible for most of cutaneous vasodilatation. Even in older subjects, nitric oxide contributed to only 60% of the vasodilatation. Thus, other mechanisms were involved and possibly other relaxation factors altering smooth muscle tone in the vascular beds \[[@B27]\].
Rosiglitazone is a PPAR gamma ligand that has been shown to increase skin circulation \[[@B15],[@B28],[@B29]\] by increasing nitric oxide production \[[@B28]\]. The fact that nitric oxide increases with administration of rosiglitazone and yet the aging effect on endothelial function was not altered supports the concept of another mechanism involved here. The improvement in glucose control could also contribute to the improvement in endothelial function seen in this experiment. Finally, the use of Beta blockers in a few of the subjects and physical activity may have had some effect on these results. Hypertension, in itself may have caused endothelial damage here. However, there was no difference in the group of subjects with beta blockers and the smokes in their response to RSG. While it is well accepted that the response to vascular occlusion is a means of examining endothelial function \[[@B30]\], and that nitric oxide production is reduced in diabetes \[[@B31]\] and with ageing in vascular endothelial cells \[[@B32]\], the exact mechanism of the damage, e.g. nitric oxide, prostaglandins etc can not be fully elucidated without additional studies. More detailed studies are needed to understand these mechanisms.
Conclusion
==========
**1.**Aging is associated with a reduction in both resting forearm blood flows and post ischemic blood flows.
**2.**Diabetes further exacerbates the cardiovascular damage, reducing vascular endothelial cell reactivity.
**3.**Administration of RSG and improvement of glucose control appear to reverse the damage to vascular endothelial cells associated with diabetes but not aging.
**4.**The reduction in endothelial cell reactivity to occlusions with aging and diabetes are probably due to different mechanism.
Abbreviations
=============
RF- Resting forearm blood flows
RSG- Rosiglitazone
PF- Post occlusion blood flows
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
Dr Jerrold Petrofsky developed the studies, collected data, analyzed the data and drafted the manuscript. Dr Scott Lee developed the studies, recruited subjects and helped draft the manuscript. Miss Cúneo collected data and analyzed data and helped draft the manuscript. All authors read and approved the final manuscript.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Demographics of Control Subjects
:::
number Age (years) Height (cm) weight (kg) BMI
------- -------- ------ ------------- ------------- ------------- ------
men 19 mean 54.0 176.2 86.3 24.5
sd 18.2 9.7 16.7 30
women 11 mean 57 168.1 80.1 28.3
sd 19.1 11.4 16.7 4.7
group 30 mean 55.2 174.2 84.2 26.4
sd 22.3 17.2 22.3 5.7
:::
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6823/5/4/prepub>
Acknowledgements
================
This work was supported in part by a grant from Galaxo Smith Kline. 0313473100.
|
PubMed Central
|
2024-06-05T03:55:55.207275
|
2005-3-24
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079810/",
"journal": "BMC Endocr Disord. 2005 Mar 24; 5:4",
"authors": [
{
"first": "Jerrold",
"last": "Petrofsky"
},
{
"first": "Scott",
"last": "Lee"
},
{
"first": "Maria",
"last": "Cuneo"
}
]
}
|
PMC1079811
|
Background
==========
Clinically significant psychosocial and mental health problems affect up to 20% of children and youth presenting for primary care services\[[@B1]\] (we will refer to children and youth as \'children\' for purposes of brevity). Primary care physicians, chiefly pediatricians in the United States and parts of Canada, and general and family practitioners (GP/FPs) in other parts of the world, are increasingly expected to participate in, if not assume responsibility for, the care of these children\[[@B2],[@B3]\]. While GP/FPs feel confident in managing common mental health concerns of adults\[[@B4]\], much less is known about their role or about their level of confidence in working with mental health concerns of children, such as depression, attention-deficit/hyperactivity disorder (ADHD), and behavioral and emotional problems\[[@B5]\]. Such information would aid the design of effective health care services for these children, and relevant educational programming for GP/FPs.
The literature relevant to these questions is limited and difficult to interpret. US studies tend to group GP/FPs and pediatricians together, while studies specific to GP/FPs, from European countries and Australia, often focus on approaches to a single condition. The latter group of studies show that GP/FPs frequently lack confidence in relation to ADHD\[[@B6]\], are reluctant to diagnose and manage cases\[[@B7],[@B8]\], and are doubtful that it can be managed solely within primary care.\[[@B9]\] These GP/FPs express a higher level of confidence in care of children with emotional problems, depression and conduct disorders.\[[@B6],[@B9]\] In US studies, primary care physicians are generally not comfortable with the treatment of children with depression, \[[@B5],[@B10]\], and are more likely to refer children with emotional problems than with ADHD to specialized mental health services\[[@B5]\]. Interestingly, GP/FPs tend to be more comfortable than pediatricians in working with children with depression, and more likely to treat such cases themselves\[[@B11]\].
Gender and self-confidence are among the factors that influence GP/FP practice with mental health issues. Female physicians report care of children\'s psychosocial problems to be less burdensome than do male physicians\[[@B12]\], and, in the management of depression among children and youth, female pediatricians are somewhat more likely to provide counselling and involve other family members \[[@B10]\]. Physician confidence is associated with making fewer referrals to specialists, at least for adult depression \[[@B4]\]. However, little research has examined the factors that affect whether GP/FPs take primary responsibility for behavioral and emotional concerns in children, rather than referring for specialist care, or factors that affect the GP/FPs\' self-efficacy in dealing with these children.
To address these issues, we surveyed a representative sample of GP/FPs in the province of British Columbia (BC), Canada. Our aims were to ascertain the preferred level of involvement of GP/FPs in a range of common psychosocial and mental health problems among children, their perceived comfort and competence in dealing with these problems, and the factors associated with differences in levels of involvement and comfort and skill. We enquired about GP/FPs\' involvement with conditions that would be included in both formal diagnostic nomenclature (ADHD and mood disorders), and the more loosely-defined set of conditions described collectively as behavior problems and social-emotional difficulties. The latter terms have been commonly used in large scale surveys of primary care practitioners \[[@B5]\], and include a range of psychosocial problems that are mild or \'subthreshold\' for formal psychiatric diagnosis \[[@B5]\].
To understand the health services context of this study, GP/FPs provide for most of children\'s primary care needs in BC, with pediatricians and psychiatrists providing mainly consultative services. The costs of \'medically necessary services\' for residents of the province are covered under a universal health insurance coverage plan known across Canada as Medicare, although BC residents who earn above a certain level must contribute to annual premiums. Most physicians are paid on a fee-for-service basis through the provincial Medical Services Plan, based on specialty and service-specific fee schedules. Patients have access to any medical practitioner licensed to practice in BC, but specialists require a referral from a family practitioner to obtain specialist rates of pay. There are also mental health clinics throughout the province, most of which require a referral from medical practitioners or other health care professionals. Access to specialized mental health services is significantly limited by a shortage of these services, especially in rural areas, where pediatrician consultation may the only medical resource available to GP/FPs for child mental health problems. Access to psychologists\' services is also limited because these are not covered by Medicare, and not all residents carry insurance for these services. GP/FPs are only remunerated for a limited number of appointments per patient for counselling, which may affect their interest in ongoing treatment.
Methods
=======
Study design
------------
Mail out survey of GP/FPs practicing in British Columbia, Canada.
Participants
------------
In order to obtain a final respondent sample of 10% of eligible GP/FPs in the province, we randomly selected 801 names from a total of 3,953 in a registry provided by the College of Physicians and Surgeons of BC after stratification by Health Region (n = 20) to ensure a representative geographic sample. To participate in this study, physicians had to be in active general or family practice, and to see at least five children per month. From the College registry, we obtained each GP/FP\'s name, office address and phone number; medical specialty status, including certification in Family Medicine from the College of Family Physicians of Canada; date of birth; date of graduation, and university of graduation. Ethical approval for this study was obtained from the University of British Columbia Behavioral Research Ethics Board.
Survey instrument
-----------------
The authors developed a 22-item survey questionnaire, through consultation and consensus, to cover five areas. In relation to children with four clinical conditions, namely behavior problems, social-emotional difficulties, ADHD, and mood disorders, we enquired about (1) number of patients presenting in a typical month; (2) the GP/FP\'s preferred approach to involvement; and (3) his/her comfort/skill and effectiveness in dealing with each condition. In addition, we assessed (4) general beliefs about children with psychosocial problems; and obtained (5) information about demographic and practice characteristics. Because our expressed aim in this survey was \"to learn about how primary care physicians in BC experience providing care to children and youth with behavioral and emotional problems\", we did not enquire as to details of the physicians\' diagnostic or treatment methods.
Behavior problems were defined as including \"disruptive, non-compliant, aggressive, antisocial, oppositional and overactive behaviors\"; and social-emotional difficulties as including \"low self-esteem, social withdrawal and fearfulness\"; and mood disorders as including \"anxiety and depression\". Our intention in using these terms was to allow the kind of latitude and subjectivity that physicians are accustomed to in their use of these terms in daily practice. We note, however, that the examples given in the Questionnaire to operationalize these terms, would point respondents to think of \'behavior problems\' as suggestive of a possible disturbance in the range of disruptive, oppositional and antisocial disorders; and of \'social-emotional difficulties\' as possibly suggestive of anxiety or depression. A draft version of the questionnaire was pilot tested with eight GP/FPs who commented on the measure\'s ease of use, clarity, completeness and relevance. These comments were incorporated into the final version.
*Preferred approach to case involvement*was ascertained by enquiring about the GP/FP\'s usual pattern of practice regarding children with each clinical condition. Options ranged from \"evaluate and manage the problem yourself\" to \"refer out for evaluation and management\", with intermediate arrangements for combining care with specialists.
*Self-reported levels of comfort, skill and effectiveness*GP/FPs were asked to rate their confidence and skill across four domains of clinical activity (comfort with diagnosis/evaluation; comfort with management; skill in diagnosis/evaluation; and effectiveness in management) for each clinical condition. Ratings were made on 7-point scales for each item, anchored by Very Uncomfortable/Unskilled/Ineffective (score of 1) and Very Comfortable/Skilled/Effective (score of 7).
*Beliefs about mental health problems in children*were measured by GP/FPs\' degree of agreement with four statements relating to the nature, etiology, evaluation and management of such problems (two of them adapted from the Management of Childhood Depression in Primary Care questionnaire \[[@B11]\]), also using 7-point scales, anchored by Strongly Disagree (score of 1) and Strongly Agree (score of 7).
*Physician personal, demographic and practice characteristics*included gender; practice type (Solo/Group/Walk in/other); practice location (Urban/Rural/Other); hours per week spent in patient care; and participation in continuing medical education activities (CME) covering children with behavioral and emotional problems over the past 5 years; as well as the number of children per month presenting with each condition type, or in whom the GP/FP would consider one of these diagnoses.
Procedures
----------
Survey packages, including questionnaire, covering letter, prepaid return envelope, and a personalized \$15 cheque (that recipients could keep or return as a donation to the hospital\'s charitable foundation), were mailed out in waves of approximately 200 each between November 2001 and March 2002. The initial mailout was followed 2 weeks later by a thank you/reminder letter, and 6 weeks later by a duplicate survey package to non-respondents. Two weeks later, we phoned the offices of non-responding GP/FPs to encourage participation, elucidate reasons for non-participation, and to try to confirm the GP/FP\'s eligibility status.
Data reduction and analysis
---------------------------
In an initial data reduction step, the four comfort/skill and effectiveness items for each condition were subjected to principal components analyses (PCA). These analyses resulted in one component solutions for each condition type that accounted for a large percentage of the total variation in item scores, ranging from 76% to 83%. Each component reflected \"dealing with\" a particular condition type. For example, the four items concerning comfort and skill in assessing/diagnosing and managing, and effectiveness and skill in managing behavior problems, loaded uniformly highly onto a derived component for \"Comfort/skill in dealing with behavior problems\". Component scores were therefore created by averaging the four items for each condition type, yielding four scores: Comfort/Skill with behavior problems, social-emotional difficulties, ADHD, and mood disorders, respectively.
To compare Comfort/Skill scores across condition types, we used repeated measures ANOVAs. In all analyses, we adopted a more stringent alpha level of 0.01 to indicate statistical significance (to protect against Type I error rate inflation) and an effect size of 0.5 (Cohen\'s d) to indicate clinical significance. To discern trends in GP/FP comfort and skill across condition types more clearly, we also characterized respondents\' reported scores of 1, 2 or 3 on the relevant items as having \"Low\" self-reported Comfort/Skill, and those reporting scores of 5, 6 or 7 as having \"High\" self-reported Comfort/Skill.
Multiple linear regression analyses were used to explore factors associated with Comfort/Skill scores (dependent variable) for each of the condition types. Independent variables included GP/FP background and demographic characteristics; practice characteristics; preferred approach to case involvement; and beliefs regarding care of children with behavioral/ psychosocial concerns.
To explore variables associated with GP/FPs\' preferred approach to case involvement, we performed a series of logistic regression analyses using the two most clearly differentiated forms of involvement for each condition type (primarily manage by yourself vs. primarily refer out). The same independent variables were used as listed above, with the addition of respondents\' Comfort/Skill scores for each condition type.
Results
=======
Sample
------
Of 801 questionnaires mailed, 567 were returned, and 427 completed. A total of 172 recipients were identified as ineligible to participate, including 150 who did not complete questionnaires and 22 who completed the survey form in error, as they reported seeing no children with any of the study conditions in a typical month. The remaining 405 questionnaires formed the study database, a response rate of 64.4% of eligible recipients. Table [1](#T1){ref-type="table"} presents characteristics of the respondent group. In comparison with the non-responders, more respondents were female (37% vs. 29%, respectively; χ^2^(1) = 4.2; P \< 0.05); certified as specialists in Family Medicine (40% vs. 28%, respectively; χ^2^(1) = 9.9; P \< 0.005); and in group, as compared to solo, practice (63% vs. 51%; χ^2^(1) = 9.3; P \< 0.005).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Characteristics of the respondent sample
:::
**N** **(%)\***
------------------------------------------------------------------------------- ------- -----------
**Gender**
Female 149 (37)
**Type of practice**
Solo 93 (23)
Group 237 (59)
Walk-in 31 (8)
Other 44 (11)
**Location of practice**
Urban 263 (65)
Rural 134 (33)
Other 7 (2)
**Time spent in patient care**
40 hrs/wk or less 209 (52)
\> 40 hrs/wk 191 (48)
**Recency of graduation**
Prior to 1978 139 (34)
1978 to 1988 142 (35)
Since 1989 124 (31)
**Place of graduation**
Canadian University 327 (81)
**Specialty certificate from the College of Family Practitioners of Canada** 166 (41)
**Participated in CME for psychosocial problems in children in past 5 years** 171 (42)
**Children seen per month: behaviour problems**
None 28 (6.9)
1 to 4 324 (80.4)
5 to 9 46 (11.4)
\>9 5 (1.2)
**Children seen per month: social-emotional difficulties**
None 25 (6.2)
1 to 4 302 (75.0)
5 to 9 56 (13.9)
\>9 12 (3.0)
**Children seen per month: ADHD/possible ADHD**
None 81 (20.1)
1 to 4 303 (75.2)
5 to 9 18 (4.5)
\>9 1 (0.2)
**Children seen per month: mood disorders/possible mood disorders**
None 28 (6.9)
1 to 4 315 (78.2)
5 to 9 47 (11.7)
\>9 13 (3.2)
Note: number of respondents varies slightly in tables of results because respondents did not always answer all questions fully
\* Percentages rounded to nearest whole number
:::
Volume of patients seen with each condition
-------------------------------------------
Between 75 and 80% of respondents reported seeing 1 to 4 children newly presenting with behavior problems or social-emotional difficulties, or in whom they would consider or make the diagnosis of ADHD or mood disorders, each month. Between 11% and 14% reported seeing 5 to 9 such cases for all conditions except ADHD, for which the proportion was 4.5%. Conversely, 20% of GP/FPs reported seeing no children in whom they would consider or diagnose ADHD, as compared with other condition types, for which the proportion varied from 6.2% to 6.9%.
Preferences for case involvement
--------------------------------
Respondents frequently reported combining personal involvement with referral. GP/FPs most commonly commence evaluation and management themselves, and then refer the patient for consultation (Table [2](#T2){ref-type="table"}). We contrasted the frequency of GP/FPs who primarily evaluate and manage these cases themselves without referral, with the frequency of physicians who primarily refer out for evaluation and management or who refer out for evaluation and then take over management, across condition types. For children with behavior problems or ADHD, more GP/FPs referred cases than handled them by themselves (χ^2^(1) = 9.0; P \< 0.005; and χ^2^(1) = 103.9; P \< 0.001, respectively), whereas no differences in case involvement were found for children with social-emotional difficulties or mood disorders.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
GP/FPs\' preferred approaches to case involvement\*
:::
**Clinical Condition Type**
----------------------------------------------------------- ------------------------------- ----------------------------------------- ---------------- --------------------------
**Behavioral Problems N (%)** **Social-Emotional Difficulties N (%)** **ADHD N (%)** **Mood Disorders N (%)**
Evaluate and manage the problem yourself 44 (11.1) 50 (12.7) 21 (5.3) 74 (18.8)
Evaluate and manage yourself, then refer for consultation 150 (38.0) 156 (39.5) 103 (26.0) 171 (43.4)
Evaluate yourself and then refer out for management 119 (30.1) 115 (29.1) 106 (26.8) 72 (18.3)
Refer out for evaluation and management 52 (13.2) 52 (13.2) 75 (18.9) 29 (7.4)
Refer out for evaluation and then take over management 25 (6.3) 17 (4.3) 82 (20.7) 36 (9.1)
Other combination of options 5 (1.3) 5 (1.3) 9 (2.3) 12 (3.0)
\*Number (percentage) of GP/FPs that preferred each approach to involvement with each clinical condition type.
:::
Self-reported comfort/skill
---------------------------
Comfort/Skill scores by clinical condition type clustered closely around the midpoint of the 1 to 7 scale: behavior problems (mean ± s.d) 3.6 ± 1.1; social-emotional difficulties 3.8 ± 1.1; ADHD 3.9 ± 1.3; and mood disorders 4.4 ± 1.3. There was a statistically and clinically significant difference between Comfort/Skill with behavior problems and mood disorders (F \[3,1155\] = 84.0, P \< .0001; effect size = 0.67), with respondents more positive about their ability to deal with mood disorders than behavior problems. This finding is supported and extended by the data on the frequency of physicians reporting \"Low\" vs. \"High\" Comfort/Skill (Table [3](#T3){ref-type="table"}). High raters outnumbered Low raters by about 2:1 for all clinical activities related to mood disorders, whereas for most aspects of dealing with behavior problems, this pattern was reversed.
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Number (%) of GP/FPs reporting Low and High ratings of comfort, skill and effectiveness with each clinical conditions \*
:::
**Behavior Problems N (%)** **Social-emotional difficulties N (%)** **ADHD N (%)** **Mood Disorders N (%)**
------------------------------------ ------ ----------------------------- ----------------------------------------- ---------------- --------------------------
Comfort with diagnosis/ evaluation Low 147 (50.3) 111 (39.1) 136 (47.6) 86 (27)
High 145 (49.7) 173 (60.9) 150 (52.4) 232 (73)
Skill in diagnosis/ evaluation Low 183 (61.6) 152 (53) 144 (51.8) 96 (31.1)
High 114 (38.4) 135 (47) 134 (48.2) 213 (68.9)
Comfort with management Low 213 (70.5) 172 (59.1) 156 (51.1) 115 (35.5)
High 89 (29.5) 119 (40.9) 149 (48.9) 209 (64.5)
Effectiveness in management Low 223 (76.4) 192 (67.4) 148 (50.7) 104 (34.7)
High 69 (23.6) 93 (32.6) 144 (49.3) 196 (65.3)
\*Percentages are proportion of respondents with High or Low ratings, not of the entire sample.
:::
Beliefs about psychosocial problems in children
-----------------------------------------------
On the 1 to 7 scales, respondents indicated an overall tendency to disagree with the following belief statements: \"These problems/conditions are usually mild and transient, so specific intervention or treatment is not usually required\"; and \"The role of primary care physicians should be very limited with these kinds of problems/conditions\", with mean (s.d) ratings of 2.2 (± 1.2) and 2.4 (± 1.2), respectively. Conversely, there was a tendency to agree with the following statements: \"These problems/conditions are usually related to stresses in the family which are hard to manage\" and \"Diagnosis/evaluation of these problems is often subjective and difficult\", with ratings of 4.9 (± 1.3) and 4.4 (± 1.5), respectively (Table [4](#T4){ref-type="table"})
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Number (%) of GP/FPs reporting different levels of agreement with belief statements about primary care management of children and youth with psychosocial problems
:::
**Belief\*** **Level of Agreement**
-------------- ------------------------ ------------ ------------
**1** 270 (68.2) 121 (30.5) 5 (1.3)
**2** 19 (4.8) 245 (61.7) 133 (33.5)
**3** 252 (63.5) 73 (34.5) 8 (2.0)
**4** 61 (15.4) 232 (58.8) 102 (25.8)
\* Belief statements:
1\. These problems/conditions are usually mild and transient, so specific intervention or treatment is not usually required.
2\. These problems/conditions are usually related to stresses in the family which are hard to manage.
3\. The role of primary care physician should be very limited with these kinds of problems/ conditions.
4\. Diagnosis/evaluation of these problems is often subjective and difficult.
:::
Factors associated with self-reported comfort/skill
---------------------------------------------------
In the multivariate models, two variables were related to higher Comfort/Skill across all condition types: having participated in CME to do with children\'s psychosocial problems in the past 5 years; and disagreement with the belief that evaluation of these conditions in children is subjective and difficult. Three other variables were related to Comfort/Skill across most of the condition types: seeing more children per month with that condition; disagreement with the belief that the role of the GP/FP in addressing these problems should be limited; and disagreement with the belief that these problems are related to stresses in the family that are hard to manage (see Table [4](#T4){ref-type="table"}). No individual factor independently explained more than 7% of the variance in Comfort/Skill scores, but together each set of variables explained up to 20% of the variance in Comfort/Skill scores for each condition.
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
Factors associated with higher self-rated comfort/skill
:::
** β ** **P** **Unique Variance Explained (%)** **Overall Variance Explained (%)**
--------------------------------------------------------------------------------------- --------------------------- ------- ----------------------------------- ------------------------------------
**Behavioral Problems** **19**
Participated in CME\~ 0.16 0.001 2.5
See more than 5 children per month with behavior problems 0.14 0.003 1.9
Belief: problems are mild & transient, specific intervention not required 0.12 0.01 1.4
Belief: these problems are related to stresses in the family that are hard to manage -0.13 0.007 1.5
Belief: role of GP should be very limited -0.19 0.000 2.9
Belief: evaluation of these problems is often subjective and difficult -0.22 0.000 4.2
**Social-emotional Difficulties** **19**
Participated in CME\~ 0.20 0.000 4.1
Belief: role of GP should be very limited -0.24 0.000 5.2
Belief: evaluation of these problems is often subjective and difficult -0.26 0.000 6.7
**ADHD**° **20**
Male gender 0.20 0.000 3.9
Participated in CME\~ 0.21 0.000 4.4
See more than 5 children per month with ADHD° 0.21 0.000 4.3
Certified as specialist in Family Medicine 0.14 0.002 2.0
Belief: these problems are related to stresses in the family that are hard to manage -0.14 0.003 1.8
Belief: evaluation of these problems is often subjective and difficult -0.13 0.005 1.6
**Mood Disorders** **19**
Participated in CME\~ 0.22 0.000 4.5
See more than 5 children per month with Mood Disorders 0.17 0.000 3.4
Spend more than 40 hours/wk in patient care 0.10 0.04 0.9
Belief: these problems are related to stresses in the family that are hard to manage -0.09 0.05 0.8
Belief: role of GP should be very limited -0.18 0.000 2.9
Belief: evaluation of these problems is often subjective and difficult -0.16 0.001 2.3
\~CME: continuing medical education activities
°ADHD: attention-deficit/hyperactivity disorder
:::
Factors associated with preferred approach to case involvement
--------------------------------------------------------------
In our logistic regression analyses, only Comfort/Skill was related to preference for case involvement. Higher ratings of Comfort/Skill were strongly associated with lower likelihood of GP/FPs referring these cases out, with odds ratios (95% confidence intervals) as follows: for behavior problems, 0.16 (0.07 -- 0.35; n = 114); social-emotional difficulties, 0.21 (0.10 -- 0.44; n = 108); ADHD 0.26 (0.14 -- 0.50; n = 170); and mood disorders, 0.21 (0.11 -- 0.40; n = 128). Higher Comfort/Skill was therefore predictive of a tendency for GP/FPs to deal with these cases themselves. The proportion of variance (Nagelkerke R^2^) in GP/FPs\' likelihood of referring out (rather than handling by him/herself) explained by Comfort/Skill was 61% for behavior problems, 53% for social-emotional difficulties, 34% for ADHD, and 60% for mood disorders.
Discussion
==========
GP/FPs in the Canadian province of BC, who are the predominant providers of primary health care to children, report a fairly consistent exposure to common childhood behavioral and emotional concerns in their practices. They report an intermediate level of comfort and skill in dealing with these patients overall, with the highest level for mood disorders and lowest for behavior problems. Caring for these children frequently involves collaboration with consultants. Self-reported comfort/skill was an important predictor of a GP/FP\'s tendency to take primary responsibility for a case, and self-reported comfort/skill was in turn related to previous educational exposure to this field, and beliefs about mental health problems in children. These findings have implications for physician education and primary care practice and organization of services.
Our results are consistent with other studies that have found an important role for physician self-efficacy in predicting physician practices \[[@B13],[@B14]\], as would be predicted from social cognitive theory \[[@B15]\]. A previous study found pediatrician confidence in diagnosis and management of children and youth with depression to be associated with higher perceived responsibility for treating these cases, which is in turn predictive of the physician\'s prescribing medication and scheduling further appointments \[[@B10]\]. Although our results explain only a portion of the variation in physicians\' comfort and skill, the consistent predictions from CME related to children\'s behavioral and emotional problems, and physicians\' beliefs about the care of these problems, are notable. Participation in educational programs has been found to increase physicians\' sense of professional efficacy in other studies \[[@B16]\], underlining the potentially important role for appropriate CME. Our finding that participation in CME and physician beliefs are both important factors in self-efficacy, raises the intriguing possibility that some of the benefits of participation in CME may be mediated through effects on attitudes and beliefs over-and-above simple knowledge acquisition. It has, in fact, been suggested that focusing on attitudes and beliefs may be a legitimate and important way to enhance skills and confidence among GP/FPs in relation to mental health problems \[[@B17]\].
We would point out that our results should not be interpreted as showing a direct benefit of physician CME on physician practice with children with behavioural and emotional problems, or on patient outcomes. The likelihood of physician CME resulting in changes in physician behavior and patient outcomes depends in large part on the type of CME activity undertaken, with certain types of CME being relatively ineffective and others being quite effective \[[@B18]-[@B20]\]. Our survey did not enquire about the types of CME activities in which physicians had participated. In relation to educational effects more broadly, previous studies from the USA have reported mixed results for effects of specialized training in psychosocial issues on the practice of primary care physicians with children with these problems. In one study, no effect of such training on treatment decisions was found \[[@B21]\], while in another study, more intensive levels of advanced training did result in better identification and management practices \[[@B22]\].
Our study raises questions about why GP/FPs feel more comfortable and confident in dealing with children with mood disorders, and less so those with behavior problems, and why they tend to refer children with behavior problems and ADHD more frequently than dealing with these cases themselves. Children\'s behavior problems may pose challenges because clear diagnostic criteria and management algorithms are lacking for this diffuse and heterogeneous group of clinical scenarios. An increased emphasis during physician training on understanding child development, and on learning practical strategies (such as contingent reinforcement) to deal with aberrant behavior, may enable GP/FPs to feel more confident with these problems. ADHD, on the other hand, continues to be a source of concern for many GP/FPs, in spite of accepted diagnostic criteria, evidence-based practice guidelines, and effective treatments.\[[@B23]\] Previous research from Australia identified diagnostic issues and complexities, time intensiveness, and insufficient education and training as contributing to GP/FP\'s reluctance to take primary responsibility for children with ADHD\[[@B8]\], while a US study of primary care providers found that higher severity of child psychosocial problems, and poorer family functioning, predicted referral to specialized mental health services \[[@B21]\].
The higher reported level of involvement and comfort/skill in dealing with mood disorders may reflect primarily work with adolescents, in which GP/FPs utilize strategies that are successful with adult patients, and drawing on the fairly extensive exposure to psychosocial medicine in family practice training\[[@B24]\]. We cannot address this possibility directly from data obtained from respondents in our survey, but we note that over 2/3 of pediatric cases of depression seen in a recent study of US pediatricians, did indeed involve youth 13 -- 18 years of age \[[@B10]\]. We note, however, that the diagnostic and treatment approaches used for adults may not be appropriate for younger children, especially in light of concerns about the effectiveness and safety of specific serotonin reuptake inhibitor (SSRI) medications that have arisen since the time of our survey \[[@B25]\]. An apparent trend for more respondents to report high rather than low levels of comfort/skill for mood disorders compared with social-emotional difficulties, may be attributable in part to the terminology used in the survey. Although behavioral and social-emotional problems that would be considered subthreshold for specific diagnoses occur relatively frequently in primary care settings \[[@B26],[@B27]\], it is possible that these non-specific and somewhat ambiguous terms may evoke uncertainty in the mind of clinicians trained in diagnostic and management approaches aimed at clearly delineated entities. Physician\'s responses in this survey might therefore reflect either uncertainty about how to cope with such problems in children, or perhaps about what was being referred to in the survey.
Strengths of our study include its careful sampling base of GP/FPs from across the province of BC, and its breadth and depth of scope. This is one of very few studies to examine how GP/FPs as a group, deal with children presenting with a range of behavioral and emotional problems, while at the same time exploring factors that underlie some of the physicians\' practice patterns, preferences and perceptions. Certain limitations of this work also need to be acknowledged. Cross-sectional studies are limited in their ability to arrive at conclusions about causality. Hence we cannot be sure, for example, that attending CME increases comfort and skill levels. It is possible that physicians who attend CME on certain topics may be particularly interested and motivated in those areas. Nevertheless, our study highlights important associations, a number of which fit with causal expectations suggested by other studies and theory. We also cannot be sure that the views of our respondents reflect all GP/FPs\' perceptions and practices, given our 64% response rate, although this was considerably higher than the 54% reported for physician postal surveys overall \[[@B28]\]. Finally, the extent to which our findings from a Canadian health care context would generalize to settings where patients\' medical needs are not covered under a system of universal health insurance coverage, or where GP/FPs are less extensively involved in primary care, is unknown. However, our findings are consistent with those from other settings.\[[@B6]-[@B8],[@B11]\] It is interesting to note that while primary care pediatricians in the USA report relatively low levels of confidence in their diagnostic and management skills for depression in children and youth most still become directly involved in some aspect of these patients\' care \[[@B10]\], in a similar fashion to GP/FPs in BC. These observations underline the need for better support for primary care physicians in their role with these children, through increasing self-confidence and self-efficacy.
Conclusion
==========
It is encouraging to find that, in general, GP/FPs in BC report at least an intermediate level of comfort and skill in assessment and management. For the majority of patients, care involves the GP/FP and referral for specialist consultation. Given that specialists with skill and experience in dealing with psychosocial and mental health concerns are often in short supply, however, it would be desirable to enable GP/FPs to deal with more of these cases independently but confidently. To do so may require new educational strategies aimed at enhancing self-efficacy and capable of affecting attitudes and beliefs regarding mental health problems in children. In addition to these strategies, however, foundational changes in the way medical and support services are organized for care of children with psychosocial and mental health problems will also be required, given the barriers and challenges posed by financial and cultural factors, lack of access to specialized mental health services, and time and resource constraints within primary care \[[@B8],[@B10],[@B29]\].
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
AM, CJ, AK and SF jointly conceived the study. AM, CJ, AK, SF and MP contributed to the study design. AM was responsible for overseeing day to day coordination of the study, anddrafted the manuscript. MP performed the statistical analysis. All authors contributed to interpreting the results, and read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2296/6/12/prepub>
Acknowledgements
================
Funding for this study was provided by the British Columbia Children\'s Hospital Foundation Telethon Fund. Dr Miller is supported by an Investigator Award from the Sunny Hill Foundation for Children.
|
PubMed Central
|
2024-06-05T03:55:55.209320
|
2005-3-11
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079811/",
"journal": "BMC Fam Pract. 2005 Mar 11; 6:12",
"authors": [
{
"first": "Anton R",
"last": "Miller"
},
{
"first": "Charlotte",
"last": "Johnston"
},
{
"first": "Anne F",
"last": "Klassen"
},
{
"first": "Stuart",
"last": "Fine"
},
{
"first": "Michael",
"last": "Papsdorf"
}
]
}
|
PMC1079812
|
Background
==========
Type 2 diabetes is a common condition with high morbidity and mortality. Its diagnosis appears to be increasing \[[@B1]-[@B3]\], probably reflecting an underlying rise in prevalence. In many countries, guidelines for its management have been issued \[[@B4]-[@B6]\], and in the United Kingdom (UK), these are reflected in the National Service Framework for Diabetes, which sets national standards and defines service models for the condition\[[@B7]\]. UK Family Practitioners\' remuneration will increasingly depend on hitting targets which encompass both process of care and health related outcomes\[[@B8]\]. In this context, process includes monitoring of the disease and of modifiable risk factors that might lead to complications. In particular, it is important to manage cardiovascular risk in diabetics because diabetes accelerates vascular occlusion and much of the excess mortality is due to cardiovascular mortality\[[@B9]\].
Practitioners will be judged largely on the achievement of target levels for the process measures monitored. Achieving such targets requires pharmacotherapy in most cases. In recent years a number of new drugs for the management of diabetes have appeared and research evidence suggests that these should replace some older drugs and supplement others, whilst some drugs, such as metformin, should be used more frequently \[[@B10]-[@B12]\].
In this study, we use routinely collected computer data to examine trends in the prevalence of diagnosed Type 2 diabetes, the drug treatment of the condition and recording of measurements utilised in its management, including body mass index (BMI), blood pressure and glycosylated haemoglobin (HbA1c). Finally we examine trends in the achievement of management target levels for these measurements as suggested by national guidance.
Methods
=======
The Doctors\' Independent Network (DIN) is an anonymised, computerised UK primary care database comprising 142 Family Practices that we consider to be good quality data providers\[[@B13]\]. Morbidity and drug data are coded using Read codes (4 byte). This study uses data from 74 practices that had continuous recording from 1994 to 2001, eliminating possible spurious trends caused by having different practices in the sample over time. We included patients in a given year if they remained registered on 31^st^December of that year and for 6 months previously.
Diabetics were identified by electronically searching the database for all diabetic Read codes. Diabetics on insulin were classified as type I or type II based on an algorithm (available from authors) which used date of diagnosis and date of first being seen in the practice, in combination with the timing of treatment with oral hypoglycaemics (British National Formulary (BNF) section 6.1.2) and/or insulin (BNF section 6.1.1). The decision was largely straightforward for diabetics newly diagnosed while they were registered with practices using their electronic database. Diabetics were classified as type 1 if a Read code \"Diabetes + ketoacidosis -no coma\" was present or if insulin was given within 90 days of first diagnosis. The problems arose for those diabetics diagnosed in the past who were using insulin at the point of registration with the practice or when the practice started using the electronic database. While dates of first diagnosis were usually available, as were date of registration and first recording of events within practice, we did not have information on when insulin treatment was started prior to electronic recording. In this instance it was necessary to base the decision on the age of the patient at first diagnosis (≥35 years implies type 2) as well as the time lapse between when the patient was first seen in the practice and when an insulin prescription was issued (\>180 days implies type 2). Based on a random sample of 150 records of diabetics receiving insulin, the algorithm had 94% sensitivity and 93% specificity for classifying type 2 diabetics compared to blind assessment by a clinician (SDeW) of the same electronic records. Type 2 diabetics were further classified by treatment received in a given year into: diet, oral only, or insulin (with or without oral).
The records of Type 2 diabetics were searched for Read codes for Body Mass Index (BMI), blood pressure, total cholesterol, and glycosylated haemoglobin (HbA1c). Read codes for these data may be indicative (e.g. \"cholesterol raised\") or have a numerical value assigned. In earlier years indicative codes tended to be used more, and for HbA1c, these were the only recording option before 1997. If an indicative code (or a biologically implausible numerical code) was used then we counted the measurement as having occurred, but set its value to be missing.
Where multiple measurements took place in a given year, we used the last recorded measurement in that year. Numerical values were selected in preference to indicative codes even if they were not the last recorded measurement in a year. BMI values were supplemented by weight records, if an existing height measurement was present in the record, allowing BMI to be calculated.
In order to examine trends in the achievement of management targets, data were compared with targets set in guidance published by the National Institute for Clinical Excellence (NICE). The NICE targets are: BMI \<25 kg/m^2^, blood pressure \<140/80 mmHg or \<160/100 mmHg depending on coronary event risk, HbA1c \<6.5% or \<7.5% depending on coronary event risk, and total cholesterol \<5 mmol/l\[[@B4],[@B5]\]. No target figure is set for smoking reduction.
To test whether any trends in the achievement of management targets were influenced by the improvements in the recording of data, we restricted the analyses to 16 practices with a consistently high level of recording and low levels of missing values. To establish whether any trends in the percentage of diabetics meeting targets were due to differences in the way that newly diagnosed diabetics were being managed, analyses were also repeated after restriction to newly diagnosed diabetics.
Statistical methods
-------------------
To adjust for the changing age structure of type 2 diabetics between 1994--2001 we used the 2001 population of DIN to age standardise prevalence rates using the direct method. Descriptive analyses of practice variation in recording of risk factor data (Figure [3](#F3){ref-type="fig"}) are summarised using box and whisker plots using the SAS procedure BOXPLOT (SAS Institute Inc., North Carolina, USA). Tests for trend across years are based on fitting the logit of the percentage of diabetics in a practice with the relevant risk factor recorded in a given year using SAS procedure GENMOD.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Distribution of the practice percentage of diabetics with a risk factor measured by year**. -- Boxes indicate the median, lower and upper quartiles Whiskers extend to the practice immediately proceeding 1.5 times the interquartile range from the median. Practices lying outside this range are individually plotted. -- % of all measurements that were numeric and valid were for each risk factor: BMI (95%), Blood Pressure (97%), HbA1c (61%) and Cholesterol (91%).
:::
:::
Tests for trends of targets met across years was based on individual data using a logit link adjusting for age, sex and practice, which was included as a categorical variable with 74 levels. Year was fitted as a linear variable to test for trend. The model was fitted using SAS procedure GENMOD with a first order autoregressive structure to allow for higher correlation of measurements on the same individuals when measured in adjacent years.
In order to assess whether changes in HbA1c over years were influenced by the changes in BMI, we regressed the recorded HbA1c levels on age, sex, practice and year before further including BMI. Various models were fitted using the SAS procedure MIXED, allowing us to take account of individuals contributing data to variable numbers of years.
Results
=======
Diabetes prevalence trends
--------------------------
The population totalled approximately half a million patients annually (Table [1](#T1){ref-type="table"}). Between 1994 and 2001, the prevalence of diagnosed Type 2 diabetes increased steadily, rising from 18 to 27 per 1000 in men and from 16 to 23 per 1000 in women. The age standardised rates (Table [1](#T1){ref-type="table"}) were almost identical. In both sexes, the most notable increases were in the 65--74 age group (men increased from 68 to 101 per 1000, women from 47 to 73 per 1000).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Prevalence rates (per 1000) of type 2 diabetics from the DIN database (74 practices)
:::
**1994** **1995** **1996** **1997** **1998** **1999** **2000** **2001**
---------------------------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- -----------
*Total males* *237,872* *239,427* *243,618* *251,005* *255,833* *259,670* *260,383* *262,290*
\- Diet controlled only 1,689 1,788 1,830 1,940 1,971 2,086 2,218 2,285
\- On oral drug only 2,111 2,186 2,374 2,573 2,833 3,128 3,417 3,779
\- On insulin 542 569 637 704 785 894 985 1,052
**Total type 2 diabetics** **4,342** **4,543** **4,841** **5,217** **5,589** **6,108** **6,620** **7,116**
0--34 years 1 1 1 1 1 1 1 1
35--44 7 7 7 8 8 8 9 9
45--54 18 18 19 21 22 23 26 27
55--64 45 47 49 50 52 54 56 57
65--74 69 71 75 79 85 91 96 101
75--84 72 78 77 82 86 94 99 109
85+ 79 72 74 72 71 77 81 89
**Crude overall rate** **18** **19** **20** **21** **22** **23** **25** **27**
Age Std Rate\* 19 20 21 22 23 24 26 27
95% CI 18--21 19--21 20--22 21--23 22--24 23--25 25--27 26--28
*Total females* *241,995* *242,996* *246,885* *253,383* *258,695* *262,241* *263,475* *264,763*
\- Diet controlled only 1,404 1,496 1,565 1,658 1,722 1,791 1,885 2,065
\- On oral drug only 1,824 1,915 2,070 2,244 2,427 2,586 2,844 3,048
\- On insulin 528 532 579 632 712 801 888 941
**Total type 2 diabetics** **3,756** **3,943** **4,214** **4,534** **4,861** **5,178** **5,617** **6,057**
0--34 years 1 1 1 1 1 2 2 2
35--44 6 6 6 7 7 8 9 9
45--54 12 13 14 15 15 16 17 19
55--64 33 34 34 34 36 38 38 40
65--74 47 50 53 59 63 66 70 73
75--84 53 56 59 59 61 67 72 76
85+ 48 48 53 52 56 58 61 68
**Crude overall rate** **16** **16** **17** **18** **19** **20** **21** **23**
Age Std Rate\* 16 17 17 18 19 20 22 23
95% CI 15--17 16--18 16--19 17--19 18--20 19--21 21--23 22--24
\* - Age standardised rate is made to the 2001 DIN Population
:::
Trends in therapy
-----------------
During the 1990\'s there were steady increases in the prevalence of diet only, oral treatment only and insulin treated type 2 diabetes (Figure [1](#F1){ref-type="fig"}). However, the proportion of type 2 diabetics treated by diet alone fell between 1994 and 2001 (from 38.2% to 33.0%), while those treated with any insulin (13.2% to 15.1%) and oral agents only (48.6% to 51.8%) both increased.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Prevalence of type 2 diabetic treatment groups in the DIN database over time**. Blue -- On insulin, Red -- On oral drug only, Green -- Diet controlled only
:::
:::
Trends in oral therapy are shown in Figure [2](#F2){ref-type="fig"}. Use of ultra-long acting sulphonylureas had almost completely ceased by 2001 when it was received by 0.1% of diabetics. The use of long acting sulphonylureas and guar gum (0.2% in 1994 to 0.04% in 2001 -- not shown in Figure [2](#F2){ref-type="fig"}) has also reduced. In contrast, the use of short acting sulphonylureas increased (from 23.3% to 35.1%), as did the use of metformin (from 22.6% to 38.9%). The use of the intestinal alpha glucosidase inhibitor, acarbose, increased up to 1999, rising from 1.7% in 1994 to 3.8% in 1999, then declined back to 2.4% in 2001. Thiazolidinediones and meglitinides started being used in the final years of the study.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Changes in the non-insulin treatment of type 2 diabetics over time**. Light Blue -- Metformin, Red -- Short acting sulphonylurea, Green -- Long acting acting sulphonylurea, Orange -- Ultra long acting sulphonylurea, Purple -- Acarbose, Yellow -- Meglitinides, Dark Blue -- Thiazolidinediones/Glitazones **Footnote:**Treatment groups are not mutually exclusive
:::
:::
Recording of indicators of process of care
------------------------------------------
Trends in recording of measurements linked to the process of diabetic care where targets are set in NICE guidance are shown in Figure [3](#F3){ref-type="fig"}. Recording of all measures has increased (P \< 0.001), especially the recording of HbA1c (practice median 34 % in 1994 to 74% in 2001) and cholesterol (practice median 17% in 1994, 61% in 2001). The practice median for recording of BMI was 43% of patients in 1994 rising to 55% in 2001; for blood pressure it rose from 65% in 1994 to 82% in 2001. Less than 10% of all measurements made were classified as \"missing\" for all except HbA1c. For HbA1c 39% of measurements recorded as being made were \"missing\" due to the sole existence of indicative codes before 1997. There was variation in recording between practices in any given year which, while it diminished over time, was still marked in 2001, with one practice still not recording any of the measures.
Outcomes: achievement of national quality targets
-------------------------------------------------
Trends in the percentage of diabetics whose management achieved NICE targets are shown in Figure [4](#F4){ref-type="fig"}. There has been a marked increase (P \< 0.001) in the percentage of diabetics with a total cholesterol \<5 mmol/l. (46.2% in 2001), and a steady increase in patients with blood pressure readings below the target levels (P \< 0.001), although only 22.5% had a blood pressure less than 140/80 in 2001.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**(NICE/NSF) Risk factor targets achieved in all type 2 diabetics by year**. (Results are age & sex standardised to the 2001 population) Denominators are the number of subjects with a valid numerical value recorded. HbA1C results start in 1997 as numerical codes could not be recorded prior to 1997.
:::
:::
The percentage of patients meeting either of the target HbA1c figures has however tended to fall, except for 2001 when there was a small increase (trend tests P \< 0.001 for lower target, P = 0.14 for higher target) : 22.5% of patients in 2001 had an HbA1c of less than 6.5% compared to 28.9% in 1997; 52.3% of patients in 2001 had an HbA1c less than 7.5% compared to 53.7% in 1997. A steady fall in the percentage of patients with a BMI less than 25 kg/m^2^has also occurred (P \< 0.001) with 19.4% of patients achieving this criterion in 2001, compared with 27.0% in 1994.
In order to assess whether the lack of improvement in glucose control over time might be due to the steady increase in the obesity of the diabetic population, we estimated the mean HbA1c levels adjusted for age, sex and practice and then further adjusted for BMI. The mean levels unadjusted for BMI reflected the same pattern as in figure [4](#F4){ref-type="fig"}; that is the mean HbA1c rose from 7.73% in 1997 to 7.82% in 2000 before falling back to 7.69% in 2001. While BMI was highly significantly related to HbA1c, adjustment for it had little effect on the estimated HbA1c means for the different years.
Trends in practices with high levels of recording of target related data
------------------------------------------------------------------------
When the above analyses were restricted to 16 practices with consistently high levels of data recording (corresponding to those above the upper quartile in Figure [3](#F3){ref-type="fig"} for each factor), the trends in meeting targets as described above remained almost unaltered (data not shown).
Treatment of newly diagnosed diabetics
--------------------------------------
The trends in treatment and other management in newly diagnosed diabetics were almost identical to those seen for all diabetics (data not shown).
Discussion
==========
Principal findings
------------------
This study confirms that the prevalence of all diagnosed type 2 diabetes continues to rise. The data are consistent with data published from the General Practice Research Database (GPRD) to 1998\[[@B3]\]. However, that study failed to identify diet treated diabetics or to include those type 2 diabetics treated with insulin. Our figures for overall prevalence and type of therapy amongst type 2 diabetics compares well with the only other UK source we have identified. In that much smaller study, the prevalence of type 2 diabetes was 20.3 per 1000 in males and 16.7 per 1000 in females in 1997, of whom 34% were treated by diet only, 53% were oral only and 14% were on insulin\[[@B14]\]. It is reassuring that two studies based on very different methodologies give such similar results.
The prevalence of diabetes has risen steadily in developed and developing countries throughout the second half of the 20^th^century\[[@B1]\]. Rising levels of obesity in the general population are believed to be one of the principal drivers\[[@B1]\], and in a recent Danish study it was concluded that the rise in diabetes between 1974/75 and 1996/97 was entirely attributable to the concurrent rise in body mass index\[[@B15]\]. It seems likely that this underlies the steady increase we have observed, given that the percentage of adult males in England with a BMI over 30 rose from 13.8% in 1994 to 21.0% in 2001 while for women the trend was from 17.3% to 23.5% \[[@B16]\]. While there is a theoretical possibility that changes in the definition of diabtetes introduced from 1998 on may have had some effect, it seems likely to have been limited given that the steady increase in the prevalence of type 2 diabetes predated the change in diagnostic criteria; in 1998--9 the definition of diabetes shifted from those with a 2 hour post load plasma glucose \> 11.1 mmol/l or a fasting level ≥ 7.8 mmo//l to a fasting plasma glucose ≥ 7 mmol/l\[[@B17]\].
Family doctors have altered drug therapies in accordance with research evidence and best guidance \[[@B10]-[@B12]\] and have introduced new drugs as they become available. The broad trends in therapy are consistent with published data from the Prescription Pricing Authority\[[@B18]\], the advantage being that our data are patient rather than prescription based and also specific to type 2 diabetes.
Measurement and electronic recording of data of value in the management of diabetes has steadily increased. The only measurement not to show substantial improvement and an increase to a high level of recording is the BMI. It is not clear why this should be, but improvement could be easily achieved. The steady improvement in data recording predates the issue of national guidance and targets and implies that doctors have improved care through the assimilation of best evidence without the need for other inducements.
It has recently been suggested that quality of data recording is not a valid indicator of quality of care\[[@B19]\] and, notwithstanding improvements in monitoring and in drug therapy, it is striking to note that glycaemic control has not improved over the period of observation. This is paralleled by a decrease in the percentage of patients achieving a BMI of \<25 kg/m^2^. The increasing weight of the diabetic population reflects the general increase in weight of the UK population\[[@B20]\], but also may reflect the tendency for many diabetic medications to produce weight gain as an unwanted effect. Data from the US suggest that obesity is rising in newly diagnosed diabetics and that this is associated with poorer prognosis\[[@B21]\]. It is ironic that BMI is the least well recorded measure in this study.
It is tempting to conclude that the lack of improvement in HbA1c control is attributable to the increasing obesity of the diabetic population. However, our analyses of this suggest that the increase in BMI amongst diabetics only had a small effect on the observed HbA1c levels. Further investigation of this issue is certainly warranted, but the statistical analysis is complicated due to the unbalanced data whereby individuals contribute varying numbers of observations, while data recording standards are changing over time, raising the possibility that missing observations are not random. It would also be important to distinguish between changes in BMI as a result of diabetic therapies, and the increasing BMI of the general population.
Other targets in diabetic care are increasingly being met, especially in the area of lipid control. Increased use of drug therapy probably accounts for this and mirrors the increased use of lipid lowering therapy in secondary prevention of ischaemic heart disease\[[@B22]\]. Blood pressure control has improved but is a long way from meeting the lower targets for most patients as suggested by guidance\[[@B4]\]. This may again be consequent on the increasing obesity of the diabetic population and suggests that blood pressure targets in general practice populations may be very difficult to achieve. It should also be noted that blood pressure target levels in UK guidance are rather modest compared to those in US guidance\[[@B6]\].
In contrast to our findings, a recent Swedish study, based on a national diabetes register, reported improvements in HbA1c, as well as in blood pressure levels and increased use of statins, between 1996 and 1999; the Swedish study also reported rising BMI levels \[[@B23]\]. A Dutch study \[[@B24]\] reported similar improvements in blood pressure and cholesterol levels between 1993 and 1999. However, that study also showed improvements in HbA1c. No data on obesity were presented.
Finally, when assessing targets it is important to recognise that complete success is unrealistic. Some patients who comply will not respond to treatment, others will opt-out of treatment, while for others with major co-morbidities the GP may choose not to treat \[[@B24]\]. Further work to assess the realistic target achievable would be worthwhile.
Limitations of the study
------------------------
Our study only reflects what was recorded on the computer systems. It seems reasonable to focus on such data given that these are increasingly being used for monitoring performance in achieving targets. However, it is important to consider whether a rising level of recording or the changing population of diabetics might account for the changes observed. As similar trends were seen in a subset of practices that had always had high levels of recording and low levels of missing values, it is unlikely that rising levels of recording explain our observations. This is consistent with the findings regarding data completeness and quality of care previously alluded to\[[@B19]\]. Comparable trends were also seen in new as well as established diabetics and observed in the control of blood pressure and cholesterol in patients with cardiovascular disease; suggesting that there is consistent improvement in the control of these risk factors.
With DIN, as with all primary care databases, it is important to consider whether the prevalence of disease, recording of risk factors and achievement of targets are nationally representative. The age-sex structure of DIN is identical to the UK average, but more practices are in the south and lower socio-economic groups are under-represented\[[@B13]\]. Nevertheless we have demonstrated that period prevalence rates for a wide range of diseases are similar to those in GPRD\[[@B22],[@B25]\], and this also holds for diabetes (data not shown). There is also evidence that research and non-research practices are similar in disease outcomes\[[@B26]\]. It seems likely that levels of electronic risk factor recording will be above average in an electronic database, although many practices were still doing poorly. While in theory higher levels of recording might improve control of risk factors, there was no clear evidence of this within DIN practices -- at practice level there was no association between level of recording and the lower management target for HbA1c (Pearson correlation = 0.07 in 2001). In summary we believe that the trends seen are typical of those seen nationally, though the overall level of electronic recording is likely to be higher in DIN.
Conclusion
==========
This study highlights the improvements in the process of diabetic care that have been achieved in primary care without the inducements of national targets. Despite this, glycaemic control has not improved and obesity has increased. While increasing obesity does not straightforwardly explain why glycaemic control is not improving it is of itself associated with worse prognosis\[[@B21]\]. Doctors have limited power to affect the weight of their patients\[[@B20]\], and unless weight control in the general population becomes a matter of more importance in national policy, it seems likely that the incidence of diabetes will continue to rise while targets for diabetic control will not be met in most patients.
Competing interests
===================
SdeL is a member of the DIN board -- DIN is a registered charity. NR is a director of a company that provides DIN data commercially.
Authors\' contributions
=======================
SdeL and DC conceived the idea for this paper. CS and IC were responsible for the analyses under the direction of DC. NR advised on issues relating to the DIN database. SdeL, SDeW and DC wrote the paper. All authors commented on drafts of the paper
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2296/6/13/prepub>
Acknowledgements
================
This work was funded by a grant from DIN. IC and SDeW are funded by Wellcome Trust Grant 065177/Z/01/Z
|
PubMed Central
|
2024-06-05T03:55:55.214075
|
2005-3-22
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079812/",
"journal": "BMC Fam Pract. 2005 Mar 22; 6:13",
"authors": [
{
"first": "Simon",
"last": "de Lusignan"
},
{
"first": "Charalambos",
"last": "Sismanidis"
},
{
"first": "Iain M",
"last": "Carey"
},
{
"first": "Stephen",
"last": "DeWilde"
},
{
"first": "Nicky",
"last": "Richards"
},
{
"first": "Derek G",
"last": "Cook"
}
]
}
|
PMC1079813
|
Background
==========
Primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC) are chronic cholestatic liver diseases characterized by a usually slowly progressive course \[[@B1],[@B2]\]. Many patients remain in good clinical condition for many years but may suffer from fatigue interfering with normal activities and general quality of life during a significant part of their life \[[@B3]-[@B5]\]. Fatigue is not related to the severity or activity of the liver disease, and its pathophysiology remains unknown \[[@B3],[@B4],[@B6]\]. In several non-hepatological conditions amino acids, in particular tryptophan and tyrosine, have been reported to be involved in the pathophysiology of fatigue \[[@B7],[@B8]\]. Plasma amino acid abnormalities have been studied extensively in patients with liver failure and hepatic encephalopathy \[[@B9]\]. In patients with less advanced liver disease of various etiologies, significant differences with respect to plasma amino acid concentrations and tyrosine metabolism have been reported in comparison with control individuals. These studies were performed more than two decades ago, at a time when fatigue had not been identified as a significant problem in cholestatic liver disease. Thus far, the potential role of abnormalities in amino acid metabolism in fatigue associated with cholestatic liver disease has not been evaluated and relevant data in PSC are completely lacking.
The present study aimed to identify abnormalities in plasma concentrations of several amino acids and their relation to fatigue and quality of life in patients with PBC and PSC.
Methods
=======
The study was approved by our institution\'s medical ethics committee and informed consent was obtained from each patient. Patients with a diagnosis of PBC (45) or PSC (27) visiting the hepatology outpatient clinic of the Erasmus Medical Center between October 2001 and June 2002 were invited to participate. Exclusion criteria were an age of less than 18 years and incomplete understanding of the Dutch language. As controls with respect to amino acid concentrations, a group of 22 patients with untreated chronic hepatitis C virus infection (HCV) and a group of 73 healthy individuals were included. Fatigue in patients with PBC and PSC was quantified using a visual analogue scale (VAS) and the Fisk fatigue severity scale (FFSS). The FFSS has been validated for use in PBC, and quantifies fatigue in a physical, social and cognitive domain \[[@B6],[@B10]\]. Quality of life was quantified using the SF-36, a widely used quality of life questionnaire \[[@B11]\]. These questionnaires were also obtained from a separate group of 18 age and sex-matched controls, because no questionnaires could be obtained from the previous group of 73 healthy individuals in whom amino acid concentrations were determined. Total serum bilirubin, serum albumin, prothrombin time and serum activities of alkaline phosphatase (AP) and aspartate aminotransferase (AST) were obtained as markers of disease activity and severity. The presence of cirrhosis was determined on the basis of histological and, if not available, clinical criteria (ultrasound findings compatible with cirrhosis if supported by the presence of thrombocytopenia or esophageal varices).
Amino acid measurement
----------------------
Immediately after the venapuncture plasma was prepared by a 20 min centrifugation step at 2650 g and stored at -80°C. The amino acids phenylalanine, tyrosine, tryptophan, isoleucine, leucine and valine were measured by means of high-performance liquid chromatography as described elsewhere \[[@B12]\]. The tryptophan ratio, which is the ratio of tryptophan to the summed concentrations of phenylalanine, tyrosine, isoleucine, leucine and valine, was determined as a measure for central availability of tryptophan for serotonin synthesis. The tyrosine ratio was determined as a measure for central availability of tyrosine for dopamine and norepinephrine synthesis and was calculated as the concentration of tyrosine divided by the sum of the concentrations of phenylalanine, tryptophan, isoleucine, leucine and valine.
Statistics
----------
Testing for differences between groups was performed using Student\'s t-test and the χ^2^test. Correlations were tested using Pearson\'s correlation method. The normality of amino acid distributions was assessed visually using histograms, and non-parametric tests were used where appropriate. The relations between amino acid concentrations and fatigue scores were tested by calculating correlation coefficients for VAS and FFSS domain scores and plasma amino acid concentrations. In these tests, a p-value \<0.01 was considered to be statistically significant. In order to quantify the impact of the differences in amino acid on fatigue, for those amino acids which significantly correlated with fatigue, patients were divided into groups with amino acid concentrations within the 95% confidence interval for healthy controls and patients with concentrations outside this range. Testing for differences in fatigue, quality of life and laboratory parameters between these two groups was performed using Student\'s t-test. Multivariate regression analysis including the biochemical tests of disease activity and severity and the presence of histological or clinical cirrhosis was performed in order to assess the independent association of amino acid abnormalities and fatigue. In all tests other than the correlation tests, a two-sided p-value \<0.05 was considered statistically significant. Statistical analyses were performed using SPSS (Version 9.0, SPSS Inc, Chicago, IL, U.S.A).
Results
=======
Patient characteristics
-----------------------
Patient characteristics for patients with PBC and PSC are shown in table [1](#T1){ref-type="table"}. As was expected because of the unbalanced sex distribution in these diseases, the majority of patients with PBC were female and the majority of patients with PSC were male. The frequency of cirrhosis, serum bilirubin and albumin and serum activities of alkaline phosphatase, AST and ALT did not significantly differ for patients with PSC or PBC.
Amino acids in patients and controls
------------------------------------
Table [2](#T2){ref-type="table"} shows the plasma concentrations of amino acids and the tryptophan and tyrosine ratio\'s for patients with PBC, PSC, HCV and healthy controls. Plasma concentrations of the aromatic amino acids tyrosine and phenylalanine were increased in patients with PBC, whereas in HCV only tyrosine concentration was increased compared to controls.
In PSC, neither of the aromatic amino acids was increased. Tryptophan concentration was decreased in patients with PBC and HCV. Plasma concentrations of the branched chain amino acids valine, isoleucine and leucine were significantly lower in both patients with PBC and PSC. The tryptophan ratio was significantly decreased in patients with PBC and HCV. The tyrosine ratio was significantly increased in all three patient groups.
Within the group of healthy controls, no differences in amino acid concentrations were found for different age groups or sex.
Amino acids and markers of disease activity and severity
--------------------------------------------------------
In patients with PBC, significant inverse correlations were present between the branched chain amino acids valine (p = 0.002), isoleucine (p = 0.006) and leucine (p = 0.007) and total serum bilirubin concentrations. Plasma concentrations of the aromatic amino acids tyrosine (p \< 0.001) and phenylalanine (p = 0.003) correlated inversely with serum albumin concentrations. There was a significant inverse correlation between plasma valine and the serum activity of AST (p = 0.005). Patients with cirrhosis had significantly increased tyrosine (p = 0.004) and phenylalanine (p = 0.03) concentrations and an increased tyrosine ratio (0.004) compared to non-cirrhotics.
However, all differences in amino acid concentrations retained their significance in when only patients without cirrhosis and with normal bilirubin and albumin were compared to healthy controls.
In patients with PSC, no significant correlations were found between any of the markers of disease activity or severity and fatigue or quality of life.
Patients with PSC and inflammatory bowel disease had significantly decreased concentrations of valine, isoleucine and leucine compared to patients with PSC alone (p = 0.02). The concentrations of tyrosine, phenylalanine and tryptophan were not significantly different.
Amino acids, fatigue and quality of life
----------------------------------------
In patients with PBC a significant negative correlation was found between tyrosine concentrations and all fatigue tests. In addition, in these patients a significant negative correlation between tryptophan concentrations and the cognitive domain of the FFSS was found, whereas trends towards significant correlations were found for the other FFSS domains. For the other amino acids, no correlations with fatigue were found (Table [3](#T3){ref-type="table"}). In patients with PSC, no significant correlations between amino acids and fatigue were found.
Comparing PBC patients with normal tyrosine concentrations with patients with increased concentrations resulted in significant differences in VAS (p = 0.03), all domains of the FFSS (p = 0.03, p \< 0.001 and p = 0.01 for the physical, cognitive and social domains, respectively) and the role functioning physical (the extent to which physical health interferes with work or other daily activities) (p = 0.001), bodily pain (p = 0.001), general health (p = 0.03), vitality (p = 0.004), social functioning (p = 0.005), role functioning emotional (the extent to which emotional problems interfere with work or other daily activities) (p = 0.008) and mental health (p \< 0.001) domains of the SF-36 (Figures [1](#F1){ref-type="fig"} and [2](#F2){ref-type="fig"}). In order to assess confounding by disease severity or activity, we performed multivariate analyses for the measurements of fatigue in PBC including plasma tyrosine concentrations and those laboratory tests which correlated with the amino acid, as well as the presence of cirrhosis, although these laboratory tests and the presence of cirrhosis themselves did not correlate with fatigue or quality of life. These analyses showed that only the plasma tyrosine concentration, and not the laboratory tests or the presence of cirrhosis was significantly and independently associated with fatigue.
Comparing patients with normal tyrosine concentrations with healthy controls resulted in the following significant differences: VAS (p \< 0.001), the physical (p \< 0.001) and social (p = 0.004) domains of the FFSS and the physical functioning (p \< 0.001), role functioning physical (p \< 0.001), bodily pain (p = 0.004), general health (p \< 0.001), vitality (p \< 0.001), social functioning (p = 0.001), role emotional functioning (p = 0.05) and mental health (p = 0.04) domains of the SF-36. There was no significant difference in the cognitive domain of the FFSS. Comparing patients with increased tyrosine concentrations with healthy controls showed no significant differences in any of the tests except for worse scores in the general health (p = 0.03) and better scores in the mental health (p = 0.02) domains of the SF-36 for patients with high tyrosine concentrations.
The mean VAS scores were 6.1 and 3.3 for patients with normal and increased tyrosine concentrations, respectively (p = 0.01). Patients with a VAS score \> 5 had a mean tyrosine concentration of 68 μMol/l, whereas patients with a score \< 5 had a mean concentration of 86 (p = 0.02).
Tests for differences in fatigue for patients with normal or decreased tryptophan concentrations did not show significant differences between the two groups.
Discussion
==========
The present study confirms previous findings that significant differences in plasma amino acid concentrations between patients with PBC and healthy controls do exist \[[@B13],[@B14]\]. We found increased concentrations of the aromatic amino acids tyrosine and phenylalanine and decreased concentrations of tryptophan and the branched chain amino acids valine, isoleucine and leucine. Tyrosine concentration correlated with all measurements of fatigue, whereas tryptophan concentrations correlated only with the cognitive FFSS domain. PBC patients with increased tyrosine concentrations reported less fatigue and better quality of life compared to patients with normal concentrations. For PSC, no previous studies on amino acid patterns are available for comparison. We found significant decreases in the plasma concentrations of the branched chain amino acids, and trends towards decreased tryptophan and increased tyrosine and phenylalanine concentrations. However, in contrast to PBC, no relationship with fatigue was found. In addition, we found that valine, isoleucine and leucine concentrations were even lower in patients with PSC and inflammatory bowel diseases than in patients with PSC alone. To our knowledge, no previous data on amino acid concentrations in inflammatory bowel disease are available for comparison.
In several previous studies, mostly on hepatic encephalopathy in patients with advanced cirrhosis, plasma concentrations of amino acids have been studied \[[@B9],[@B15]\]. However, we could identify only two studies including patients with non-cirrhotic PBC. Given the supposedly normal liver function in these patients, these studies somewhat surprisingly found marked differences between patients and controls comparable to those observed in the present study \[[@B13],[@B14]\]. In addition, although the differences appeared to be somewhat smaller, comparable results were obtained in patients with PSC. It remains unclear which mechanisms are responsible for these differences. Although correlations with the markers of disease severity were found, these do not adequately explain the differences in amino acid concentrations, since only a small proportion of the variation in amino acid concentrations could be explained by differences in these markers, and significant differences existed in the majority of patients without cirrhosis and with normal albumin and bilirubin concentrations. Therefore, we suggest other mechanisms, rather than inflammation of the liver or an overall decreased liver function, may be responsible for the noted abnormalities. The nature of these mechanisms, however, remains unknown.
Tyrosine and phenylalanine are mainly metabolized in the liver, suggesting that decreased liver function might result in increased plasma levels. The decreased tryptophan concentrations found in our study might be explained by increased use of tryptophan as a result of immune activation \[[@B15],[@B16]\]. We did not analyze dietary factors that supposedly could influence amino acid concentrations. Previous studies found no evidence to suggest that this is a factor of importance \[[@B13],[@B14]\]. Nearly all patients in the present study were being treated with ursodeoxycholic acid while previous studies reporting comparable plasma amino acid patterns in PBC were performed in the pre-UDCA era \[[@B13],[@B14]\]. Therefore, a role for UDCA in causing these altered patterns seems unlikely.
Fatigue is a significant problem in many patients with PBC and PSC, and has been studied extensively in recent years \[[@B3],[@B4],[@B6],[@B17]\]. However, so far, no specific etiological or pathogenic factors have been identified. Especially, no relation has been found with laboratory parameters for the activity or severity of the disease or histological stage.
An effective medical treatment for fatigue associated with PBC and PSC is not available. Two recent studies specifically addressing PBC-associated fatigue, indicate that treatment with antioxidants is ineffective \[[@B18],[@B19]\].
The present study suggests an association between fatigue and normal tyrosine concentrations in PBC. Concentrations above the 95% confidence interval for healthy controls corresponded with statistically significantly less fatigue and better quality of life scores. Although this suggests that increased tyrosine concentrations may \'protect\' against fatigue and normal concentrations may \'cause\' fatigue, it may well be that that tyrosine plasma concentration alterations are an epiphenomenon and that both these and fatigue are caused by a so far unknown confounding factor or mechanism. Tyrosine is a precursor in the synthesis of dopa, dopamine, epinephrine and norepinephrine, all of which are important neurotransmitters that might play a role in fatigue. Experimental catecholamine depletion has been reported to worsen fatigue, suggesting that a (relative) lack of tyrosine might be associated with fatigue \[[@B20],[@B21]\]. Further, beneficial effects of tyrosine administration in the prevention of exhaustion and fatigue after physical activity in both animals and humans have been reported \[[@B22]-[@B24]\]. Since tyrosine concentrations, and not the tyrosine-ratio was significantly associated with fatigue, a peripheral instead of a central role for tyrosine in the development of fatigue is suggested, which is supported by previous findings supporting peripheral mechanisms in the development of PBC associated fatigue \[[@B17]\]. In addition, other mechanisms, which we did not study, such as abnormalities in the hypothalamo-pituitary-adrenal axis, for example abnormal CRH-release, or manganese homeostasis, might be involved in the development of fatigue in these diseases \[[@B25]\]. In addition, cytokine release as a result of an inflammatory response might also play a role, although studies supporting this hypothesis are lacking. Studies into these mechanisms might therefore be of interest.
It remains unclear why we found a relation between fatigue and tyrosine only in patients with PBC and not in patients with PSC. Although it is likely that fatigue would have a similar etiology in these diseases, the cause of fatigue remains unknown and therefore different mechanisms may occur in these related diseases. Another explanation might be a lack of power to detect a difference in patients with PSC, because less patients with PSC than with PBC were included. On the other hand, although the relation between tyrosine and fatigue was highly significant and occurred for all measures of fatigue, the current finding could be the result of chance or be an epiphenomenon not involved in the pathogenesis of fatigue.
Further studies are therefore required to confirm the present findings and to evaluate the effect of tyrosine suppletion in PBC patients with fatigue.
Conclusion
==========
In conclusion, we showed that in patients with PBC and PSC, marked abnormalities in plasma amino acids occur. In addition, in patients with PBC, normal tyrosine concentrations, compared to increased tyrosine concentrations, may be associated with fatigue and diminished quality of life. This association was independent from the activity and severity of the disease.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
All authors participated in study design and read and approved the final manuscript. PB and HB initiated and coordinated the study, performed patient recruitment and inclusion and prepared the manuscript. JV performed recruitment and inclusion of patients with hepatitis C. PB and HB performed the statistical analysis.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-230X/5/11/prepub>
Acknowledgements
================
None.
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Fatigue and tyrosine.**Mean VAS scores and scores of the physical, cognitive and social domains of the FFSS for PBC patients with normal tyrosine concentrations (blue), high tyrosine concentrations (red) and controls (yellow).
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Quality of life and tyrosine.**Mean SF-36 scores for PBC patients with normal tyrosine concentrations (blue), high tyrosine concentrations (red) and controls (yellow). PF = Physical Functioning, RF = Role Functioning Physical, BP = Bodily Pain, GH = General Health, VI = Vitality, SF = Social Functioning, RE = Role Emotional Functioning, MH = Mental Health.
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Patient characteristics
:::
PBC (n = 45) PSC (n = 27)
--------------------------------------------------- --------------- ----------------
Age (mean, range in years) 58 (34--78) 45 (23--68)
Sex (male / female) 4/45 19/8
Fatigue (yes / no) 32/15 15/12
Cirrhosis (yes / no) 12 / 33 2 / 25
Inflammatory bowel disease (yes / no) 0 / 45 16 / 11
Total serum bilirubin (median, range in μmol/l) 10 (4--98) 15 (6--37)
Serum albumin (median, range in g/l) 42 (31--48) 42 (33--47)
Prothrombin time (median, range in sec.) 12 (10--18) 13 (11--16)
Serum ALT activity (median, range in U/l) 43 (17--484) 35 (10--662)
Serum AST activity (median, range in U/l) 37 (21--241) 31 (18--338)
Serum Alkaline Phosphatase (median, range in U/l) 154 (68--441) 147 (53--1141)
Normal values: bilirubin \<18 μmol/l, albumin 35--50 g/l, ALT \< 41 U/l (male) \< 31 U/l (female), AST \< 37 U/l (male) \< 31 U/l (female), alkaline phosphatase \< 117 U/l, prothrombin time \< 13 sec.
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Amino acid measurements (μMol/l)
:::
----------------------------------------------------------------------------------------------------------------------
PBC (n = 45) PSC (n = 27) HCV (n = 22) controls
---------------------------- ------------------------------ -------------- -------------- -------------- -------------
Aromatic amino acids Tyrosine (mean ± SD) 76 ± 26\ 69 ± 24\ 73 ± 25\ 62 ± 14.4
p = 0.0002 p = 0.07 p = 0.01
Phenylalanine (mean ± SD) 65 ± 15\ 59 ± 11\ 54 ± 8\ 56 ± 6.9
p \< 0.0001 p = 0.11 p = 0.26
Tryptophan (mean ± SD) 39 ± 8\ 43 ± 11\ 40 ± 8\ 46 ± 6.1
p \< 0.0001 p = 0.085 p = 0.0003
Branched chain amino acids Valine (mean ± SD) 211 ± 47\ 208 ± 42\ 232 ± 45\ 238 ± 41.7
p = 0.002 p = 0.002 p = 0.56
Isoleucine (mean ± SD) 55 ± 16\ 53 ± 13\ 67 ± 19\ 70 ± 18.8
p \< 0.0001 p \< 0.0001 p = 0.51
Leucine (mean ± SD) 107 ± 30\ 101 ± 25\ 119 ± 28\ 132 ± 27.1
p \< 0.0001 p \< 0.0001 p = 0.053
Calculated ratio\'s Tryptophan-ratio (mean ± SD) 7.7 ± 1.5\ 8.9 ± 1.8\ 7.5 ± 1.0\ 8.3 ± 1.14
p = 0.02 p = 0.051 p = 0.004
Tyrosine-ratio (mean ± SD) 16.3 ± 6.5\ 15.1 ± 5.0\ 14.0 ± 3.5\ 11.4 ± 2.17
p \< 0.0001 p \< 0.0001 p \< 0.0001
----------------------------------------------------------------------------------------------------------------------
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Correlation between amino acid concentrations and fatigue for patients with PBC
:::
-------------------------------------------------------------------------------------------------------------------------------------
VAS FFSS, physical domain FFSS, cognitive domain FFSS, social domain
---------------------------- ------------------ -------------- ----------------------- ------------------------ ---------------------
Aromatic amino acids Tyrosine -0.38\ -0.37\ -0.40\ -0.37\
**p = 0.01** **p = 0.01** **p = 0.006** **p = 0.01**
Phenylalanine -0.27\ -0.29\ -0.25\ -0.31\
p = 0.08 p = 0.05 p = 0.10 p = 0.04
Tryptophan -0.25\ -0.36\ -0.42\ -0.33\
p = 0.10 p = 0.02 **p = 0.004** p = 0.03
Branched chain amino acids Valine -0.10\ -0.21\ -0.06\ -0.04\
p = 0.50 p = 0.16 p = 0.70 p = 0.79
Isoleucine -0.04\ -0.09\ 0.06\ 0.09\
p = 0.76 p = 0.54 p = 0.70 p = 0.58
Leucine -0.09\ -0.16\ 0.01\ 0.02\
p = 0.57 p = 0.29 p = 0.98 p = 0.89
Calculated ratio\'s Tryptophan-ratio -0.02\ -0.07\ -0.26\ -0.19\
p = 0.87 p = 0.66 p = 0.09 p = 0.22
Tyrosine-ratio -0.30\ -0.21\ -0.34\ -0.32\
p = 0.048 p = 0.15 p = 0.02 p = 0.03
-------------------------------------------------------------------------------------------------------------------------------------
:::
|
PubMed Central
|
2024-06-05T03:55:55.217569
|
2005-3-24
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079813/",
"journal": "BMC Gastroenterol. 2005 Mar 24; 5:11",
"authors": [
{
"first": "Pieter CJ",
"last": "ter Borg"
},
{
"first": "Durk",
"last": "Fekkes"
},
{
"first": "Jan Maarten",
"last": "Vrolijk"
},
{
"first": "Henk R",
"last": "van Buuren"
}
]
}
|
PMC1079814
|
Background
==========
Pancreatic cancer carries a poor prognosis due to the advanced level of tumor invasion and metastasis often encountered at the time of diagnosis \[[@B1]\]. A clear understanding of the cellular processes involved in pancreatic cancer invasiveness and metastasis may be useful in the development of novel forms of therapy. The receptor tyrosine kinase epidermal growth factor receptor (EGF-R) and its ligands are over expressed in pancreatic cancer tissues and in pancreatic cancer cell lines \[[@B2],[@B3]\], with coexpression of receptor and ligand correlating with tumor invasiveness \[[@B4]\]. The mitogenic effects of EGF-R stimulation in pancreatic cancer are well established \[[@B5],[@B6]\]. The importance of EGF-R signalling in promoting and maintaining pancreatic cancer growth is highlighted by studies showing decreased growth in mice treated with a receptor tyrosine kinase blocker \[[@B7]\] and by an EGF-R blocking antibody \[[@B8],[@B9]\].
In contrast to the role of EGF-R and its ligands on cell proliferation, the mechanisms involved in EGF-R mediated invasiveness in pancreatic cancer cells are unclear. Specifically, the role of EGF-R signalling with respect to cellular adhesion to extracellular matrix, cellular motility and invasion through extracellular matrix in pancreatic cancer cells are not known. The involvement of EGF-R and related erbB receptor tyrosine kinases in cancer cell invasion is suggested by studies in colon and mammary carcinoma cell lines. In colon cancer cells, EGF-R activation correlates with the development of liver metastases \[[@B10],[@B11]\]. In breast cancer cells, heregulin-alpha (HRG-α), an EGF family member that binds to erbB3 and erbB4 and facilitates heterodimerization of these receptors with the erbB2 receptor, regulates the actin cytoskeleton and promotes invasion \[[@B12]\]. Signalling via the erbB family of receptor tyrosine kinases therefore is likely to play an important role in mediating pancreatic cancer invasiveness as well.
A combination of signals from cytokine, growth factor and adhesion receptors regulates cell motility. In cancer cells, motility may become dysregulated leading to increased invasive potential. A model of cancer cell invasion would include the loosening of attachment to the primary tumor mass in a process involving detachment from extracellular matrix. The detached cells would then penetrate the extracellular matrix thereby leading to cell migration and invasion. Because EGF and related erbB receptor ligands have been implicated in cellular adhesion and migration \[[@B13]-[@B17]\], we set out to investigate the role of EGF in adhesion and invasion in a pancreatic cancer cell line. We chose Capan-1 cells, as these are well-differentiated cells originally isolated from liver metastases in a patient with pancreatic cancer \[[@B18]\]. We also performed adhesion and invasion studies on MIA PaCa-2 cells, which were originally isolated from a primary pancreatic adenocarcinoma, to determine whether adhesion and invasion characteristics were shared between these cells and Capan-1 cells.
Methods
=======
Materials
---------
Eagle\'s minimum essential media (EMEM), FBS, trypsin-EDTA, penicillin-streptomycin, phalloidin-TRITC, fibronectin, collagen I, laminin IV, AG825, wortmannin and Cell Dissociation Solution were from Sigma (St. Louis, MO). RPMI 1640 media was from GIBCO BRL (Grand Island, NY). Matrigel was from Becton Dickenson Biosciences (Bedford, MA). Recombinant human EGF was from Calbiochem (San Diego, CA). Recombinant heregulin-α was from R&D Systems (Minneapolis, MN). Vitrogen was from Celtrix (Palo Alto, CA). Transwell cell culture plates (diameter 24 mm; pore size 3.0 μm) were from Costar (Cambridge, MA). Six well and 96 well plates were from Becton Dickinson (Franklin Lakes, NJ). Enhanced chemiluminescence detection reagents (ECL-Plus) were from Amersham Pharmacia (Piscataway, NJ). Calcein-AM, To-Pro-3 iodide and fluorescein-conjugated goat anti-rabbit IgG were from Molecular Probes (Eugene, OR). The anti-human mouse monoclonal EGF-R and the anti-human rabbit polyclonal EGF antibodies was from Oncogene (Boston, MA). Anti-human mouse monoclonal erbB2 and erbB3 antibodies were from NeoMarkers (Fremont, CA). For western blotting, α2 integrin and β1 integrin anti-human mouse monoclonal antibodies were from Transduction Laboratories (Lexington, KY). The rabbit anti-human GAPDH antibody was from Novus (Littleton, CO). The human EGF ELISA kit was from Biomedical Technologies (Stoughton, MA).
Cell lines and cell culture
---------------------------
Capan-1 cells were obtained from American Type Culture Collection (Manassas, VA) and cultured in RPMI 1640 supplemented with 15% FBS and 2 mM L-glutamine. MIA PaCa-2 cells were also obtained from American Type Culture Collection and cultured in Dulbecco\'s modified Eagle medium supplemented with 10% FBS and 4 mM L-glutamine. Cells were passaged when confluent. Once confluent, cells were switched over to serum free media gradually over three successive feedings on alternate days, then treated with growth factors in serum free media for the indicated time periods. Human fibroblasts were derived from human gallbladders and cultured on six-well plates as previously described \[[@B19]\].
Adhesion assay
--------------
Adhesion assays were conducted in 96 well plates coated overnight at 4°C with 10 μg/ml collagen I, Matrigel, laminin IV, fibronectin, hyaluronic acid or 1% BSA. After coating, the wells were incubated with 1% BSA in PBS for 1 hour at 37°C, then washed with PBS before use. Cells were labelled for 1 hour with 1 mM Calcein AM at 37°C, washed with PBS, and then detached with 5 mM EDTA in PBS or with Cell Dissociation Solution. Detached cells were washed once more in PBS, resuspended in serum free media and added to the wells (5 × 10^4^cells per well). The plates were then incubated at 37°C for 1 hour. Each well was then washed three times in PBS, and then overlaid with serum free media prior to quantification. The percentage of adherent cells was determined using a Cytofluor 4000 fluorescence plate reader (Applied Biosystems, Foster City, CA), with excitation wavelength 485 nm and emission wavelength 530 nm. Results are expressed as the ratio of fluorescence from wells containing matrix to uncoated control wells containing 5 × 10^4^cells.
For the adhesion assays using inhibitors, growth factor treatment duration was 8 hours. Cells were also treated with the indicated inhibitors at the following final concentrations and durations of preincubation prior to addition of growth factor: cycloheximide 10 μM for 30 minutes; wortmannin 100 nM in DMSO for 60 minutes; AG825 25 μM in DMSO for 60 minutes; erbB2 and erbB3 blocking antibodies 10 μg/ml for 30 minutes. Initial experiments were performed with a range of doses, with the indicated doses being the ones for which data are shown. Inhibitors were present throughout the duration of exposure to growth factor or vehicle. Cell toxicity was monitored using trypan blue exclusion during incubations with inhibitors, with no treatment leading to \>5% cell death.
Invasion assay
--------------
Invasion assays were performed using a modification of a previously described method \[[@B20]\]. Briefly, 8 μM pore HTS Fluoroblok filters (Becton Dickenson) were coated overnight with 300 μl Matrigel or collagen I diluted to 100 μg/ml in PBS at 4°C. The matrix was rehydrated for two hours with serum free media prior to use, then washed once with PBS. Cells were labelled for 1 hour with 1 mM Calcein AM at 37°C, washed with PBS, and then detached with 5 mM EDTA in PBS or with Cell Dissociation Solution. Detached cells were washed once more in PBS, resuspended in serum free media and added to the invasion chamber (1 × 10^5^cells per well). The chambers were incubated at 37°C for up to 48 hours. The percent of cells invading to the bottom of the membrane was then quantified using a Cytofluor 4000 fluorescence plate reader. Results are expressed as the ratio of fluorescence from cells invading to control wells containing 1 × 10^5^cells.
Confocal laser scanning immunofluorescence microscopy
-----------------------------------------------------
Cells were cultured in 24 mm diameter Transwell inserts until confluent. Cells were washed twice with ice cold PBS, then a square of membrane was excised. This square was then fixed in methanol/acetone (1:1; v/v) for 15 minutes at -20°C. The fixed cells were then washed in ice cold PBS and incubated for 30 minutes in the presence of PBS containing 1% BSA. The cells were then washed in PBS and incubated in primary antibody diluted in PBS containing 1% BSA. After one hour, the cells were again washed and incubated with secondary antibody in PBS containing 1% BSA. The secondary antibody was removed after one hour by washing with PBS. To-Pro-3 iodide was applied for 1 hour. Cells were washed three times with PBS and sealed in Vectashield Hard Set mounting medium (Vector Labs, Burlingame, CA). Scanned images were acquired with 400× optical and 3× digital magnification using a laser scanning spectral confocal microscope system (Leica DM-R upright fluorescence microscope and Leica TCS-SP confocal scanner). Single serial section images in the xy plane were acquired. Cells treated with secondary antibody only did not show detectable signals at the parameters used for acquiring images.
Immunoblotting
--------------
Cells were cultured to confluency on six well plates and harvested with SDS loading buffer (250 mM Tris, ph 6.8; 4% SDS; 10% glycerol; 0.006% bromphenol blue; and 2% β-mercaptoethanol). Protein content of cell extracts was measured by the Lowry method. SDS-PAGE was performed followed by transfer to PVDF membrane. The membranes were blocked with 10% BSA in PBS/Tween 20 (0.05%; v/v) at 4°C for 16 hours and then incubated with the anti-α~2~integrin or β~1~integrin antibodies for 1 hour at room temperature. The membrane was then washed with 0.05% Tween 20 in PBS and incubated with peroxidase-conjugated anti-mouse IgG antibody for 1 hour at room temperature. The membrane was washed and incubated with ECL-Plus and autoradiography performed. We also ran blots in parallel that were hybridized with rabbit anti-human GAPDH antibody to confirm equal protein loading.
ELISA assay
-----------
Capan-1 cells and human gallbladder myofibroblasts were cultured until confluent in 6 well plates. Upon achieving confluency, cells were then cultured in media containing 10% FBS or in serum free media. Following 48 hours of incubation, media was collected, centrifuged to pellet cell debris and floating cells, and soluble EGF levels quantitated using a human EGF ELISA kit, using 96 well plates and a microplate fluorescence reader per the manufacturer\'s instructions. The sensitivity of this method was to 0.5 ng/ml soluble EGF.
Statistical analysis
--------------------
A minimum of three separate experiments (with each treatment condition performed in triplicate) was performed for the adhesion and invasion assays. Results of experiments are expressed as the mean percentage ± SD of cells that are adherent or invading compared to cells from control wells. Student\'s t test was used, and p \< 0.05 was considered significant.
Results
=======
Capan-1 adhesion to collagen I is inhibited by EGF
--------------------------------------------------
We first examined the ability of Capan-1 cells to adhere to various extracellular matrix components. As shown in Figure [1](#F1){ref-type="fig"}, Capan-1 cells adhered to a variety of matrices, including laminin IV, fibronectin, collagen I and Matrigel. Capan-1 cells did not adhere to any significant degree to hyaluronic acid or to BSA.
We then determined the effects of EGF on adhesion to these extracellular matrix components. For this set of experiments, EGF treatment was provided for 8 hours prior to the initiation of the adhesion assay. Adhesion to collagen I decreased upon treatment with EGF. Similarly, adhesion to Matrigel decreased with EGF treatment. In contrast, treatment with EGF increased adhesion to fibronectin. There was no difference in adhesion to laminin IV. Finally, even though adhesion to hyaluronic acid was low at baseline, EGF treatment increased adhesion. No significant change in adhesion was noted on BSA with EGF treatment. EGF therefore has varied effects on adhesion to various extracellular matrix components, mediating detachment from collagen I and Matrigel, while potentiating adhesion to fibronectin and hyaluronic acid.
Heregulin-α (HRG-α) belongs to the same family as EGF, binds to erbB3 and erbB4 receptors and is capable of heterodimerization with the erbB2 receptor in other cell types \[[@B12]\]. We therefore used HRG-α as an indirect means to determine whether the EGF-mediated effects on adhesion were shared by activation of erbB3 and erbB4 receptors. We tested the effects of HRG-α on the adhesion of Capan-1 cells to the same set of extracellular matrix components and for the same duration of treatment. In contrast to the results with EGF, there was no significant inhibition of adhesion to collagen I with HRG-α treatment. In addition, HRG-α did not alter adhesion to laminin IV, fibronectin, hyaluronic acid or BSA. There was a significant decrease in adhesion to Matrigel (from 57.95 % ± 2.5 to 33.6 % ± 2.9; p = 0.003) with HRG-α treatment. These results imply that the inhibition of adhesion to collagen I occurred without involvement of erbB3 or erbB4 receptors. The inhibition of adhesion to Matrigel, however, appeared to involve receptors activated by either EGF or HRG-α.
Evidence that the EGF-mediated detachment from collagen I in Capan-1 cells involves erbB1 and erbB2 receptors and the phosphatidylinositol-3 kinase signalling pathway
----------------------------------------------------------------------------------------------------------------------------------------------------------------------
We focused on adhesion to collagen I by Capan-1 cells given the marked inhibition of adhesion seen with EGF. In order to begin to delineate the pathways involved in this detachment, we examined the effect of various inhibitors: cycloheximide, a protein synthesis inhibitor; wortmannin, an inhibitor of phosphatidylinositol-3 kinase; AG825, a pharmacologic inhibitor of the erbB2 receptor; a blocking antibody against the erbB2 receptor; and a blocking antibody against the erbB3 receptor. Each of these inhibitors did not alter adhesion to collagen I to any significant degree in the absence of growth factor treatment (Figure [2](#F2){ref-type="fig"}). EGF-mediated de-adhesion to collagen I was not reversed by cycloheximide, suggesting that new protein synthesis was not involved. Similarly, treatment with a blocking antibody against the erbB3 receptor did not alter this EGF-mediated detachment. On the other hand, treatment with wortmannin reversed the EGF-mediated detachment, implying a phosphatidylinositol-3 kinase mediated pathway was involved (EGF = 30.5 % ± 2.1 vs. EGF + wortmannin = 50.5 % ± 3.8; p \< 0.001; n = 10). Similarly, an inhibitor of the erbB2 receptor, AG825 and a blocking antibody against erbB2 reversed the EGF-mediated detachment (to 49.9 % ± 1.7; p = \<0.001; n = 10; and to 48.6 % ± 2.4; = \<0.001; n = 5, respectively). The similar magnitude of this inhibition of the EGF-induced detachment suggested that the erbB2 receptor signalling pathway and the phosphatidylinositol-3 kinase pathway were involved simultaneously.
We also examined the effects of this panel of inhibitors on cells adhering to collagen I and treated with HRG-α. There was no effect on adhesion to collagen I with HRG-α treatment, a finding that was not altered with treatment with wortmannin or the erbB2 receptor inhibitor AG825 (Figure [2](#F2){ref-type="fig"}). Furthermore, blocking antibodies against the erbB2 and erbB3 receptors did not alter this lack of effect of HRG-α on adhesion to collagen I (data not shown). Cycloheximide decreased adhesion in the presence of HRG-α, (from 57.0 % ± 1.6 to 50.4 % ± 2.1; p = 0.02; n = 10), suggesting that protein synthesis was involved to a limited degree in maintaining adhesion to collagen I with HRG-α treatment. Overall, these results suggested that the detachment from collagen I was an EGF-specific effect, which was not shared by the related growth factor, HRG-α. This de-adhesive effect is therefore likely mediated through activation of the EGF-R and partially through activation of the erbB2 receptor, and involves the phosphatidylinositol-3 kinase signalling pathway.
Lack of a de-adhesion response to Collagen I of EGF in a primary pancreatic cancer cell line (MIA PaCa-2)
---------------------------------------------------------------------------------------------------------
MIA PaCa-2 cells are derived from a primary, non-metastatic pancreatic adenocarcinoma, unlike Capan-1 cells, which were isolated from liver metastases. We therefore asked whether the adhesion response to EGF on collagen I was a property shared between these two pancreatic cancer cell lines. Adhesion to collagen I was markedly diminished in MIA PaCa-2 cells under control conditions compared to Capan-1 cells. Treatment with EGF (10 nM for 8 hours) did not show a significant change in adhesion. Strikingly, however, treatment with HRG-α showed a marked increase in adhesion to collagen I (Figure [3](#F3){ref-type="fig"}). This effect of HRG-α was abrogated by pre-treatment with either cycloheximide or the erbB-2 inhibitor, AG825, suggesting that protein synthesis and activation of the erbB-2 receptor were involved in the HRG-α response. HRG-α, and by implication, activation of the erbB3 and/or erbB4 receptors in concert with erbB-2 receptor activation, appears to mediate adhesion to collagen I in this cell line.
EGF stimulates invasion through collagen I and Matrigel in Capan-1 cells
------------------------------------------------------------------------
We next examined the effects of EGF on invasion through collagen I and Matrigel. We chose collagen I because of the clear detachment response to EGF noted above. We also chose to examine invasion through Matrigel, as this is a more complex matrix that simulates extracellular matrix *in vivo*and serves as a more physiologic substrate on which to measure invasion capabilities of Capan-1 cells. Furthermore, the studies on adhesion had shown parallel effects with EGF treatment, suggesting that collagen I and Matrigel possessed similar properties with respect to adhesion and invasion. Capan-1 cells showed an increase in invasion following addition of EGF which was seen with both collagen I (data not shown) and with Matrigel (Figure [4](#F4){ref-type="fig"}). No effect on invasion through either collagen I or Matrigel was seen with HRG-α.
EGF stimulates invasion through Matrigel in MIA PaCa-2 cells
------------------------------------------------------------
We also assessed invasion through Matrigel with and without growth factor treatment on MIA PaCa-2 cells. As shown in Figure [4](#F4){ref-type="fig"}, EGF treatment enhanced invasion to a similar extent as in Capan-1 cells, whereas HRG-α had no effect. Therefore, the de-adhesive response to EGF is absent in the MIA PaCa-2 cells, whereas the invasive response to Matrigel is preserved.
Time course of EGF-mediated inhibition of adhesion and stimulation of invasion
------------------------------------------------------------------------------
Given the EGF-mediated effects on both adhesion and invasion, which were not seen with HRG-α, erbB1/erbB2 receptor signalling pathways were likely involved in mediating both detachment from extracellular matrix and the stimulation of invasion noted in Capan-1 cells. We explored the effects of EGF signal duration on detachment and invasion in Capan-1 cells. EGF treatment led to a time-dependent detachment from collagen I, which was slowly reversible (Figure [5](#F5){ref-type="fig"}). The peak inhibition of adhesion occurred with 8 hours of EGF treatment.
The time course of invasion through Matrigel showed that the EGF-mediated increase in invasion peaked at 24 hours of treatment, and then began to slowly return to pre-treatment levels (Figure [6](#F6){ref-type="fig"}). Therefore, the EGF-mediated inhibition of adhesion was followed by an EGF-mediated increase in invasion, suggesting a temporal relationship between loss of attachment to extracellular matrix and acquisition of the ability to invade.
Dose response of EGF-induced detachment from collagen I and invasion through Matrigel
-------------------------------------------------------------------------------------
This temporal relationship was supported by the finding that the EGF dose response for both adhesion and invasion were similar (Figure [7](#F7){ref-type="fig"}). Concentrations greater than 0.01 nM were required to see effects on both of these phenomena. For these studies, the duration of treatment with EGF was held constant at 24 hours, a time when both detachment and invasion were statistically significantly different between treated and non-treated cells. The dose of EGF chosen for the earlier assays on adhesion and invasion (10 nM) was justified by these dose response studies, as at that concentration a plateau of both detachment and invasion were seen.
Expression of EGF receptor, α~2~integrin and β~1~integrin in Capan-1 cells
--------------------------------------------------------------------------
EGF effects are mediated predominantly through the EGF receptor (EGF-R). EGF-R expression was demonstrated in Capan-1 cells by confocal immunofluorescence microscopy (Figure [8A](#F8){ref-type="fig"}). As effects of adhesion and invasion were likely to involve integrins, we sought to demonstrate expression of the integrins involved in adherence to collagen I, the α~2~and β~1~integrins. As shown in Figure [8B](#F8){ref-type="fig"}, α~2~integrin was expressed on the cell surface of Capan-1 cells. Similarly, the β~1~integrin was expressed on the surface of Capan-1 cells (Figure [8C](#F8){ref-type="fig"}). These results demonstrate that the EGF-mediated effects of adhesion and invasion in Capan-1 cells were likely to involve the EGF receptor, as well as the collagen I receptor, the α~2~β~1~integrin heterodimer.
We next asked whether α~2~/β~1~integrin expression levels were affected by EGF treatment. No significant difference in signal for either α~2~integrin or for β~1~integrin was noted, either by immunofluorescence or immunoblot analysis, following treatment with 10 nM EGF (data not shown).
EGF treatment induces rearrangements in the actin cytoskeleton
--------------------------------------------------------------
ErbB1/erbB2 receptor tyrosine kinase signalling mediated by EGF binding has been linked to integrin expression via effects on the focal adhesion complex \[[@B21]\]. One important downstream effect of changes in the focal adhesion complex is actin cytoskeleton rearrangement. Therefore, we looked for changes in actin cytoskeleton architecture using phalloidin staining in cells treated with EGF. As shown in Figure [9](#F9){ref-type="fig"}, phalloidin staining showed focal areas of staining suggestive of focal adhesion complexes in cells treated with EGF compared to untreated cells.
Expression of EGF in Capan-1 cells does not correlate with EGF secretion
------------------------------------------------------------------------
*In vivo*, the source of EGF that drives a cancer cell to become invasive may be adjacent cancer cells, adjacent mesenchymal cells, or the cancer cell itself. In order to determine the source of EGF involved in these EGF-mediated effects on adhesion and invasion in Capan-1 cells, we looked for evidence of endogenous EGF synthesis and secretion. An ELISA for human EGF showed no soluble EGF secreted by Capan-1 cells, whereas cultured human myofibroblasts secreted soluble EGF, consistent with previous reports (Figure [10A](#F10){ref-type="fig"}) \[[@B22]\]. On immunofluorescence microscopy with an EGF antibody, however, EGF was expressed on the plasma membrane of Capan-1 cells (Figure [10B](#F10){ref-type="fig"}). These results suggest that Capan-1 cells synthesize but do not secrete EGF.
Discussion
==========
Drugs that target the EGF receptor offer promise for more effective treatment of pancreatic cancer \[[@B1]\]. For example, Gefitinib, a selective EGF receptor tyrosine kinase inhibitor, inhibits pancreatic cancer cell growth and invasion \[[@B23]\]. Understanding the mechanisms of EGF-induced cellular invasion in pancreatic cancer cells therefore is critical, as such insights could provide a means to define subpopulations of tumors more likely to be responsive to these agents and allow for more focused targeting strategies in future drug design \[[@B24]\]. In the current work, we have provided data to support a model in which EGF-mediated effects on de-adhesion and invasion occur via the EGF-R and erbB-2 receptor involving the PI3 kinase signalling pathway in Capan-1 cells. The involvement of the PI3-kinase pathway has also been reported in glial cell line-derived neurotrophic factor-induced migration and invasion in pancreatic cancer cells, suggesting that this signalling mechanism may be common to the invasion phenotype in pancreatic cancer cells \[[@B25]\]. Furthermore, we provide evidence that EGF-mediated de-adhesion and invasion are temporally linked phenomena, occurring in cells that are capable of synthesizing both EGF and EGF-R. The lack of EGF secretion by these cells suggests that the drive to invasiveness in these cells may be provided by cell-cell interactions that promote EGF-R activation, although alternative mechanisms involving EGF have not been ruled out by our results.
It is interesting to compare and contrast the results obtained by Stefani et al with our data, because these investigators used the same pancreatic cancer cell lines (Capan-1 and MIA PaCa-2), and assessed the response to EGF on adhesion on the same ECM components as in our study \[[@B26]\]. In our study, EGF-mediated de-adhesion in Capan-1 cells occurred with collagen I and Matrigel, and not with other extracellular matrix components. This finding is similar to results reported by Stefani et al in Capan-1 cells, although in that study, the de-adhesion response to EGF in Capan-1 cells did not reach statistical significance and no further studies involving the mechanism involved in this response were conducted. One important difference between the methodologies used in these studies was in the time course of the adhesion assay. Stefani et al treated their cells with daily EGF treatments for 5--6 days prior to performance of the adhesion assay. In contrast, our studies were performed on a much shorter time scale. Indeed, our results can be reconciled by examination of the time course results presented in Figure [5](#F5){ref-type="fig"}, which shows that continuous exposure to EGF for more than 32 hours leads to an abrogation of the de-adhesive response. Therefore, the de-adhesive response to EGF in Capan-1 cells reported at 6 days by Stefani et al may have been less significant than our results because the effect of EGF was more pronounced at earlier time points. A similar explanation would account for the differences in adhesion to collagen I measured with and without EGF treatment in MIA PaCa-2 cells.
The α~2~/β~1~integrin heterodimer is the collagen I receptor, which is found in numerous tissues \[[@B27],[@B28]\]. The adhesion assay results with collagen I implicated the α~2~/β~1~integrin heterodimer, which are expressed in Capan-1 cells \[[@B29]\]. Higher expression of α~2~integrin in pancreatic cancer tissue compared with normal pancreas has been reported previously \[[@B30]\]. However, the lack of change in expression of α~2~/β~1~integrin with EGF treatment in Capan-1 cells suggested that functional activation of these integrins were more important than absolute expression levels in mediating the effects of EGF on adhesion. This effect of EGF differs from that described in human cutaneous squamous carcinoma cells, in which treatment with this growth factor induces expression of α~2~/β~1~integrins and enhances adhesion and migration on type I collagen \[[@B31]\]. Therefore, the lack of increase in expression of α~2~/β~1~integrin, and the de-adhesion response, are features that distinguish the EGF-induced adhesion effects in Capan-1 cells from that in squamous carcinoma cells.
The studies with phalloidin, showing changes in localization of actin microfilaments upon EGF exposure, implicate a functional effect on focal adhesion complexes as well. EGF receptor activation has been linked to focal adhesion complex changes via activation of the focal adhesion kinase, FAK \[[@B21]\]. As motility signalling derived from EGF-R activation required FAK \[[@B21]\], the de-adhesion and invasion responses noted in Capan-1 cells likely involve FAK activation also. Further experiments are necessary to prove that this link between EGF-R activation and FAK activation leading to focal adhesion complex regulation and cell motility is in fact present in Capan-1 cells as it is in other types of cancer cells \[[@B32]-[@B34]\].
The de-adhesive response was temporally and dose-dependently linked to stimulation of invasion in Capan-1 cells, suggesting that these responses are linked. The effects of EGF on invasion through collagen I and Matrigel were also significant, and differed from HRG-α, which did not display this effect. β~1~integrins have been implicated to play a key role in pancreatic cancer cell invasion \[[@B35]\]. Furthermore, α~2~integrins have also been implicated in EGF-induced invasion through collagen in keratinocytes \[[@B36]\]. Taken within this context, the finding that EGF (but not HRG-α) induces invasion through collagen I and Matrigel implicates the α~2~/β~1~integrin in this response, acting via EGF-R and erbB2 receptors. MIA PaCa-2 cells share these properties with respect to invasion through Matrigel. Metastatic dissemination mediated by increased expression of α~2~/β~1~integrin has been shown in ovarian carcinoma cells, so integrin-mediated invasion capabilities in Capan-1 and MIA PaCa-2 cells may also be operative \[[@B37]\]. In this respect, Capan-1 and MIA PaCa-2 cells also differ from breast carcinoma cells, in which re-expression of α~2~/β~1~integrins correlates with abrogation of the malignant phenotype \[[@B38]\].
Recent reports have implicated the MUC4 mucin as an intramembranous ligand for erbB2 \[[@B39]\]. Specifically, the rat Muc4/Sialomucin complex interacts with erbB2 and induces its autophosphorylation and translocation from the basolateral to the apical membrane in CaCo-2 cells \[[@B40]\]. Antisense plasmid-mediated down-regulation of MUC4 expression in a pancreatic cancer cell line enhanced adhesion and decreased motility, findings that are consistent with our data \[[@B41]\]. While the pancreatic cancer cell line used in that study was not Capan-1, and while adhesion to and migration through plastic was assessed, the fact that the general findings between the report of Singh et al and our data are in agreement suggests that erbB-2 signalling effects on adhesion and invasion may not be limited to one particular pancreatic cell line. In this context, the contrasting results on the collagen I adhesion assay between Capan-1 cells and MIA PaCa-2 cells indicate that the de-adhesive response to EGF appears to be a feature characteristic of a metastatic cell, whereas stimulation of adhesion to collagen I by HRG-α appears to be a feature of a primary pancreatic adenocarcinoma cell. Further experiments comparing these and other pancreatic cancer cell lines would need to be performed to determine whether these differential responses to EGF and HRG-α hold true.
Finally, the finding that Capan-1 cells expressed but did not release EGF into the media has implications for the self-sustaining stimulation of the EGF-R with respect to invasion capabilities. Pancreatic cancer cells over express EGF, along with other members of the EGF family such as Cripto and epiregulin \[[@B42],[@B43]\]. EGF-R, erbB-2 and related members of the EGF receptor tyrosine kinases are also over expressed in pancreatic cancer \[[@B44]\]. The co-expression of these EGF families of ligands and receptor tyrosine kinases suggests that an autocrine mechanism is operative in pancreatic cancer, thereby allowing autonomous activation of signalling pathways that manifests as an aggressive tumor phenotype. \[[@B4],[@B6]\]. Our data suggests that, at least in Capan-1 cells, this autocrine mechanism may need to be refined so that cell-cell interactions between membrane-expressed but non-secreted EGF stimulates EGF-R on adjacent cells. Pancreatic cancer cells destined to invade extracellular matrix may therefore need to perform this function in groups rather than as individual cells.
Conclusion
==========
In summary, our studies on Capan-1 cells have shown that EGF-R and erbB2 receptor are likely involved in mediating a temporally linked de-adhesive and invasion response. These cellular functions involve the participation of α~2~/β~1~integrins, which induce changes in the actin cytoskeleton at points of contact with the extracellular matrix. Finally, the autocrine nature of this growth factor signalling likely involves cell-cell interactions as EGF is synthesized but not secreted by these cells. These insights refine our understanding of the complex cellular phenomena that occur *in vivo*in the setting of pancreatic cancer invasion and metastasis, and open up avenues for further investigation into the mechanisms involved.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
AJS participated in the conception and design of the study, analyzed the data, and performed the experiments. RK participated in the conception and design of the study, analyzed the data, and wrote the manuscript. Both authors read and approved the final version of the manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-230X/5/12/prepub>
Acknowledgements
================
This work was supported by NIH grant DK 02609 to RK.
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Adhesion of cells to various extracellular matrix components and the effect of EGF.**Adhesion was tested under conditions of no growth factor (open symbols) and treatment with 10 nM EGF (closed symbols) for 8 hours. Results are expressed as % adhesion ± SEM from three separate experiments, each performed in triplicate wells. \* p \< 0.003; \*\* p \< 0.05.
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Effects of various inhibitors on adhesion of cells to collagen I.**Adhesion assays on collagen I were performed with no growth factor (open bars), 10 nM EGF (closed bars) or 10 nM HRG-α (hatched bars). Shown are the results of three separate experiments, each performed in triplicate. \* p \< 0.001. \*\*p \< 0.02.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Effects of cycloheximide and AG825 on adhesion of MIA PaCa-2 cells to collagen I.**Adhesion assays on collagen I using were performed with no growth factor (open bars), 10 nM EGF (closed bars) or 10 nM HRG-α (hatched bars). Shown are the results of three separate experiments, each performed in triplicate. \* p \< 0.001.
:::
:::
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Effects of EGF and HRG-α on invasion through Matrigel in Capan-1 and MIA PaCa-2 cells.**Cells were cultured on invasion chambers coated with Matrigel, and invasion measured following treatment with no growth factor, 10 nM EGF, or 10 nM HRG-α. Shown are the results of 3 experiments, each performed in triplicate wells. \*p \< 0.05 between the no growth factor group and the EGF treatment group for each cell type.
:::
:::
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Time course of EGF-mediated inhibition of adhesion.**The adhesion assay was performed and the experiment terminated at various time points. Adhesion was measured without and with EGF treatment. Shown are the results of one experiment performed in triplicate wells, which was repeated twice with similar results.
:::
:::
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**The time course of EGF-mediated invasion through Matrigel.**Cells were labeled and invasion through chambers performed in the absence or presence of 10 nM EGF. Shown are the results of one experiment performed in triplicate wells, which was repeated twice with similar results.
:::
:::
::: {#F7 .fig}
Figure 7
::: {.caption}
######
**Dose response of EGF-induced detachment from collagen I and stimulation of invasion through Matrigel.**Adhesion and invasion were measured following incubation with escalating doses of EGF. Shown are the results of three experiments, each performed in triplicate.
:::
:::
::: {#F8 .fig}
Figure 8
::: {.caption}
######
**Expression of EGFR, α~2~/β~1~integrin in Capan-1 cells.**Cells were cultured on Transwell inserts, and confocal immunofluorescence microscopy performed following incubation with primary antibodies against EGF-R (A), α~2~integrin (B) and β~1~integrin (C). Negative controls with primary antibody omitted did not show a detectable signal.
:::
:::
::: {#F9 .fig}
Figure 9
::: {.caption}
######
**Phalloidin-TRITC staining of cells with and without EGF treatment.**Cells were cultured on Transwell inserts, and confocal immunofluorescence microscopy performed following incubation with TRITC-conjugated phalloidin, in the absence (A) or presence (B) of 10 nM EGF. Arrow shows focal area of actin staining consistent with a focal adhesion complex.
:::
:::
::: {#F10 .fig}
Figure 10
::: {.caption}
######
**Soluble EGF levels in incubation media from Capan-1 cells compared to media from human gallbladder myofibroblasts.**Cells were cultured in media with or without 10% FBS. Shown are the results of one experiment, which was repeated with similar results (A). B. Confocal immunofluorescence image of confluent Capan-1 cells stained for EGF.
:::
:::
|
PubMed Central
|
2024-06-05T03:55:55.219843
|
2005-3-31
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079814/",
"journal": "BMC Gastroenterol. 2005 Mar 31; 5:12",
"authors": [
{
"first": "Andrew J",
"last": "Shirk"
},
{
"first": "Rahul",
"last": "Kuver"
}
]
}
|
PMC1079815
|
Background
==========
The term \"isolation by distance\" (IBD) was first used by Sewall Wright \[[@B1],[@B2]\] to describe patterns of population genetic variation that derive from spatially limited gene flow. IBD is defined as a decrease in the genetic similarity among populations as the geographic distance between them increases. Statistical tests for IBD can be conducted using populations or individuals as the units of replication, although analyses at the individual level typically utilize spatial autocorrelation statistics \[[@B3]\]. When individuals can be grouped into populations, it is possible to calculate 1) a matrix of genetic similarity or distance between all population pairs for comparison with 2) a matrix of pairwise geographic distances. A nonparametric Mantel test is typically used to test for non-random associations between these two matrices. Bohonak \[[@B4]\] has also suggested performing statistical tests of the IBD slope and intercept using bootstrapped pseudoreplicates, and recently, Yang \[[@B5]\] has developed a likelihood-based approach for IBD analysis. The ever-growing volume of population genetic data and a continued desire to interpret spatial processes make computational tools for rapidly identifying patterns of isolation by distance highly desirable.
In this paper we describe a sophisticated, yet user-friendly, Common Gateway Interface (CGI) implementation of statistical software for detecting IBD patterns. Isolation by Distance Web Service (IBDWS) is a major expansion and upgrade of Isolation By Distance (IBD), a standalone computer program written in C and originally compiled for Windows and Macintosh \[[@B4]\]. IBDWS includes the following highly useful features for analysis of genetic structure in natural populations:
\(1) [Analysis of raw data]{.underline}: For a raw data set with codominant markers (e.g., microsatellites), IBDWS calculates basic summary statistics including heterozygosity, *F*~*ST*~between all pairs of populations, and standard transformations of *F*~*ST*~.
\(2) [Statistical tests]{.underline}: IBDWS performs Mantel tests, partial Mantel tests, reduced major axis (RMA) regression of IBD slope and intercept, and error estimation for slope and intercept using multiple methods.
\(3) [Simple form interface]{.underline}: Various parameters (e.g., number of randomizations) are easily configured on the IBDWS front page and are checked for validity.
\(4) [Analysis selection]{.underline}: Multiple analyses may be performed via checkbox selections.
\(5) [Publication quality graphics]{.underline}: IBD scatter plots are generated in Postscript format for download and further modification.
\(6) [Rapid analysis]{.underline}: IBDWS is hosted on a fast server with minimal wait times (see Results and discussion).
The majority of these features were not included in IBD. Additionally, IBDWS includes thirty-one updates and bug fixes to the original IBD statistical package. Altogether, these changes make IBDWS a major advancement on the original release in terms of both performance and utility.
Implementation
==============
IBDWS was written in C++, compiled with the Unix g++ compiler, and runs on a server using Apache 2.0.50 and Fedora Linux. Postscript file plotting capability (see below) was implemented using CGraph, a C++ plotting library. The UNIX command \"convert\" is called by IBDWS to create a JPEG for web-based viewing from the postscript file. Both files are held on the server for 24 hours until routine deletion.
Depending on the specific configuration options, IBDWS performs four primary functions. The first is generating standard summary statistics from raw data (if provided). These include a matrix of the genetic distance *F*~*ST*~\[[@B1]\] between all pairs of populations, and several standard transformations of this statistic \[see \[[@B6],[@B7]\]\]. Second, the scatter plot of genetic vs. geographic distance is generated in JPEG format (for viewing on the web page) and in vector-based Postscript format (for downloading and manipulating in a graphics design program). The Reduced major axis regression line overlays the plot (see Figure [1](#F1){ref-type="fig"}). An algorithm that evaluates the range of values on each axis chooses an appropriate set of evenly spaced units for labelling.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Example scatter plot.**The plot was generated using the example raw allelic data and geographic information from the IBD.2 example data set available on the IBDWS website. The RMA regression line overlays the scatter plot. The plot was downloaded in postscript format from the website and slightly modified using Adobe Illustrator.
:::
:::
Third, IBDWS assesses whether there is a statistically significant relationship between the genetic distance (or similarity) matrix and the comparable matrix of geographic distances. As in other software applications, a Mantel test \[[@B8]\] is used to test this relationship, as it properly accounts for the population as the unit of replication, rather than population pairs (i.e., the matrix entries). In addition, IBDWS calculates the slope and intercept of the IBD relationship using reduced major axis (RMA) regression. RMA regression is more appropriate than standard linear regression when both the dependent and the independent variables are measured with error. Error for the slope and intercept are estimated using standard linear approximations, using a one-delete jackknife across population pairs, and by bootstrapping pseudoreplicates over *independent*population pairs (see \[[@B4]\] and the online IBD manual). The final two methods are preferred, as they correctly treat the population as the unit of replication. Figure [2](#F2){ref-type="fig"} provides examples of IBDWS statistical output.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**IBDWS example statistical results output.**The figure shows a partial list of statistical output calculated from the IBD.2 sample data set. Results of Mantel tests and other calculations are also provided. All results can be downloaded as text files.
:::
:::
Different researchers have recommended logarithmic (base 10) transformation of genetic distance, geographic distance or both, prior to IBD analysis \[[@B6],[@B9]\]. IBDWS automatically performs each of these transformations and provides an interface that switches among the four possible scatter plots.
Results and discussion
======================
Input
-----
IBDWS analyzes genetic data with or without corresponding geographical information. Without geographic distances, summary statistics can be calculated from raw data sets that include diploid genotypes at one or more loci. These statistics include heterozygosity, and estimates of genetic similarity between all population pairs. The additional input of geographic distance between all population pairs facilitates the variety of IBD calculations described above. The geographic distance matrix can be calculated using simple Euclidean distances, or with more complex algorithms that follow biologically relevant landscape features. Input data can be uploaded as an ASCII text file or the data can be inserted into an HTML form on the IBDWS web page (Figure [3](#F3){ref-type="fig"}). White space delimiters are used between fields. If the population genetic distance matrix is entered directly (rather than individual-level data), the input file may be in matrix or list format (examples of each are provided on the IBDWS home page). Indicator labels (e.g., GENETIC\_DISTANCE, GEOGRAPHIC\_DISTANCE) are required to separate blocks in the input file or test string, assuring that it is parsed correctly. Additional documentation regarding input for partial Mantel tests is provided in the IBD manual (available at the web site).
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**IBDWS forms interface.**The figure shows a portion of the CGI web interface. Data can be uploaded from a text file using the Browse button or pasted directly into the form. Below the data input portion is a partial list of checkboxes for tailoring analyses. A total of twelve parameters are modifiable by the user.
:::
:::
Checkboxes and radio buttons allow the user to configure which analyses will be performed (Figure [3](#F3){ref-type="fig"}). These include the calculation of Rousset\'s genetic distance \[[@B7]\] from raw data, options to log-transform genetic and/or geographic distances, replacement values for undefined transformations, and the number of bootstrap randomizations per analysis (up to 30,000).
Output
------
A full analysis of raw data for six loci, two to three alleles per locus, and 326 individuals required an average wait time of 10 seconds. When an input without any geographic distances is used, wait times are negligible. If raw genotypic data are supplied, IBDWS returns an HTML table with basic summary statistics, the genetic distance *F*~*ST*~between all population pairs, the genetic similarity statistic M \[[@B6]\] between all population pairs, and Rousset\'s \[[@B7]\] distance *F*~*ST*~/ (1 - *F*~*ST*~) if desired. If geographic distances are included, IBDWS will display a scatter plot of genetic distance vs. geographic distance for all pairs of populations, and overlay the RMA regression line (Figure [1](#F1){ref-type="fig"}). If log-transformation of the data is selected, a pull-down menu below the plot allows the user to choose data sets with one or both axes log-transformed. Clicking on any plot downloads a Postscript version to the user\'s computer. Statistical analyses of the IBD relationship, including Mantel tests, partial Mantel tests (if appropriate) and the RMA regression analyses described above, are summarized in HTML tables as well. A summary of all analyses can downloaded as a static HTML file, or a tab-delimited text file suitable for importation into a spreadsheet application.
Future developments
-------------------
At present, IBDWS requires the user to calculate all pairwise genetic distances between populations for DNA data sets. In future versions, we plan to include algorithms for calculating genetic distances using both sequence and allelic data, including genetic distances based on phylogenetic trees under different models of evolution. When implemented, the user will simply have to supply the raw data and geographic distances before calculations are performed. IBDWS will also be updated with new population genetics analyses as they become available.
Conclusion
==========
In summary, IBDWS is a substantial expansion of the previous PC-based versions of IBD, providing geneticists with a simple interface, multiple statistical calculations, rapid analysis, and high-quality graphics. It will be useful to a significant number of molecular ecology, conservation genetics, population genetics and genomics researchers, and will be updated with newer analyses in the future.
Availability and requirements
=============================
IBDWS is currently hosted at <http://phage.sdsu.edu/~jensen/>. Documentation of the statistical calculations and biological context are available on IBD\'s original website at <http://www.bio.sdsu.edu/pub/andy/IBD.html>. The programming source files have been made available at <http://sourceforge.net/projects/ibdws/> under the GPL (GNU Public License).
List of abbreviations used
==========================
IBD: Isolation by distance.
IBDWS: Isolation by distance web service.
RMA: Reduce major axis (regression).
Authors\' contributions
=======================
JLJ designed and implemented IBDWS. This effort included creating the CGI, debugging and enhancing IBD, implementing postscript graphing libraries and extensive testing of the software. JLJ also wrote the initial draft of the manuscript. AJB wrote the original IBD software, the precursor to IBDWS. AJB also upgraded the code, added graphing algorithms, provided technical advice and assistance to JLJ and contributed to the writing of the manuscript. STK was involved in the early training of JLJ for the project as a mentor in the High Tech High Internship program. STK also guided the development of the project, provided technical advice, and contributed substantially to the writing of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
================
The authors are grateful to Forest Rohwer for hosting IDBWS on his web server. We also thank Pat McNairnie for helping us with web configuration, account management and troubleshooting tips. We thank Varykina Thackray for editorial suggestions on the manuscript. Finally, we thank Jay Vavra, Stacey Praizner, and members of the High Tech High School Biotechnology program for arranging the internship that lead to this research project.
|
PubMed Central
|
2024-06-05T03:55:55.223211
|
2005-3-11
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079815/",
"journal": "BMC Genet. 2005 Mar 11; 6:13",
"authors": [
{
"first": "Jeffrey L",
"last": "Jensen"
},
{
"first": "Andrew J",
"last": "Bohonak"
},
{
"first": "Scott T",
"last": "Kelley"
}
]
}
|
PMC1079816
|
Background
==========
Transposable elements (TEs) constitute a large fraction of plant genomes, and exert critical effects on the formation of the current genomes \[[@B1]\]. With the genome sequences available from a few model species, the differential amplification of TEs helps to explain the C-value paradox in cereal grass species \[[@B2]\]. The TEs have also proven to be a robust source of allelic and subsequent genetic diversity in plants \[[@B3],[@B4]\].
Two classes of transposable elements, classes 1 and 2, have been delineated in plants \[[@B5]\]. Class 1 TEs integrate into host chromosomes via RNA intermediates, using element- encoded reverse-transcriptase, culminating in the production of highly abundant copies in the host genome \[[@B2]\]. The class 1 TEs include the retro-elements, the long terminal repeat (LTR) retrotransposons, the long interspersed nuclear elements (LINEs, also known as non-LTR retrotransposons), and the short interspersed nuclear elements (SINEs). Class 2 elements transpose via DNA intermediates, usually resulting in relatively low copy numbers (usually \<100 copies per genome) \[[@B6]\]. The class 2 elements are also characterized by short terminal inverted repeats (TIRs), and are divided into two groups, autonomous and non-autonomous elements, depending on their transposability. Autonomous elements, such as *Ac*and *Spm*, transpose themselves autonomously, as they harbor all the genes necessary for transposition. Non-autonomous elements, including *Ds*and *dSpm*, only transpose in the presence of autonomous elements in the genome, as they are usually derivatives of autonomous elements, or possess defects in critical regulatory sequences \[[@B7]\]. Another family of class 2 TE, MITEs (miniature inverted-repeat transposable elements) were found in plants \[[@B8],[@B9]\]. Unlike other DNA elements, MITEs are present in very high copy numbers in the genome \[[@B1]\]. However, the mechanisms by which they achieve these high copies have yet to be clearly elucidated \[[@B10]\]. CACTA is another family of transposable elements present at high copy numbers in plants \[[@B11]\]. CACTA was first isolated in maize as a subfamily of *En/Spm*\[[@B12]\], and its name was designated by virtue of its inverted repeats, which terminate in a conserved CACTA motif.
Regardless of its small genome, the *Oryza*species contains all classes of TEs \[[@B13]\]. TEs of both classes have been found to contribute 19.9% of the 910 kb of the rice genome sequence, as evidenced by a high-resolution computer-based survey \[[@B14]\]. While the number of elements in the class 2 TEs outnumbered the class 1 TEs (166 to 22), class 1 TEs constituted a greater sequence contribution (12.2% to 6.6%). Mao *et al*. \[[@B13]\] also noted a variety of TE elements occurring in a survey of 73,000 sequence-tagged-connectors (STC), which can be converted to one STC for every 9 kb across the 430 Mb rice genome, and found that 6848 STCs shared homology with regions of the known TE sequences. A CACTA-like element was identified in rice from a RNA transcript, *Rim2*, in response to the fungal pathogen, *Magnaporthe grisea*\[[@B31]\]. Upon subsequent analysis, the *Rim2*transcript was revealed to belong to the CACTA superfamily, and designated as a *Rim2/Hipa*element \[[@B21]\]. The *Rim2/Hipa*element was estimated to be present several hundred copies or more in the rice genome.
Phenotypic changes due to TE mobilization have provided powerful genetic and molecular tools for the discovery and isolation of genes, using both forward and reverse genetic strategies \[[@B15],[@B16]\]. MITE-transposon display (MITE-TD), a modification of conventional AFLP (amplified fragment length polymorphism) techniques \[[@B17],[@B18]\] using the consensus sequences of the MITE transposons, demonstrated high allelic variations occurring in a segregating maize mapping population \[[@B3],[@B4]\]. The MITE-TD technique proved quite efficient in the construction of recombinant genetic maps. More recently, the MITE-TD technique was approved as an effective method for the evaluation of genetic diversity and species relationships in the *Oryza*species \[[@B19],[@B20]\]. We have modified the MITE-TD, allowing us to utilize the *Rim2/Hipa*CACTA consensus sequences \[[@B21]\] to develop a new set of transposon display (TD) markers in rice. Here, we report the detailed protocols with regard to *Rim2/Hipa*-TD in the *Oryza*species.
Results
=======
Amplification profiles in *O. sativa*
-------------------------------------
As shown in Figure [1](#F1){ref-type="fig"}, *Rim2/Hipa*-TD generated multiple bands with abundant polymorphic profiles among the *O. sativa*accessions. The overall amplification profiles were similar to the AFLP profiles. Depending on primer combinations, the number of amplified fragments ranged from 60 to 80 bands, in a size range from 100 to 700 bp. The number of amplified bands was reduced by increasing the number of selective bases, and the best resolution was obtained with 2 selective bases, as shown in Figure [1](#F1){ref-type="fig"}. In Figure [1](#F1){ref-type="fig"}, we can see the disparity in many of the major bands of the *Indica*and *Japonica*ecotypes, although we were not, at that time, attempting to differentiate between *O. sativa*ecotypes. In order to verify the reproducibility of this technique, *Rim2/Hipa*-TD was conducted on two different thermocyclers, using different brands of Taq DNA polymerases, in 5 different primer combinations. In all of these trials, the amplification profiles were proved to be highly reproducible, as shown in Figure [2](#F2){ref-type="fig"}.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
*Rim2/Hipa*-TD profiles of 6 Indica-type and 5 Japonica-types of *Oryza sativa*species using three different primer combinations (KRMP-AG, KRMP-AT and KRMP-CA).
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Reproducibility of the *Rim2/Hipa*-TD profiles in 5 accessions of *O. sativa*spp. *Japonica*with 2 different thermocyclers and 3 different Taq polymerases. The primer employed was KRMP-GA.
:::
:::
Segregation and chromosomal distribution of the *Rim2/Hipa*-TD markers
----------------------------------------------------------------------
Segregation of the highly polymorphic bands was assessed using F~5~plants derived from an inter-specific hybrid between *O. sativa*Ilpoombyeo/*O.rufipogon*W254 (Fig. [3](#F3){ref-type="fig"}). There were 50 and 45 recordable markers being segregated in the AT and CA primer combinations, respectively. As we were unable to ascertain whether these markers were dominant or co-dominant using the F~5~population, we utilized *Rim2/Hipa*-TD to analyze the F~2~population. Most of the *Rim2/Hipa*-TD markers were found to be segregated as dominant markers, with the exception of a few co-dominantly segregated markers. However, the co-dominant segregating markers constituted less than 1% of the total segregating markers. In Figure [3](#F3){ref-type="fig"}, the segregating markers indicated by stars were odd, since they were found to be present in both parents. We also attempted to confirm the segregation pattern in the F~14~RIL lines (M/G lines) derived from an intra-specific hybrid, during which we also observed one or two odd segregating markers in each primer set amplification (data not shown). Whether or not these odd markers were derived from transposon movement after hybridization remains unknown.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Genetic segregation profile of the *Rim2/Hipa*-TD markers in F5 RIL lines derived from an inter-specific hybrid *O. sativa/O. rufipogon*. Arrowheads represent the segregating markers. The markers designated by stars represent odd non-parental segregating markers.
:::
:::
Figure [4](#F4){ref-type="fig"} displays the chromosomal distribution of the *Rim2/Hipa*-TD markers in chromosome 1 of rice using the F~14~M/G RIL lines, in which the *Rim2/Hipa*-TD markers are distributed proximal half to the centromere in both arms, which was also observed similarly in other chromosomes.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
Distribution of the *Rim2/Hipa*-TD markers in rice chromosome 1. The left is the *Rim2/Hipa*transposon map and the right is the SSR map \[37\]. Note that the *Rim2/Hipa*markers are distributed in the half proximal to the centromere in both arms of the chromosome.
:::
:::
Diversity and distribution of *Rim2/Hipa*-TD markers in *Oryza*genus
--------------------------------------------------------------------
The distribution of the *Rim2/Hipa*elements was evaluated by *Rim2/Hipa*-TD in 15 different *Oryza*species, 8 of which were AA diploids and 7 of which were found to be other genomes, including BB, CC, EE, BBCC, and CCDD (Fig. [5](#F5){ref-type="fig"}). Although the *Rim2/Hipa*-TD generated ample amplification profiles among the AA diploid *Oryza*species, the other species exhibited amplified fragments which, in general, were larger than 400 bp, and these fragments could not be clearly distinguished. These results were consistent with the results obtained with other primer combinations. Among the AA diploid species, amplification profiles were distinct between species, with the notable exceptions of *O. glaberrima*and *O. barthii*. *O. barthii*, however, is believed to be a direct ancestor of *O. glaberrima*\[[@B22]\].
::: {#F5 .fig}
Figure 5
::: {.caption}
######
*Rim2/Hipa*-TD profile of species in the *Oryza*genus. Note the high profile among the AA diploid species, as compared to the high molecular clustered bands in other genome species. The employed primer was KRMP-GA.
:::
:::
Overall profiles were also fairly consistent with the geographical distribution of the *Oryza*species in Asia (*O. sativa*, *O. rufipogon*, *O. nivara*), Africa (*O. glaberrima*, *O. barthii*, *O. longistaminata*), and Australia (*O. meridionalis*). This was confirmed by the phenetic dendrogram \[Fig. [6](#F6){ref-type="fig"}\]. The phenetic relationship between AA diploid *Oryza*species, as shown in Figure [6](#F6){ref-type="fig"}, was similar to those obtained by RFLP \[[@B23]\] and AFLP \[[@B24]\]. The *Oryza*species, except for the AA diploids, could not be included in the phenetic analysis, as the *Rim2/Hipa*-TD marker bands in these species were difficult to match with their corresponding homologous bands in the AA diploids (Fig. [5](#F5){ref-type="fig"}).
::: {#F6 .fig}
Figure 6
::: {.caption}
######
Phenetic dendrogram of the AA diploid *Oryza*species based on the *Rim2/Hipa*-TD markers. The numbers in the horizontal bar at the bottom represent the genetic similarity at the corresponding nodes. The numbers at the nodes represent bootstrap values in each node. The countries in the right column are the origins of each accession.
:::
:::
Discussion
==========
The CACTA transposon superfamily is abundant in most plants. Similar sequence organization has been observed in its terminal regions which are flanked by short TIRs of 10 -- 28 bp in size, which terminate in a CACTA sequence motif \[[@B11],[@B12]\]. Using representational difference analysis (RDA), a CACTA-like transposon, *hipa*, was identified in the rice genome \[[@B25]\], which had previously been characterized as *Rim2*\[[@B21]\]. Although rice has the smallest genome among cereal grass species, various transposon types can be found in the rice genome. In a survey of 910 kb of the rice genomic sequences, class 1 and 2 transposons together constitute approximately 20% of the genome, and CACTA transposons alone contribute 0.5% to these total transposons \[[@B14]\]. Based on cloning and data mining in 230 Mb of the rice genome, the *Rim2/Hippa*CACTA element was estimated to comprise about 600--700 elements of the entire rice genome \[[@B21]\], suggesting that there would be several thousands of the CACTA elements in the entire rice genome. We have utilized the unusually high copy numbers of the *Rim2/Hipa*CACTA transposons and the sequence conservation TIRs of the *Rim2/Hippa*element as genetic markers, using the conventional AFLP protocol, with minor modifications \[[@B17]\].
Conventional AFLP detects restriction site polymorphisms by adaptor ligation to the restricted ends, and selective amplification of restriction fragments using complementary primers to the adaptors. Rather than using two different restriction enzymes, TD employs a single restriction enzyme (usually *Mse*I). Therefore, in addition to the restriction site polymorphisms which flank the transposons, TD also detects polymorphisms of the presence or absence of transposons at specific loci. This constitutes a marked advantage when TDs are utilized for genetic markers, as the integration or excision of transposons can induce allelic diversity in the genes \[[@B3],[@B4]\]. Kanazawa *et al*. \[[@B26]\] also noted that the presence or absence of MITE elements in the *Stowaway*family was significantly associated with speciation in the AA diploid *Oryza*species. The advantages of the AFLP technique over other molecular markers include the reproducibility of the AFLP profile, as well as its ability to detect multiple loci within a PCR amplification. The amplification of the *Rim2/Hipa-*TD was also proved to be highly reproducible, which we confirmed by conducting trials with different thermocyclers and Taq DNA polymerases, and the resolution profile was equivalent to that of AFLP. As primers with two selective bases result in optimum amplification, 16 primer combinations are possible. The average number of amplified fragments in each primer combination is approximately 50 -- 60, when surveying over 800 fragments. Therefore, *Rim2/Hipa*-TD appears to be another effective protocol for the genetic analysis of *Oryza*species, as is shown in Figure [5](#F5){ref-type="fig"}.
The large number of segregation markers detected in the inter- and intra-specific hybrid mapping populations represents a very favorable circumstance for *Rim2/Hipa*-TD, especially with regard to the construction of genetic maps and the tagging of genes of interest. However, its dominant segregation characteristics may limit the use of *Rim2/Hipa*-TD in the F~2~population, although the band intensity enables us to differentiate between homozygotic and heterozygotic genotypes. This intensity-differentiating typing method should be carefully scored, as unequivocal genotyping has proved impossible for some markers, as illustrated by Lee *et al*\'s experiences with maize F~2~mapping using MITE-TD \[[@B27]\]. One notable feature of the *Rim2/Hipa*-TD markers is their distribution of regions proximal to the centromeres in both arms (Fig. [4](#F4){ref-type="fig"}), which was unexpected since the *Stowaway*MITE *Pangrangja*markers were also evenly distributed among 12 linkage groups in rice \[[@B28]\]. In the linkage analysis of *Heartbreaker*MITE markers in maize, the MITE markers were determined to be evenly distributed in all 10 linkage groups \[[@B3]\]. Therefore, the chromosomal distribution of the *Rim2/Hipa*CACTA and MITE transposons may be different in rice. Chromosomal localization of other transposons, such as MITEs and SINEs, is being under investigated with the F~14~M/G RIL lines.
The appearance of non-parental bands is also intriguing. Although we did not, in our analysis, attempt to calculate the frequency of non-parental bands, similar results were reported in a RIL mapping population of maize with *Heartbreaker*MITE-TD markers \[[@B3]\]. In that study, the frequency of non-parental fragments ranged from 0.2 % to 2.5%, depending on the enzyme/primer combination, and the authors explained this non-parental band appearance in terms of the loss of some degree of parental variation over subsequent generations of inbreeding. The mutations in the restriction sites were also proposed to explain the appearance of non-parental bands. Therefore, further analyses of our materials are warranted. The frequency of and mechanisms underlying the appearance of non-parental bands requires determination in future research.
The distribution of the *Rim2/Hipa*CACTA elements among *Oryza*species is particularly prominent in AA genome diploid species, which was corroborated by the results of Southern hybridization using *Rim2/Hipa*CACTA element \[[@B21],[@B25]\]. In the current study, the *Rim2/Hipa*TIR sequence for TD analysis was derived from the sequences of *O. sativa*var. Nipponbare in the NCBI data base. Therefore, poor amplification in species with other genomes may derive from these slight differences in the TIR sequences, resulting in the reduction of primer annealing at the target sites during PCR amplification. Similar results have also been reported by Park *et al*., in an analysis of *Oryza*species using a *Stowaway*MITE *Pangrangja*element \[[@B29]\]. In Southern analysis with the *Pangrangja*probe, more abundant copies of the *Pangrangja*sequences were found among AA diploids than in any other *Oryza*species. Subsequent TD analysis with the *Pangrangja*primer also indicated that the amplified profiles were more prominent in the AA diploid species than in any other species \[[@B20]\]. In the CACTA superfamily, several subfamilies, namely, *Casper*, *Mandrake*, *Isaac*, *Baldwin*, *Jorge*, *Enac*, and *TAT-1*, were isolated and characterized in the *Gramineae*species \[[@B11]\]. As all of them shared the CACTA nucleotide, containing TIR sequences and constituting significant fractions of the cereal genomes, the amplification profile in the current study may represent these CACTA subfamilies.
Conclusion
==========
The *Rim2/Hipa*-TD generated abundant polymorphisms between different *O. sativa*ecotypes. Many segregating markers in inter- and intra-specific hybrids were distributed to regions proximal to the centromeres of the rice chromosomes. The phenetic relationship occurring among AA diploid *Oryza*species, as based on the *Rim2/Hipa*-TD markers, matched well with their geographical distributions, and this was corroborated with results obtained with other marker systems. Therefore, the *Rim2/Hipa*-TD technique will provide another effective protocol for the development of linkage maps and phenetic analyses in rice.
Methods
=======
Plant materials and DNA extraction
----------------------------------
A few representative accessions were analyzed from each of 13 *Oryza*species (Table [1](#T1){ref-type="table"}). The seeds of *Oryza*species, kindly provided by Dr. M.T. Jackson, at the Genetic Resources Center, International Rice Research Institute, Los Banos, Philippines, were germinated in a nursery field. Plant DNA was extracted from young leaves according to the method described by Dellaporta *et al*. \[[@B30]\].
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Name of the species, accessions and genomes of the *Oryza*species tested.
:::
Species Accession Genome
---------------------- ------------------------------------------------ ----------
*O.sativa(Japonica)* Odaebyeo, Jinbubyeo, Jinmibyeo, Hwasungbyeo AA
Ilpumbyeo
*O.sativa(Indica)* China1039, IRRI57313, IRRI60820, IRRI61009 AA
IRRI72
*O.rufipogon* 100647, 100657, 100678, 100692, 100898, 100916 AA
100926, 101173, 101193
*O.nivara* 100593, 100918, 104405, 105701, 100898, 100951 AA
100967, 103834, 103835, 103836
*O.glaberrima* 100982, 100983, 100984, 100980, 101297, 101303 AA
100140, 100149, 100152, 100854
*O.longistaminata* 101202, 101206, 101207, 101210, 101211, 101214 AA
101228, 101230, 101431
*O.barthii* 86447, 100117, 100122, 100223, 100224, 101257 AA
100933, 100936, 100941, 101226
*O.glumaepatula* 100969, 105662, 105670 AA
*O.meridionalis* 101145, 101146, 101147, 101148, 101446, 105289 AA
*O.minuta* IR21 101082, IR39 103865 BBCC
*O.punctata* IR15 100886, IR43 105137 BB, BBCC
*O.officinalis* IR55 105328 CC
*O.alta* IR8 100161, IR47 105222 CCDD
*O.grandiglumis* IR28 101405, IR60 105669 CCDD
*O.latifolia* IR9 100172, IR45 105145 CCDD
*O.australiensis* IR35 103303, IR51 105271 EE
:::
Transposon Display with *Rim2/Hipa*CACTA transposon
---------------------------------------------------
The MITE-AFLP protocols of Casa *et al*. \[[@B3]\] and Park *et al*. \[[@B20]\] were modified for the CACTA transposon display. Using the basic information provided by He *et al*. \[[@B31]\], the CACTA primer and adaptors were designed from consensus sequences obtained from the GenBank database. The primer and sequence information are shown in Table [2](#T2){ref-type="table"}.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Nucleotide sequences of the adaptors and anchors used in *Rim2/Hipa*-TD.
:::
Primer name Sequence
------------------- -----------------------
Adaptor
KRMA-1 GACGATGAGTCCTGAG
KRMA-2 TACTCAGGACTCAT
*MseI*anchors
KRMP-0 GACGATGAGTCCTGAGTAA
KRMP-AA GACGATGAGTCCTGAGTAAAA
KRMP-AC GACGATGAGTCCTGAGTAAAC
KRMP-AG GACGATGAGTCCTGAGTAAAG
KRMP-AT GACGATGAGTCCTGAGTAAAT
KRMP-CA GACGATGAGTCCTGAGTAACA
KRMP-CC GACGATGAGTCCTGAGTAACC
KRMP-CG GACGATGAGTCCTGAGTAACG
KRMP-CT GACGATGAGTCCTGAGTAACT
KRMP-GA GACGATGAGTCCTGAGTAAGA
KRMP-GC GACGATGAGTCCTGAGTAAGC
KRMP-GG GACGATGAGTCCTGAGTAAGG
KRMP-GT GACGATGAGTCCTGAGTAAGT
KRMP-TA GACGATGAGTCCTGAGTAATA
KRMP-TC GACGATGAGTCCTGAGTAATC
KRMP-TG GACGATGAGTCCTGAGTAATG
KRMP-TT GACGATGAGTCCTGAGTAATT
*Rim2/Hipa CACTA*
*Rim2/Hipa*-MAP AGATGGTTTCTCCACCAGTG
:::
The genomic DNA (100 ng) was fully digested with *Mse*I endonuclease, and the adaptor was ligated with the digested DNA in a volume of 20 μl at 22° for 3 hours. Pre-amplification was carried out with the KRMIP-0 primer and with either *Rim2/Hipa*MAP primer. The PCR reaction was carried out with 0.5 μM of each primer, 0.2 mM dNTP, 1.5 mM MgCl~2~, and 1.5 units of Taq Pol (Biotool, Spain) in a total volume of 50 μl. PCR reaction control was as follows: one cycle of 72° for 2 min and 94° for 3 min; 25 cycles of 94° for 30 sec, 56° for 30 sec, 72° for 1 min; and a final extension at 72° for 5 min before completion of the reaction. For selective amplification, the pre-amplified products were diluted by 50-fold. Three μl of the dilution was mixed with 0.5 μM of *Rim2/Hipa*MAP primer, 0.5 μM of one of the *Mse*I selective primers, 0.2 mM dNTP, 1.5 mM MgCl~2~, and 1 unit of Taq Pol (Biotool, Spain) in a total volume of 30 μl. PCR reaction control was as follows one cycle at 94° for 5 min; ten \"touchdown\" cycles of 94° for 30 sec, 64° for 30 sec, and 72° for 1 min with a decrease in annealing temperature to 1° in each cycle; 26 cycles of 94° for 30 sec, 56° for 30 sec, 72° for 1 min; and once at 72° for 5 min to terminate the reaction.
Electrophoresis and fragment detection
--------------------------------------
Five μl of the final reaction was mixed with 10 μl of electrophoresis loading-buffer (98% formamide, 0.02% BPH, 0.02% Xylene C, and 5 mM of NaOH). After being denatured and immediately cooled, two μl of the sample was loaded into 6% denaturing (7.5 M urea) acrylamide-bisacrylamide gel (19:1) in 1× TBE buffer and electrophoresed at 1800 volts and 60 watts for 130 min. Then, the separated fragments were visualized with the silver-staining kit (Promega, USA).
Genetic inheritance of the *Rim2/Hipa*-TD markers
-------------------------------------------------
Genetic inheritance of the *Rim2/Hipa*-TD markers was analyzed using F~2~and F~5~populations derived from a cross between *O. sativa*var. *Ilpoombyeo*(Japonica type variety) and *O. rufipogon*W259. Chromosomal distributions of the *Rim2/Hipa*-TD markers were analyzed with a RIL population (M/G RILs) derived from an intra-specific cross between *O. sativa*var. *Milyang*(Tongil type, Indica/Japonica) and *O. sativa*var. *Gihobyeo*(Japonica variety) since a SSR framemap had already been developed with the M/G RILs \[[@B32]\]. Linkage analysis was performed using Mapmaker version 3.0 \[[@B33]\].
Phenetic cluster analysis
-------------------------
Presence or absence of the marker bands were recorded as a binary code, 1 or 0, in each accession. Then, a phenetic dendrogram was constructed on the basis of Nei and Li\'s algorithm \[[@B34]\] using the arithmetic average option in the NTSYS-pc program \[[@B35]\]. The bootstrapping was done using the \'WINBOOT\' program developed at IRRI \[[@B36]\].
List of abbreviations
=====================
TE ; transposable elements, AFLP; amplified fragments length polymorphism, TD; transposon display, MITE; miniature-inverted transposable element, PCR; polymerase chain reaction
Authors\' contributions
=======================
SJK conducted most of the TD analysis, designed the experiment, and prepared the illustrations for the manuscript. JHK performed the TD analysis. KCP mined the TE sequences in the GenBank database and designed the primers used in the *Rim2/Hipa*-TD analysis. JKL analyzed the obtained data and participated in the discussion for preparing the manuscript. NSK was the principal investigator of the project and prepared the manuscript
Acknowledgements
================
This work was supported by a grant to NSK from the Ministry of Science and Technology, Republic of Korea, through the Crop Functional Genomics Center (Project Number CG3122). Thanks are also extended to Drs. Thomas Bureau and Myung-Hyun Wang for critically reading the manuscript.
|
PubMed Central
|
2024-06-05T03:55:55.224431
|
2005-3-14
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079816/",
"journal": "BMC Genet. 2005 Mar 14; 6:15",
"authors": [
{
"first": "Soon-Jae",
"last": "Kwon"
},
{
"first": "Kyong-Chul",
"last": "Park"
},
{
"first": "Jin-Hong",
"last": "Kim"
},
{
"first": "Ju Kyong",
"last": "Lee"
},
{
"first": "Nam-Soo",
"last": "Kim"
}
]
}
|
PMC1079817
|
Background
==========
Sex identification using genomic DNA extracted from meat, blood, hair or embryo biopsies is sometimes an important analytical tool in forensic science or in routine genotyping. In most mammals, the male is identified by amplifying the SRY gene (sex-determining region Y) which is a Y chromosome-specific sequence. When the results are negative from amplification of only the SRY gene, it cannot be assumed that the individual is female or that there was a mistake in the experimental process. A gene present in both, males and females, should be amplified in the same tube as a positive control. In sex identification of ursides by PCR, primers that amplify the SRY gene together with the mitochondrial DNA control region or the ZFX/Y region have been used \[[@B1],[@B2]\]. Furthermore, recent reports have described the use of low-stringency PCR or the detection of X/Y specific restriction fragment polymorphisms (RFLP) in sheep and goats \[[@B3],[@B4]\]. However, RFLP analysis requires an additional reaction step which might increase the risk of contamination and misdiagnosis. The amelogenin (AMEL) gene, which exists on both X and Y chromosomes, has been used to determine the sex in cattle \[[@B5]\] and humans \[[@B6]\]. The use of this gene has made the sex determination much less complicated, since only one pair of primers is required to amplify the different size fragments of the AMEL genes.
In this study we used the primers established in cattle to determine their suitability for other species, i.e. sheep and European red deer. The amplicons were isolated and sequenced, and showed a length polymorphism characteristic for the X and Y chromosome in both species.
Results
=======
The nucleotide sequences of the PCR products of the sheep and the red deer amelogenin gene are shown in Figure [1](#F1){ref-type="fig"}. Both male specific products harbour different basepair deletions (ins/del). Besides a few minor single nucleotide ins/dels, the most prominent ins/del is between positions 63 and 107 (45 bp) and positions 127 and 136 (10 bp) according to Fig. [1](#F1){ref-type="fig"}. This major ins/del leads to a shorter amplicon that can easily be detected in an agarose gel. The sheep and red deer male specific product was deposited with the EMBL/Genebank database under accession numbers AY453392 and AY453391, respectively. The comparison of the sheep and red deer with the cattle amelogenin gene DNA sequences showed a 96 % and 97 % homology for the AMELX gene and a 86 % and 90 % homology for the AMELY gene, respectively.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
DNA sequence comparison of the X and Y amelogenin gene fragments of sheep and European red deer. (.) ins/del
:::
:::
As expected, a single band for females (EMBL/Genebank accession number female sheep AY 452664 and female red deer AY 452665) and two bands for the males were detected when analysing different animal samples (Fig. [2](#F2){ref-type="fig"}). Although an additional non-specific band was observed in male sheep and red deer, there was no problem to determine the sex, based on the different banding patterns.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Detail of an ethidiume bromide stained agarose gel showing amplicons of the amelogenin gene fragments of sheep and European red deer. Lane 1--2 samples from male red deer, lane 3--4 samples from male sheep, lane 5 sample from female sheep and lane 6--7 samples from female red deer. The star 1 indicates the AMELX band, the star 2 indicates a non specific band in male sheep and male red deer and the star 3 indicates the AMELY band.
:::
:::
Discussion
==========
In this study we have established a simple and accurate method for determining the sex of sheep and red deer by using a pair of primers for the AMEL gene. The amelogenin gene encodes an important protein in the developing mammalian tooth enamel matrix \[[@B7],[@B8]\] that has been conserved during the evolution of vertebrates. The bovine amelogenin gene \[[@B9]\] is located on the X and Y chromosomes, and differences in their length are observed between the X and Y specific gene. In other species analysed so far, e.g. mice \[[@B10]\], the gene is located only on the X chromosome. Our results show that the amelogenin gene of sheep and red deer is located on both sex chromosomes and that there are two diagnostic ins/del of 55 bp together within the Y-specific gene in the region amplified.
By using this method in combination with routine genotyping more information about a material under investigation can be obtained. In addition, the amplification of the AMEL gene can also be used as an internal control. Furthermore, the contamination with human DNA is sometimes a problem during the laboratory analysis. However, the amplification of the human AMEL gene with the same primers resulted in a different and clearly distinguishable banding pattern. The conserved status of the amelogenin gene among vertebrates indicates the possibility to use the test in other wild mammal species as well.
Conclusion
==========
In conclusion our findings show that the PCR assay based on the AMEL gene is reliable for sex identification in sheep and European red deer. The advantage of this assay is that neither additional control amplicons with a second locus-specific autosomal primer pair nor restriction endonuclease steps are necessary for sex determination and control of the PCR reaction.
Methods
=======
Samples were taken from sheep (*Ovis aries*) and European red deer (*Cervus elaphus*). DNA was extracted from different tissue samples using QIAamp^®^Tissue Kit (QIAGEN GmbH, Hilden, Germany) according to the manufacturers\' handbook. Isolated DNA was diluted in 50 μl HPLC-H~2~O and used for further analyses.
We used one set of primers (SE47, SE48) for amplifying the sheep and red deer amelogenin gene. The DNA sequences of the primers were 5\'-cagccaaacctccctctgc-3\' (SE47) and 5\'-cccgcttggtcttgtctgttgc-3\' (SE48) as described by \[[@B5]\].
Amplifications were performed in a final volume of 20 μl in 10 × PCR buffer (15 mM MgCl~2~, pH 8.3) and Q-solution (QIAGEN GmbH, Hilden, Germany), 100 μM for each dNTP, with 1 M Taq DNA Polymerase and 10 pmol of each primer. Four microlitres of the DNA-extract were added to the PCR mix. The amplification was carried out with initial denaturation at 95°C for 10 min, followed by 35 cycles of one denaturation step at 94°C for 50 sec, primer annealing at 56°C for 50 sec and primer extension at 72°C for 50 sec in a Hybaid Omnigene thermocycler (MWG Biotech, Ebersberg, Germany). A final extension step was not included. PCR-products were purified using the QIAEX II Gel Extraction Kit (QIAGEN GmbH, Hilden, Germany) according to the manufacturers\' instructions. Sequencing was performed using ABI-Prism™ BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Weiterstadt, Germany) in a 10 μl volume containing 2 μl purified PCR-product and 5 pmol of primer. Sequencing reactions underwent 27 cycles of 30 sec at 94°C, 30 sec at 55°C and 3 min at 60°C in a Techne Gene E Thermocycler (Burkhardtstorf, Germany). The dye terminators were removed by sephadex-G45 column purification (Millipore). Sequencing reactions were electrophoresed for 2 h on an ABI Prism^®^3100 Genetic Analyzer (Applied Biosystems, Weiterstadt, Germany) according to the manufacturers\' instructions.
Authors\' contributions
=======================
IP performed the DNA extractions, PCR analysis and the DNA sequencing.
BB was responsible for funding, supervision of the research project, manuscript writing and editing as well as scientific correspondence.
Acknowledgements
================
This work was supported by a grant of the Erxleben Research & Innovation Council ERIC to B. Brenig (ERIC-BR1959-2002--03).
|
PubMed Central
|
2024-06-05T03:55:55.226765
|
2005-3-16
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079817/",
"journal": "BMC Genet. 2005 Mar 16; 6:16",
"authors": [
{
"first": "I",
"last": "Pfeiffer"
},
{
"first": "B",
"last": "Brenig"
}
]
}
|
PMC1079818
|
Background
==========
From the selfish DNA perspective \[[@B1],[@B2]\], gonads are fundamentally important organs of an organism. During the first meiotic division of gametogenesis, crossing-over enhances the re-assortment of information carried in parental DNA molecules and virtually immortal genetic information is then transferred to next generations of mortal individuals via the final products of gametogenesis, spermatozoa and eggs. Moreover, testes and ovaries are the only niches where the paternal and maternal DNA interacts with a different environment. The dissimilar gonadal environment enables sex-dependent epigenetic modifications of paternal and maternal DNA such as reactivation of the X chromosome in female germ cells \[[@B3],[@B4]\], inactivation of a single X chromosome in pachytene spermatocytes \[[@B5]-[@B7]\] or differential establishment of imprinting marks on paternally or maternally imprinted genes \[[@B8],[@B9]\]. Spermatogenesis also serves as an important checkpoint filtering out many *de novo*occurring gene mutations \[[@B10],[@B11]\] and chromosomal rearrangements \[[@B12],[@B13]\] by making their carriers sterile. A special form of meiotic checkpoint is represented by hybrid sterility, which facilitates creation of new species. Obeying the Haldane\'s rule, hybrid sterility preferentially affects gametogenesis in testis in species with heterogametic (XY) sex \[[@B13]-[@B15]\]. Molecular analyses of these phenomena are hindered by the fact that testis is a complex organ with many types of intimately intermingled somatic and germline cells. Moreover, the spermatogenic differentiation is almost impossible to achieve *ex vivo*, in a cell culture system. The main cell types can be fractionated, via gravity sedimentation, centrifugal elutriation or fluorescence activated cell sorting, but the time required can be fairly long to exclude possible artificial changes of mRNA levels.
In the present work we used Serial Analysis of Gene Expression (SAGE) \[[@B16]\] to characterize the transcriptome of mouse total testis. We created a catalogue of genes expressed in the adult mouse testis of the C57BL/6J (abbreviated here B6) inbred strain. The B6 inbred strain has been chosen because its genome has been recently sequenced \[[@B17]\] and since it has been selected as a recipient strain for creation of two sets of Chromosome Substitution Strains, C57BL/6J-Chr\#^A/J^\[[@B18]\] and C57BL/6J -Chr\#^PWD/Ph^\[Gregorova S, Forejt J et al., in progress\]. Except for the characterization of the total testis transcriptome, we compared our data with the publicly available SAGE library from adult testis somatic cells \[[@B19]\] and other SAGE libraries constructed from normal mouse tissues. Furthermore, we were interested in the organization of testicular genes in the mouse genome and we present here a detailed bioinformatic analysis of the distribution of testicular genes between the X chromosome and autosomes, and the positional clustering of genes with preferential expression in testis.
Results
=======
Characterization of the SAGE libraries of B6 mouse testis
---------------------------------------------------------
We have constructed two high-quality SAGE libraries, TT 1 and TT 2, from the total mouse testis of adult B6 males (Table [1](#T1){ref-type="table"}). The libraries contain 24 975 (TT 1) and 51 879 (TT 2) tags corresponding to 10 516 and 18 848 unique tags, respectively. The tags with abundance \> 1 comprise 17 244 (69 %) and 38 457 (74 %) of the total tag mass but only 2 785 (26.5 %) and 5 426 (29 %) of the unique tags, respectively. The high average number of tags per clone (\> 30) and low contamination with linker-derived tags (\< 1 %) and duplicated ditags (\~1%) indicate that the SAGE libraries are of high quality. Both total testis SAGE libraries provided similar gene expression profiles (R^2^= 0.84 for all unique tags, Pearson correlation), which suggests a good reproducibility of SAGE data. However, a certain variation was observed in the tag abundances when 24 529 unique tags found in both total testis SAGE libraries were compared by Monte Carlo simulations. Three hundred thirteen tags exhibited significant differences in their frequency between TT 1 and TT 2 libraries at p \< 0.05 (89 tags at p \< 0.01) representing non-hereditary variations in transcription profiles and variations introduced by the experimental process. The fold factor value (defined as the ratio of normalized tag counts in TT 2 to TT 1 libraries, with ratios \< 1 converted to reciprocal negative values) for 93.5 % of the compared tags ranged between -2.2 and 2.2 (for 99% of the tags between -5 and 5). Dot plot comparison and fold factor distribution graphs (Fig. [1A,C](#F1){ref-type="fig"}) depict the similarity of both total testis libraries. Despite this variation, the SAGE method produced reproducible gene expression profiles and the libraries could be combined into the total testis SAGE library (referred to hereafter as TT 1+2) with the total of 76 854 tags and 24 529 unique tags. The raw data from the total testis SAGE libraries are deposited in the GEO repository \[[@B20]\] under accession numbers GSM34767 (TT 1) and GSM34768 (TT 2). The set of tags with abundance \> 1 in TT 1+2 SAGE library with reliable tag identification is listed in [Additional file 1](#S1){ref-type="supplementary-material"}. The testis SAGE libraries are also freely available for interactive exploration and analysis in the Mouse SAGE Site database \[[@B21]\].
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Parameters of constructed SAGE libraries from B6 mouse total testis
:::
SAGE library Total testis 1 (TT 1) Total testis 2 (TT 2)
------------------------ ----------------------- -----------------------
Sequenced clones 811 1 510
Total tags\* 24 975 51 879
Unique transcripts 10 516 18 848
Single copy tags 7 731 13 422
Quality parameters
Average tags per clone 30.8 34.4
Duplicated ditags 157 (1.2 %) 276 (1.0 %)
Linker derived tags 147 (0.6 %) 223 (0.4 %)
\* excluding tags from duplicated ditags and linker-derived tags (i.e. two linker tags TCCCTATTAA, TCCCCGTACA and all possible 1-bp linker tag variations)
:::
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Comparison of mouse testis SAGE libraries represented by dot plots and fold factor distribution graphs.**Comparison of tag counts between two total testis libraries (A), and between the combined total testis library and adult testis somatic cells library (B). Tags with significant p-chance are depicted in blue (0.01 \< p \< 0.05) and red (p \< 0.01). Tags missing in one of the libraries are plotted at -0.5 coordinates. Point size is proportional to the number of represented tags. Distribution of the fold factor between two total testis libraries (C), and between the combined total testis library and adult testis somatic cells library (D). Fold factor is the ratio of normalized tag abundances in two SAGE libraries with ratios \< 1 converted to reciprocal negatives. For tags missing in one library, normalized tag count of single copy tags was assumed. Abbreviations: TT 1 = total testis library 1; TT 2 = total testis library 2; TT 1+2 = combined total testis libraries; ATSC = adult testis somatic cells library.
:::
:::
Tag-to-gene identification in the B6 testis transcriptome
---------------------------------------------------------
Tag-to-gene identification in the TT 1+2 SAGE library was evaluated using three different criteria applied to the SAGEmap database. The first was the most commonly used SAGEmap reliable mapping \[[@B22],[@B23]\]. The second was a modified approach based on the SAGEmap full mapping file and implemented in the Mouse SAGE Site database \[[@B21]\]. In this approach, the tag-to-gene associations were considered reliable if supported by tags extracted from at least one mRNA sequence (from RefSeq, Mammalian Gene Collection or GenBank) or at least 3 ESTs with a poly(A) signal or at least 8 ESTs with no poly(A) signal \[[@B24]\]. The third approach (referred to here as RNA evidence mapping) was also based on the SAGEmap full mapping file. Tag-to-gene associations were considered reliable if supported by tags extracted from at least one mRNA sequence. The 7 481 tags with tag count \> 1 in the TT 1+2 library were subjected to the SAGEmap reliable mapping that could identify 92.6 % tags to UniGene clusters (54.3 % to single and 38.3% to multiple genes; Table [2](#T2){ref-type="table"}). When a more restricted reliable mapping from the Mouse SAGE Site was used, only 63% tags were identified to UniGene clusters (47.5% to single and 15.5 % to multiple genes) and about 29.6 % tags had unreliable identification to one or more UniGene clusters. Based only on the tags extracted from mRNA sequences, the RNA evidence mapping identified 51.3 % tags to UniGene clusters (45 % to single and 6.3 % to multiple genes) leaving 41.3% tags with unreliable identification. Using any of the tag identification methods, 7.4 % tags could not be identified to UniGene clusters and may be associated with novel genes. Further in this work, we used Mouse SAGE Site or RNA evidence mapping appropriately for a particular analysis (as indicated in Methods and Additional files).
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Identification of tags in the combined total testis SAGE library (TT 1+2) The tags matching the mitochondrial genome were omitted in this summary. Tags in group \"Unreliable matches\" (\*) are considered not reliable according to Mouse SAGE Site and RNA evidence mappings, because they are not supported by the required number of mRNA and EST sequences. These tags are, however, included in reliable single/multiple match groups in the SAGEmap reliable mapping, which results in a highly increased number of reliable multiple matches and a slightly increased number of reliable single matches.
:::
NCBI SAGEmap reliable mapping Mouse SAGE Site reliable mapping RNA evidence mapping
----------------------------- ------------------------------- ---------------------------------- ---------------------- ------- ------- -------
Reliable single match 4 061 54.3 3 553 47.5 3 367 45.0
Reliable multiple matches 2 865 38.3 1 157 15.5 472 6.3
Unreliable match(es)\* \- \- 2 216 29.6 3 087 41.3
No match 555 7.4 555 7.4 555 7.4
Total tags (tag count \> 1) 7 481 100.0 7 481 100.0 7 481 100.0
:::
Functional categories of genes expressed in total testis
--------------------------------------------------------
We associated genes and their corresponding tag counts to functional categories from the biological process ontology of GO database \[[@B25],[@B26]\] (Fig. [2](#F2){ref-type="fig"}). In the total testis transcriptome, we observed more than 1000 genes involved in metabolism, particularly in the protein metabolism (protein modification, protein targeting) and nucleic acid metabolism (chromatin assembly and modification, DNA replication, DNA repair, RNA processing, RNA modification). As expected, the genes associated with spermatogenesis (e.g., protamine 1 and 2, transition proteins 1 and 2), chromosome organization, cell cycle and cell differentiation were highly expressed. Notably represented gene functions also included transport (e.g., diazepam binding inhibitor-like 5, proteasome 26S subunit, ribosomal protein L23), signal transduction (e.g., calmodulin 1 and 2, sperm autoantigenic protein 17, A kinase (PRKA) anchor protein 3, PDZ domain containing 1, WD repeat domain 12), cytoskeleton organization (e.g., t-complex testis expressed1, t-complex-associated testis expressed 3, tubulin alpha7/alpha 3, tubulin alpha 6, thymosin beta 10) and apoptosis (e.g., Bcl2-associated athanogene 1, Bcl2-like 14, programmed cell death 5, tumor protein translationally-controlled 1). From the mitochondrial genome, ATP synthase 6, cytochrome c oxidase I and III were the most highly expressed genes.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Classification of genes expressed in total testis according to the biological process ontology of the GO database.**Bar graphs represent the sum of tag abundances corresponding to genes associated with a particular GO term. Only selected GO terms with the sum of tag abundances \> 500 are displayed. The number of genes associated with each GO term is indicated inside the bars.
:::
:::
Comparing the transcriptomes of total testis and adult testis somatic cells
---------------------------------------------------------------------------
The mouse testis is composed of two main cell types with principally different origin and functions, the germ cells that differentiate from spermatogonia to mature spermatozoa and the somatic cells that carry out all supportive functions to make the spermatogenesis and reproduction possible. Seminiferous tubules of the adult testis consist of approximately 88% germ cells and 12% somatic cells including myoid and Sertoli cells \[[@B27]\]. We compared our total testis SAGE library (TT 1+2) with a SAGE library constructed from the somatic cells of adult testis (GEO, accession GSM5435). This library was created from testes largely devoid of germ cells 60 days after busulphan treatment \[[@B19]\]. The SAGE library sizes are similar for TT 1+2 and the adult testis somatic cells (abbreviated here ATSC) comprising 76 854 and 81 478 tags, respectively. The number of unique tags (24 529 and 22 809) as well as the proportions of tags with abundance \> 1 to the total tag mass (77.8% and 81.1 %) and to the number of unique tags (30.5 % and 32.6 %) are also comparable (Table [3](#T3){ref-type="table"}). As anticipated, comparison of TT 1+2 and ATSC SAGE libraries using Monte Carlo simulations revealed extensive differences in gene expression between total testis and somatic cells of adult testis. Out of the 42 239 unique tags in TT 1+2 and ATSC libraries, the simulations detected significantly different tag abundances in 3 258 tags at p \< 0.05. Concerning the fold factor, 83 % of the compared tags stretch in the range between -2.2 and 2.2 (92.5% tags between -5 and 5). At the extreme ends, 563 tags reach \> 10-fold increase in tag counts in the ATSC library (fold factor \> 10) and 672 tags reach \> 10-fold increase in the TT 1+2 library (fold factor \<-10) (see [Additional file 2](#S2){ref-type="supplementary-material"}). Dot plot comparison and fold factor distribution graphs for TT 1+2 and ATSC transcriptomes illustrate their dramatic dissimilarities (Fig. [1B,D](#F1){ref-type="fig"}).
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Parameters of the SAGE libraries constructed from total testis and somatic cells of adult testis
:::
SAGE library Total testis Adult testis somatic cells
-------------------------------------------- -------------- ----------------------------
Total tags 76 854 81 478
Unique tags 24 529 22 809
Unique tags with count \> 1 7 481 7 435
Proportions of unique tags with count \> 1
\% of total tags 77.8 81.1
\% of unique tags 30.5 32.6
:::
Genes with predominant expression in the germinal or somatic component of testis
--------------------------------------------------------------------------------
To sort out subsets of genes with predominant expression in either germinal or somatic cells of testis we applied tentative criteria to account for the presence of somatic cells in TT 1+2 and for residues of germ cells in ATSC. Predominant expression of a gene was considered if the corresponding tag was significantly more frequent in one of the libraries (p \< 0.05, Monte Carlo simulations) and exhibited at least fivefold enrichment of tag counts (fold factor \<-5 or \> 5). According to this criterion a set of 829 genes is expressed predominantly in germ cells and 944 genes are expressed mainly from the somatic part of the testis (see [Additional file 3](#S3){ref-type="supplementary-material"}). Moreover, we identified 12 tags corresponding to 8 genes encoded in the mitochondrial genome (1 gene with increased tag counts in TT 1+2 and 6 genes with increased tag count in ATSC). A gene coding for cytochrome c oxidase III (*mt-Co3*) displayed two tags separated by 87 bp in *mt-Co3*gene mRNA. One isoform was predominantly present in the ATSC library and the other was observed exclusively in the TT 1+2 library. Substantial over-expression of mitochondrial cytochrome c oxidase complexes I, II, III and NADH dehydrogenase 3 and 4 was noted in testicular somatic cells (see [Additional file 3](#S3){ref-type="supplementary-material"}).
Exploring the dissimilarity of testis transcriptomes and transcriptomes of other mouse tissues
----------------------------------------------------------------------------------------------
We examined the similarity of B6 testis transcriptomes to other available mouse SAGE transcriptomes created from normal and diseased bulk tissues by hierarchical clustering. Thirty-two SAGE libraries containing 190 871 unique tags (including single copy tags) were used as input in this analysis (see [Additional file 4](#S4){ref-type="supplementary-material"}). We computed pair-wise library distances based on differences between normalized tag counts \[[@B28]\] and used the average agglomerative method for hierarchical clustering due to the highest cophenetic correlation (0.936). In the dendrogram of dissimilarities the two total testis SAGE libraries, TT 1 and TT 2, cluster together in contrast to the library from somatic cells of the adult testis (Fig. [3](#F3){ref-type="fig"}). The ATSC library is located separately and close to the libraries created from heart, liver and kidney in accord with the somatic origin of all these tissues. Interestingly, another SAGE library created from somatic cells of the fetal testis did not cluster with the ATSC library, but was placed close to the libraries from developing limbs, juvenile retina and whole brains. Another cluster consists of the six libraries generated from the whole adult kidneys. Several specialized brain tissues form a cluster with a brain tumor tissue (cerebellum, hippocampus, hypothalamus, medulloblastoma). An additional small cluster groups three libraries created from whole brain samples (normal male, trisomic Ts65Dn male and normal female).
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Dissimilarities of mouse SAGE libraries illustrated by a dendrogram.**Thirty-two SAGE libraries constructed from bulk tissues containing 190 871 unique tags (including single copy tags) were selected (see [Additional file 4](#S4){ref-type="supplementary-material"}). Pairwise library distances based on differences between normalized tag counts were computed according to \[28\]. The average agglomeration method was used in hierarchical clustering due to the highest cophenetic correlation (0.936) between observed and predicted distances resulting from the dendrogram. The number of tags in each SAGE library is indicated.
:::
:::
Nonrandom representation of testis-expressed genes on the X chromosome
----------------------------------------------------------------------
Previous works have shown a significant enrichment of prostate- and spermatogonia-specific genes on the X chromosome when compared to autosomes \[[@B29],[@B30]\]. We asked what proportion of testis-expressed genes maps to the X chromosome and compared it with the proportion of X-linked genes expressed in somatic (non-testis) tissues. Furthermore, we examined whether the proportion of testis-specific genes on the X chromosome differs from the proportion of X-linked tissue-specific genes in somatic tissues.
Out of the 14 222 genes expressed in SAGE libraries from total testis, adult testis somatic cells and 7 somatic tissues (brain, eye, heart, liver, kidney, limbs and adipose tissue) (see [Additional file 4](#S4){ref-type="supplementary-material"}) we considered only genes identified by corresponding tag count \> 1. The proportion of genes expressed from the X chromosome in a pool of 7 somatic tissues was 3.1 % (374 of 11 903 genes). Although the proportions of X-linked genes in somatic tissues were uneven, there were no significant differences among the tissues (3.2 % in brain, 2.7 % in limbs and eye, 2.6 % in liver, 2.5 % in kidney and adipose tissue, 2.4 % in heart; p \> 0.05, Chi-square test for brain vs. heart). In testicular somatic cells, we observed 3.2% X-linked genes (133 of 4 216 genes), while in total testis only 1.4 % genes (48 of 3 338 genes) were expressed from the X chromosome (p \< 10^-6^, Chi-square test). We can conclude that the number of expressed X-linked genes is underrepresented in the transcriptome of total testis.
The same set of 14 222 genes was examined for the distribution of tissue-specific genes on autosomes and the X chromosome. We compared the genes specific for either total testis (Table [4, a](#T4){ref-type="table"}) or adult testis somatic cells (Table [4, b](#T4){ref-type="table"}) in conjunction with somatic (non-testis) tissue-specific genes. A gene was considered to be tissue-specific if it was expressed only in one tissue type (total testis or adult testis somatic cells, brain, eye, heart, liver, kidney, limbs and adipose tissue). Moreover, the corresponding tag count \> 1 was required to guarantee that the gene is truly expressed. The tissue-specific genes were assigned to chromosomes according to the LocusLink database and the significance of their chromosomal distribution was evaluated by permutations (see Methods) and confirmed by Fisher\'s exact test (Table [4](#T4){ref-type="table"}). Out of the 395 genes specific for total testis 3.5% mapped to the X chromosome (see [Additional file 5](#S5){ref-type="supplementary-material"}). Essentially the same proportion of X-linked genes was found for genes specific for 7 somatic (non-testis) tissues. In testicular somatic cells, we detected only 81 tissue-specific genes, but 13.6% were X-linked (see [Additional file 5](#S5){ref-type="supplementary-material"}). This is a 3.2-fold increase in the proportion of testis somatic cell-specific genes on the X chromosome and represents their significant enrichment (p = 0.0024, two tailed, 100 000 permutations) in comparison to the genes specific for other tissues. All the X-linked testis-specific genes were subjected to BLAST against the whole X chromosome, which revealed no duplicated genes. The results from the permutation analysis indicate a significantly increased amount of testis-specific genes on the X chromosome in somatic cells of the testis when compared to autosomal testis-specific genes. The genes specific for 7 somatic tissues did not show a significant preference for the X chromosome. The list of X-linked genes expressed in total testis and testicular somatic cells with indicated testis-specific genes is available in [Additional file 6](#S6){ref-type="supplementary-material"}.
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Distribution of testis-specific genes on autosomes and the X chromosome The total of 14 222 LocusLink genes were identified in total testis, adult testis somatic cells and non-testis tissue SAGE libraries (see [Additional file 4](#S4){ref-type="supplementary-material"}) using RNA evidence mapping (tags matching multiple LocusLink genes were discarded). The genes identified by total tag count = 1 were then excluded from analysis. The genes expressed only in one tissue type (total testis, adult testis somatic cells, brain, eye, heart, liver, kidney, limbs and adipose tissue) were considered to be tissue-specific genes. Chromosomal distribution of genes specific for total testis (a) and testis somatic cells (b) in comparison to the non-testis tissue-specific genes was evaluated. The significance was tested by permutations (100 000 random shufflings of the chromosomes while keeping the sum of genes on autosomes and the X chromosome fixed) and confirmed by Fisher\'s exact test. Abbreviations: total t. = total testis; t. somatic = testicular somatic cells; other = non-testis tissues; ChrA = autosomes; ChrX = X chromosome.
:::
a\) Total testis: 395 genes specific for the combined total testis SAGE library (TT 1+2) Other tissues: 877 genes specific for one tissue type in the pool of other SAGE libraries
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ---------------------- ---------------------------------- ------------------------- ------------------------------- ----------- ------- -------------------
Chrom Observed gene counts Gene counts in randomized genome \% observed gene counts Ratio of observed proportions
total t. other total t. other total t. other total t./other
ChrA 381 836 378 839 96.5 95.3 1.0
ChrX 14 41 17 38 3.5 4.7 0.7
Permutations yielding \< = observed gene counts in total t. on ChrX 22 395
Permutations, p-value (two tailed) 0.4479
Fisher\'s exact, p-value (two tailed) 0.4563
Confidence interval (0.95) 0.70 -- 2.68
b\) Testis somatic cells: 81 genes specific for the adult testis somatic cells SAGE library (ATSC) Other tissues: 924 genes specific for one tissue type in the pool of other SAGE libraries
Chrom Observed gene counts Gene counts in randomized genome \% observed gene counts Ratio of observed proportions
t.somatic other t.somatic other t.somatic other t\. somatic/other
ChrA 70 885 77 878 86.4 95.8 0.9
ChrX 11 39 4 46 13.6 4.2 3.2
Permutations yielding \> = observed gene counts in t. somatic on ChrX 121
Permutations, p-value (two tailed) 0.0024
Fisher\'s exact, p-value (two tailed) 0.0013
Confidence interval (0.95) 0.13 -- 0.64
:::
Chromosomal clustering of genes with preferential expression in testis
----------------------------------------------------------------------
Based on the data from testis and other publicly available SAGE libraries (see [Additional file 4](#S4){ref-type="supplementary-material"}) we identified genes with preferential expression in testis by Preferential Expression Measure (PEM) \[[@B31]\]. PEM score controls for the genes that are highly expressed in many tissues (housekeeping genes) and reports positive values for over-expressed genes and negative values for under-expressed genes in a given tissue. Large positive PEM scores for a gene in a particular tissue indicate that the gene is unusually highly expressed in that tissue, relative to its expression in other tissues \[[@B31]\]. We considered a gene to be preferentially expressed if the PEM score reached at least 50 % of the maximum PEM value encountered in that tissue. Using this criterion, we scored expression of genes in total testis or testicular somatic cells in conjunction with their expression in 7 other tissues (brain, eye, heart, liver, kidney, limbs and adipose tissue).
Further we analyzed the genome organization of genes preferentially expressed in testis. We evaluated the expression of 14 222 genes among the studied tissues and for 12 331 genes we were able to assign a genomic position according to the NCBI mouse genome assembly (build 32, mapping 19 684 known LocusLink genes). The genomic position was resolved for 5 252 and 5 843 genes expressed in total testis and testicular somatic cells, respectively, including 1 438 (27.4%) and 1 197 (20.5%) preferentially expressed genes, respectively (see [Additional file 7](#S7){ref-type="supplementary-material"}). To evaluate the gene order of preferentially expressed genes in testis and to eliminate the effect of tandem duplications we purged the whole mouse genome of tandemly duplicated genes (see Methods). The tandem duplicate-free genome resulted in total of 16 858 LocusLink genes and preserved 1 300 and 1 050 genes preferentially expressed in total testis and testicular somatic cells, respectively. Using a search with a sliding window (see Methods) we localized chromosomal clusters containing at least three adjacent preferentially expressed genes (tight clusters). Similarly, we searched for clusters with at least three preferentially expressed genes among the six adjacent genes (loose clusters) to include genes that could be preferentially expressed but did not pass the above criterion for preferential expression or their expression was not detected by SAGE. By definition, the tight clusters form a subset of the loose clusters. The chromosomal distribution of clusters with preferentially expressed genes in testis is illustrated in Figure [4](#F4){ref-type="fig"}. We observed 44 and 36 genes preferentially expressed in total testis and testicular somatic cells located in 13 and 11 tight clusters, respectively. Two hundred and thirty and 120 genes preferentially expressed in total testis and testicular somatic cells resided in 66 and 37 loose clusters, respectively (Table [5](#T5){ref-type="table"}; [Additional file 8](#S8){ref-type="supplementary-material"}). Two of the tight clusters and eight of the loose clusters shared preferentially expressed genes between total testis and testicular somatic cells. Statistical analysis revealed that the observed number of preferentially expressed genes located in tight clusters was 2.0-fold and 3.1-fold higher for total testis and testicular somatic cells, respectively, than the average number of preferentially expressed genes located in clusters in randomized genomes (p = 0.0074 and p = 0.0005, one tailed, 100 000 permutations). Although only slighly higher (1.4- and 1.3-fold) than the average in randomized genomes, the observed number of preferentially expressed genes in testis located in loose clusters was still significant in case of total testis and nearly significant in case of testicular somatic cells (Table [5](#T5){ref-type="table"}). Not surprisingly, the most highly expressed genes detected in total testis and involved in spermatogenesis (protamine 1, 2, 3 and transition protein 2) formed one of the tight clusters on chromosome 16. The results indicate a nonrandom distribution of the genes preferentially expressed in total testis and testicular somatic cells into chromosomal clusters, which did not arise from tandem duplications.
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
Number of preferentially expressed genes in testis located in clusters within tandem duplicate-free mouse genome Out of the 19 684 known genes (LocusLink) mapped on mouse genome assembly (NCBI, build32), 16 858 genes remained in tandem duplicate-free genome, including 1 300 and 1 050 preferentially expressed genes in total testis and testicular somatic cells, respectively. Chromosome search found clusters containing at least three adjacent preferentially expressed genes (tight clusters) or at least three preferentially expressed genes among the six adjacent genes (loose clusters). The tight clusters therefore form a subset of the loose clusters. Observed gene counts were evaluated using permutations (100000 random shufflings of the expression status of genes while keeping the gene positions constant) and the average number of genes located in clusters in the randomized genomes was computed.
:::
Total testis (TT 1+2) Adult testis somatic cells (ATSC)
------------------------------------------------------ ----------------------- ----------------------------------- ------------ -------------
Observed gene counts 44 230 36 120
Proportion of preferentially expressed genes 3.4 % 17.7 % 3.4 % 11.4 %
Gene counts in randomized genomes (mean ± std. dev.) 21.9 ± 8.1 168.4 ± 20.1 11.7 ± 5.9 94.2 ± 15.6
Ratio observed/mean in randomized genomes 2.0 1.4 3.1 1.3
Permutations yielding \> = observed gene counts 741 180 52 5 722
p-value (one tailed) 0.0074 0.0018 0.0005 0.0572
:::
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Chromosomal positions of clusters containing preferentially expressed genes in testis.**The positions of 103 gene clusters according to the physical map are displayed on an ideogram with corresponding cytogenetic bands on chromosomes. The clusters revealed in total testis and testicular somatic cells are not distinguished. Tight clusters (long dashes) form a subset of loose clusters (short dashes).
:::
:::
Comparing the B6 and BDF1 total testis transcriptomes
-----------------------------------------------------
In a recent study focused on senescence changes in testis, a modified SAGE method was used to generate digital gene expression profiles of total testis from 3- and 29-month-old mice of the BDF1 strain and 14-month-old mice of the SAMP1 strain that exhibits an accelerated senescence \[[@B32]\]. Because of the different anchoring enzyme (*Rsa*I) used in construction of the libraries and the limited availability of data from the BDF1 testis transcriptome, we could perform only a rough manual comparison of our B6 testis transcriptome (76 854 tags) and the combined BDF1 testis transcriptome from 3- and 29-month-old BDF1 mice (41 221 tags). We focused on the most highly expressed testicular genes in GNF Mouse Atlas v2 \[[@B33],[@B34]\] that were detected by Affymetrix GeneChips. A set of 35 highly expressed genes in testis (average difference \> 9 000) was organized with SAGE tag counts from B6 and BDF1 testis (see [Additional file 9](#S9){ref-type="supplementary-material"}). In the B6 total testis, we detected 33 out of 35 genes (the *Serf1*gene could not be distinguished because its low complexity tag matches multiple genes and the *Cox7a2*gene is not detected because its transcript lacks *Nla*III restriction site). In contrast, only 9 genes were detected in the BDF1 testis library, 13 genes were missing due to the absence of *Rsa*I restriction site in the transcript and for 13 other genes the expression data from BDF1 testis were not publicly available. Furthermore, out of the 35 highly expressed genes in testis, 21 genes were among the top 100 most expressed genes in the B6 total testis library, but only 9 genes were among the top 100 most expressed genes in the BDF1 total testis library. It appears that our SAGE data from the B6 testis transcriptome shows better correspondence to the microarray data than the data from the transcriptome of BDF1 testis.
Discussion
==========
Serial analysis of gene expression is a high-throughput method for building a catalogue of expressed genes and their expression levels of \"normal\" as well as diseased or genetically variant tissues and organs \[[@B16]\]. The digital character of SAGE data enables addition and direct comparison of different SAGE libraries, provided they were built with the same anchoring enzyme and originated from individuals of the same species. The utilization of such global transcriptome databases is multifold, including positional cloning of mutations or quantitative trait loci \[[@B35],[@B36]\], functional genome annotation \[[@B37],[@B38]\] or analysis of a nonrandom gene order \[[@B39]\]. Admittedly, the SAGE, as used in this work, has several limitations, including a significant proportion of repetitive and low complexity tags. The SAGE is obviously more labor-intensive than transcriptome analysis based on microarrays. At present, some of these inconveniencies can be solved by applying LongSAGE or massively parallel signature sequencing technologies \[[@B38],[@B40]\].
In this study we constructed a SAGE library of the total testis of the C57BL/6J (B6) mouse inbred strain, compared it with other public available mouse SAGE libraries and analyzed localization of testis-expressed genes within the mouse genome. The B6 strain was favored for the availability of its high-quality draft genomic sequence \[[@B17]\] and because series of congenics and recently also consomic strains have used the B6 strain as a background strain \[[@B18]\] \[Gregorova S, Forejt J, personal communication\]. The combined total testis SAGE library, TT 1+2, consisted of 76 854 total tags representing 24 529 unique tags. The tag-to-gene reliable identification method used in Mouse SAGE Site \[[@B24]\] was applied to tags with frequency ≥ 2. Out of these tags, 47.5% (3 553) revealed a reliable match to single and 15.5% (1 157) to multiple UniGene clusters. Considering the size of the total testis SAGE library, medium to highly expressed genes are present in the expression profile. The library size is comparable to the recently published SAGE library of somatic cells of the mouse testis \[[@B19]\] and almost twice the size of a library constructed from the total testis of BDF1 hybrid mice using a modified SAGE method \[[@B32]\].
Contrary to microarrays, SAGE data are platform independent, which permits the use of unrelated datasets coming from various sources to compare gene expression patterns. We analyzed the mouse testis transcriptome by comparing our total testis SAGE library to the adult testis somatic cells library \[[@B19]\] and to additional publicly available SAGE libraries from 7 different tissues. We recognized three different modes of differential expression. (1) Predominant expression of genes in the germinal or somatic part of the testis, which did not consider expression in other tissues. (2) Preferential expression in testis that was defined by comparing the expression of testis to 7 somatic tissues for which SAGE data were available. (3) Testis-specific expression that was defined by null expression (at the resolution of a particular SAGE library) in SAGE libraries of seven tissues or organs other than testis. Complete lists of genes predominantly expressed in germinal or testis-somatic cells, as well as the catalogues of genes preferentially expressed in testis and testis-specific genes are available online in [Additional file 3](#S3){ref-type="supplementary-material"} , [5](#S5){ref-type="supplementary-material"} and [Additional file 7](#S7){ref-type="supplementary-material"}.
Conflicting results have been reported on the representation of male-biased genes on the X chromosome in various species. Spermatogonia-specific genes were found to be an order of magnitude more abundant on the mouse X chromosome \[[@B30]\]. In human, the prostate-specific genes were twice more frequent on the X chromosome, but the female mammary gland- and ovary-specific X-linked genes were not enriched in respective SAGE libraries \[[@B29]\]. On the contrary, under-representation or absence of male-biased genes on the X chromosome was reported in *Caenorhabditis elegans*\[[@B41]\] and in *Drosophila*\[[@B42],[@B43]\]. In the mouse, an under-representation of testis-expressed and testis-enriched genes on the X chromosome was also revealed by the analysis of microarray and EST data \[[@B5]-[@B7]\]. Our present data favor under-representation of X-linked genes in the total testis transcriptome but not in testis-somatic cells. Because the germ cells in different stages of differentiation constitute about 90% of the total cell mass of testis, the data indicate that the deficit of X-linked testis-expressed genes may reflect the lack of transcription from the X chromosome in meiotic cells. These results are in agreement with the idea of X-chromosome silencing during the first meiotic division, the phenomenon based mostly on circumstantial evidence in flies and mice \[[@B7],[@B44]-[@B46]\]. Thus, transcription at the haploid stage of spermatogenesis is expected for most of the X-linked genes expressed in total testis. The meiotic X chromosome inactivation seems to be restricted to primary spermatocytes, but Sertoli cells, which form the somatic part of seminiferous tubules, may have the X chromosome in the active state. Indeed, in the transcriptome of adult testis somatic cells the proportion of expressed X-linked genes (3.2 %) was more than twice higher than in total testis (1.4 %) and did not differ from the proportion of X-linked genes expressed in non-testis (somatic) tissues.
Testis-specific genes belong to a wider category of sex-biased genes, which according to the hypothesis of sexually antagonistic genes are more likely to spread on the X chromosome than on autosomes \[[@B47]\]. This is because on the X chromosome they will express their favorable effect in the hemizygous state (XY) while their deleterious effect will be masked by their recessivity in the other sex (XX). Consequently, accumulation of male-specific genes on the X chromosome will be possible by the effect of modifiers that narrow the expression of sex-biased genes only to the male sex \[[@B47]\]. Thus, the evolution of sexually antagonistic genes and X inactivation may act as opposing forces on the germline lineage of testis while accumulation of male-specific genes could be expected in somatic cells of testis. In accord with these assumptions the proportion of X-linked genes specific for total testis did not significantly differ from the proportion of genes specific for other tissues, while we observed a significant 3.2-fold enrichment of the proportion of X-linked genes specific for testicular somatic cells.
The eukaryotic gene order is nonrandom obviously not only due to shifting of sex-biased genes to and from the X chromosome, but also owing to a nonrandom clustering of genes within chromosomes. This somewhat unexpected conclusion (taking into account the relative autonomy of transgene regulation) is gaining gradual support from global transcriptome analyses of various eukaryotic species (see Hurst et al. for review) \[[@B39]\]. The observed examples of clustering are apparently a mixture of several unrelated phenomena, including large domains of similarly expressed genes in *Drosophila*and humans \[[@B48],[@B49]\], clustering of housekeeping genes \[[@B50]\], clustering of highly expressed genes \[[@B51]\] or genes with similar expression breadth in regions of similar GC content \[[@B52]\]. In *Drosophila melanogaster*one third of testes-specific genes occur in clusters \[[@B43]\], a phenomenon not reported in any other species. Using PEM \[[@B31]\] to define preferentially expressed genes we were able to demonstrate that in the mouse, the genes preferentially expressed in germ cells as well as in somatic cells of testis occur in tight clusters with a frequency 2.0-fold and 3.1-fold higher than the expected average frequency in randomized genomes. Moreover, our results indicate that this phenomenon is not merely a consequence of tandem duplications. Further analysis of clustering of testis-expressed genes may reveal new insights into the functional organization of the mammalian genome.
Conclusion
==========
We identified chromosomal clusters of adjacent genes with preferential expression in testis that contain a significantly higher number of genes than expected by chance. This phenomenon is not merely a consequence of tandem duplication. The genes with specific expression in testicular somatic cells are more abundant on the X chromosome, which favors the theory of accumulation of male-biased genes on the X chromosome. In contrast, the X-linked genes are under-represented in the transcriptome of total testis, which is in accordance with the idea of X-chromosome inactivation during the first meiotic division.
Methods
=======
Tissue collection and RNA isolation
-----------------------------------
Mice were housed in specific pathogen free environment and their manipulation was in accordance with the Czech Animal Protection Act No. 246/92, 162/93, and decrees No. 311/97, fully compatible with the NIH Publication No. 85-23, revised 1985. Testes were obtained from 9-week-old males of the C57BL/6J mouse strain. The animals were killed by cervical dislocation; the testes were quickly removed from the body and released from tunica. The total RNA was extracted from homogenized testes using TRIzol (Invitrogen) according to the manufacturer\'s protocol. SAGE libraries were constructed from the total RNA isolated from both testes of a single male (TT 1) and from the pool consisting of equal weight amounts of total RNA isolated from both testes of three male littermates (TT 2).
Construction of SAGE libraries, sequencing and tag extraction
-------------------------------------------------------------
SAGE libraries were constructed as described in the MicroSAGE protocol version 1.0e available from SAGE homepage \[[@B53]\] using *Nla*III as the anchoring enzyme and *Bsm*FI as the tagging enzyme. Two minor modifications of the MicroSAGE protocol were employed: the first strand cDNA synthesis reaction was incubated at 42°C and the amount of linkers used in the linker ligation step was decreased to \~10 ng. Sequencing was performed in a Beckmann Coulter CEQ 2000 DNA Analysis System. The sequence files were processed for the tag extraction using a custom Perl script. Tags were extracted only from clones containing \> 2 ditags. Duplicated ditags, linker tags and all 1-bp linker variations were removed. Data of total testis SAGE libraries are available in the GEO repository \[[@B20]\] under accession numbers GSM34767 (TT 1) and GSM34768 (TT 2).
Identification of SAGE tags
---------------------------
Tag identification to UniGene clusters was done using three methods: SAGEmap reliable mapping \[[@B22]\], Mouse SAGE Site reliable mapping \[[@B24]\] and RNA evidence mapping. The SAGEmap reliable mapping \[[@B23]\] uses a reliability score to classify tag-to-gene associations and tag-to-gene associations with the top two reliability scores are considered reliable. The Mouse SAGE Site \[[@B21]\] reliable mapping is based on the SAGEmap full mapping file and considers reliable the tag-to-gene associations that are supported by tags extracted from at least one mRNA sequence (from RefSeq, Mammalian Gene Collection, GenBank) or at least 3 ESTs with a poly(A) signal or at least 8 ESTs with no poly(A) signal. The RNA evidence mapping is also based on the SAGEmap full mapping file and considers reliable only tag-to-gene associations supported by tags extracted from at least one mRNA sequence. Mitochondrial tags were identified using all possible tags extracted from the mouse mitochondrial genome reference sequence \[GenBank:NC\_005089\].
Comparison of testis SAGE libraries
-----------------------------------
Tags significantly different between SAGE libraries were determined by Monte Carlo simulations. Using the described algorithm \[[@B54]\] a set of 100 000 random tables was generated keeping the row and column totals of the observed data fixed. For each tag, the proportion of simulations that produced a difference equal to or greater than the observed difference (p-chance) was computed. The set of 100 000 random tables was generated six times and the average p-chance was calculated. The fold factor was computed as the ratio of normalized tag counts in two SAGE libraries with values \< 1 converted to reciprocal negatives. For the tags absent in one library a normalized tag count of single copy tags was assumed.
Data sources
------------
The SAGE library from somatic cells of the adult testis \[[@B19]\] was obtained from GEO repository \[[@B20]\], accession number GSM5435. Other SAGE libraries were obtained from GEO repository or downloaded from Internet sources (see [Additional file 4](#S4){ref-type="supplementary-material"} ). The data from the BDF1 testis SAGE library were obtained from a printed table in publication \[[@B32]\] (only the top 100 genes expressed in BDF1 testis are listed in publication, the whole library is currently not publicly available). Microarray data of mouse testis, generated by the GNF Mouse Atlas v2 project \[[@B33]\], were obtained from the hgFixed database of the UCSC Genome Browser \[[@B55],[@B56]\].
Hierarchical clustering of mouse SAGE libraries
-----------------------------------------------
Thirty-two mouse SAGE libraries constructed from bulk tissues (including normal and diseased) that were publicly available to date (July 1, 2004) were selected (see [Additional file 4](#S4){ref-type="supplementary-material"}). For each pair of SAGE libraries a distance based on differences between normalized tag counts was computed \[[@B28]\]. The average agglomeration method was used in hierarchical clustering because of the highest cophenetic correlation (Pearson correlation between the observed distances and the distances calculated from the dendrogram).
Selection and preparation of mouse SAGE libraries for genomic analysis
----------------------------------------------------------------------
Twenty-seven SAGE libraries created from bulk tissues (excluding tumors) were organized into 7 groups by tissue type and tag counts from SAGE libraries within each group were combined (see [Additional file 4](#S4){ref-type="supplementary-material"}). The groups of SAGE libraries include: brain (9 libraries, 329 745 tags), eye (6 libraries, 336 399 tags), heart (1 library, 84 275 tags), liver (2 libraries, 37 118 tags), kidney (6 libraries, 87 810 tags), limbs (2 libraries, 136 650 tags) and adipose tissue (1 library, 44 974 tags). These groups were analyzed in parallel with total testis (2 libraries, 76 854 tags) and adult testis somatic cells (1 library, 81 478 tags). All tags from prepared tissue groups, total testis and adult testis somatic cells SAGE libraries were identified to UniGene clusters using RNA evidence mapping (tag-to-gene association is supported by at least one mRNA sequence) and linked to LocusLink genes. Only tags with identification to a single LocusLink gene were subjected to further analysis. Tag counts from multiple tags matching the same LocusLink gene were combined.
Distribution of tissue-specific genes on chromosomes
----------------------------------------------------
Analysis was done in parallel for testis-specific genes in total testis and somatic cells of adult testis. The tissue-specific genes were selected according to tag counts in the testis tissue and 7 non-testis tissues (see [Additional file 4](#S4){ref-type="supplementary-material"}). A gene was considered to be tissue-specific if it was expressed only in one tissue and its expression was supported by tag count \> 1. Each tissue-specific gene was then assigned to a chromosome (autosome or X chromosome) according to the LocusLink database and the group (testis or non-testis). The permutations algorithm performed 100 000 random shufflings of the chromosomes while keeping the sum of genes on autosomes and the X chromosome constant. The p-value (two tailed) was computed as doubled number of permutations yielding gene counts above/below (which of this was lower) or equal to the observed gene counts in testis tissue and the X chromosome.
Identification of chromosomal clusters of genes with preferential expression in testis
--------------------------------------------------------------------------------------
The preferential expression measure (PEM) \[[@B31]\] was used to score differential expression of genes in testis tissues. PEM for total testis (PEM~TT~) and adult testis somatic cells (PEM~ATSC~) were calculated for each gene. The gene was considered to be preferentially expressed in total testis if PEM~TT~\> = 1/2 PEM~TT(max)~, and in somatic cells of adult testis if PEM~ATSC~\> = 1/2PEM~ATSC(max)~. PEM~(max)~values represent the maximum PEM value encountered in the tissue, PEM~TT(max)~= 1.169, PEM~ATSC(max)~= 1.145.
To prepare a tandem duplicate-free mouse genome we considered 19 684 known genes from the LocusLink database that were mapped on the mouse genome assembly (NCBI build 32) \[[@B57]\]. For each LocusLink gene, we obtained a known protein sequence (NP\_ accessions) from the mouse RefSeq collection \[[@B58]\] and performed protein BLAST (standard settings) against the RefSeq known protein collection. The hits with expectation value \< 1e^-10^and with an alignment of at least 50% length and 30% identity of the query sequence were processed and identified to LocusLink genes. If a LocusLink gene located in the vicinity of the original LocusLink gene was found among the hits (considering 10 adjacent genes in both directions), both genes were considered as a tandem duplicate pair and were excluded from the genome. As a result a tandem duplicate-free genome with 16 858 LocusLink genes was obtained.
Two sets of gene clusters with preferentially expressed genes were identified -- for total testis and somatic cells of adult testis. All LocusLink genes from the tandem duplicate-free mouse genome were associated with the expression status (preferentially expressed, expressed, unknown). Each chromosome was searched using a sliding window of three adjacent genes and three consecutive preferentially expressed genes were considered as a cluster (tight clusters). Another search was performed using a sliding window of six adjacent genes and at least three preferentially expressed genes were required to form a cluster spanning from the first to the last preferentially expressed gene (loose clusters). The overlapping clusters were merged into a single cluster encompassing all involved genes (separately for tight or loose clusters). The permutations performed 100 000 random shufflings of the expression status in the genome while keeping the gene positions constant. A search with the above defined sliding windows determined the number of preferentially expressed genes located in clusters in each randomized genome. The p-value (one tailed) was computed as the number of permutations yielding greater than or equal to the observed number of preferentially expressed genes located in clusters.
Statistical evaluation
----------------------
All statistical analyses, including Monte Carlo simulations, hierarchical clustering, chromosomal and gene permutations were conducted in R statistical environment \[[@B59]\] using custom scripts.
Database versions
-----------------
The following database versions were used in all analyses: Mouse UniGene build \#136 (March 26, 2004), mouse SAGEmap (April 3, 2004) corresponding to the mouse UniGene \#136, LocusLink (April 3, 2004), mouse genome assembly NCBI build 32 (November 2003), mouse Reference Sequence collection (April 3, 2004) and Gene Ontology database (July, 2004).
Authors\' contributions
=======================
PD constructed the SAGE libraries and performed the bioinformatic analyses. CV carried out sequencing of SAGE libraries. PS participated in bioinformatic analysis of gene order. VP is Head of the Center for Integrated Genomics. JF conceived the study and coordinated work. PD and JF wrote the article. All authors read and approved the final manuscript.
Supplementary Material
======================
::: {.caption}
###### Additional File 1
**SAGE tags detected in mouse total testis**. List of 7 481 tags with tag count \> 1 in combined total testis SAGE library with reliable tag identification according to the Mouse SAGE Site database.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 2
**Tags with significantly different tag counts between total testis and adult testis somatic cells**. List of 3 258 tags with significantly different tag counts between the total testis and adult testis somatic cells SAGE libraries determined by Monte Carlo simulations (1 691 tags have increased tag counts in total testis, 1 567 tags have increased tag counts in adult testis somatic cells at p-chance \< 0.05). The reliable tag identification according to the Mouse SAGE Site database is provided.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 3
**Genes with predominant expression in germinal and somatic cells of the testis**. List of 924 and 802 genes with predominant expression in germinal and somatic cells of the testis, respectively, based on the comparison of total testis and adult testis somatic cells SAGE libraries. Tags with significantly different tag counts (p-chance \< 0.05, Monte Carlo simulation) and at least five-fold increased/decreased tag counts were selected and identified using RNA evidence mapping.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 4
**Mouse SAGE libraries used in genomic analysis and hierarchical clustering**. List of mouse SAGE libraries publicly available to date July 1, 2004 that were used in genomic analysis and hierarchical clustering.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 5
**Genes specific for total testis or adult testis somatic cells**. List of 395 and 81 genes with specific expression in total testis or adult testis somatic cells determined by comparison to SAGE data from seven non-testis tissues (brain, eye, heart, liver, kidney, limbs and adipose tissue). A gene was considered to be testis specific if the corresponding tags were present only in total testis or adult testis somatic cells SAGE libraries and missing in all non-testis libraries. Tags were identified using RNA evidence mapping.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 6
**Testis expressed genes located on the X chromosome (summary)**. Summary of the X-linked genes expressed in total testis and adult testis somatic cells. Tags were identified using RNA evidence mapping.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 7
**Preferentially expressed genes in total testis and testicular somatic cells**. Preferentially expressed genes were determined separately for total testis and adult testis somatic cells in conjunction with their expression in seven non-testis tissues. Expression of genes was scored using preferential expression measure (PEM). A gene was considered to be preferentially expressed if PEM score was above 50 % of the maximum PEM value encountered in that tissue. Tags were identified using RNA evidence mapping.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 8
**Genes preferentially expressed in testis located in chromosomal clusters within tandem duplicate-free genome**. Chromosomal clusters of genes preferentially expressed in testis were localized by the search with a sliding window. Two types of clusters were identified: tight clusters (containing at least three adjacent preferentially expressed genes in testis) and loose clusters (containing at least three preferentially expressed genes in testis among the six adjacent genes).
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 9
**Manual comparison of the most highly expressed genes in three total testis transcriptomes (GNF atlas, B6 testis, BDF1 testis)**. The list of 35 most highly expressed genes in total testis according to the GNF Mouse Atlas v2 organized with the appropriate *Nla*III and *Rsa*I SAGE tags extracted from their representative mRNA/RefSeq sequences.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
We thank Laurence D. Hurst, Adam Pavlíček and Jan Pačes for helpful comments and suggestions, Radka Storchová, Zdeněk Trachtulec and Šárka Takáčová for critically reading the manuscript. This work is supported by the project of the Czech Ministry of Education, Youth and Sports No. LN00A079 -- Center for Integrated Genomics and by the project of the Academy of Sciences of the Czech Republic No. K5052113. J.F. is supported as an International Scholar of the Howard Hughes Medical Institute.
|
PubMed Central
|
2024-06-05T03:55:55.227973
|
2005-3-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079818/",
"journal": "BMC Genomics. 2005 Mar 5; 6:29",
"authors": [
{
"first": "Petr",
"last": "Divina"
},
{
"first": "Čestmír",
"last": "Vlček"
},
{
"first": "Petr",
"last": "Strnad"
},
{
"first": "Václav",
"last": "Pačes"
},
{
"first": "Jiří",
"last": "Forejt"
}
]
}
|
PMC1079819
|
Background
==========
*Plasmodium spp*. are protozoan parasites (phylum *Apicomplexa*, class *Sporozoa*, order *Haemosporida*) with a complex life cycle, including alternating development phases in mosquito and vertebrate hosts. More than a hundred *Plasmodium*species have been described, of which four infect humans and cause malaria. The genome sequences of the most pathogenic human species, *P. falciparum*\[[@B1]\], and of *P. yoelii yoelii*, a murine malaria species \[[@B2]\], were published in 2002. Sequencing of additional *Plasmodium*species is underway. Recently, the genome of *Cryptosporidium parvum*another Apicomplexan parasite has been published \[[@B3]\], while the genome of *Toxoplasma gondii*is being sequenced. The body of information available offers new opportunities in malaria research, and will hopefully accelerate rational design of novel intervention strategies, a pressing need in view of the rapidly deteriorating efficiency of existing control tools.
In recent years, our laboratory has explored the vaccine potential of the trophozoite protein R45, a *P. falciparum*antigen exported by the parasite to the infected red blood cell membrane \[[@B4]\]. The protein contains a large central domain with 90 copies of a HK/MSDH/SN consensus hexapeptide, encoded by the gene annotated as PFD1175w in the *P. falciparum*genome \[[@B1]\]. We have shown that a recombinant antigen called R23, which contains 11 such repeats is the target of antibodies promoting phagocytosis of parasitized red blood cells \[[@B5]\] and induces strong protective immunity against lethal infection in *Saimiri sciureus*squirrel monkeys \[[@B6]-[@B8]\]. Furthermore, the response to R23 repeats is associated with protection against clinical malaria in humans \[[@B9]\]. The PFD1175w gene sequence is well conserved among all isolates of *P. falciparum*studied so far. In order to obtain information on the organisational and expression constraints of the antigen, we carried out a comparative genomic approach. This showed that the N-terminal region and the central hexapeptide repeats were *P. falciparum*-specific and unique to PFD1175w. In contrast, multiple paralogs of the C-terminal region were identified in *P. falciparum*and a single ortholog in most other *Plasmodium*species as well as in some distantly related Apicomplexan species. This C-terminal region is predicted to be a protein kinase domain \[[@B4]\], with high divergence from all other known kinase domains. We describe here some characteristics by this novel gene family and of this novel kinase domain, and discuss the possible relevance of species-specific gene amplification.
Results
=======
Identification of R45 paralogs in *P. falciparum*
-------------------------------------------------
The R45 trophozoite protein (PFD1175w) has a predicted three-exon structure, with a small (201 bp) exon 1, a large (3387 bp) exon 2 followed by a short (78 bp) exon 3. Gene structure and exon / intron boundaries were experimentally confirmed by cDNA sequencing (data not shown). Based on homology with PFD1175w, a gene family with 20 paralogs was identified in the genome of the 3D7 clone of *P. falciparum*. Data mining indicated that all 20 paralogs are transcribed at a stage or another of the life cycle, and for many members at multiple stages (Figure [1](#F1){ref-type="fig"}) \[[@B16],[@B17]\]. Twelve genes were re-annotated to fit with the gene structure and amino acid sequence of the other paralogs (see Materials and Methods). All genes have a three exon structure with a large exon 2 flanked by two small exons, except Mal7P1.175, where the small exon 1 appears to be missing (or fused to exon 2) (Figure [1](#F1){ref-type="fig"}). Two paralogs (Mal7P1.175 and PF14\_0733+4) have an internal stop codon, reflecting pseudogenes or need for read through translation \[[@B18]\].
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Schematic representation of the R45-FIKK kinase genes in the *P. falciparum*3D7 genome including salient protein features as deduced from PlasmoDB (as per release 4.3) and mRNA expression profiling as published by Bozdech et al \[16\] and Le Roch et al \[17\]. The 20 genes of the family are scattered over 11 chromosomes, as indicated. Stage of expression refers to the stages in the life cycle where the mRNA has been consistently detected and / or is maximal (R, T, Sc, Mz and Sp refer to ring stage, trophozoite, schizont, merozoite and sporozoite, respectively). The kinase domain is depicted as a black bar. The polymorphic region within the kinase domain is shown in white. Exon borders are indicated with a downwards arrow head. The location of the pentameric Plasmodium Export element (Pexel) is marked with an asterisk. The Pexel sequence is indicated for each deduced protein. It is indicated in italics when it does not entirely match with the consensus (R,K)XLX(D,Q,E). ^†^genes re-annotated in this work (see Materials and Methods).
:::
:::
The hallmark of the family is a conserved C-terminal domain showing 38 -- 64 % identity among the paralogs, with its terminal 24 -- 41 residues encoded by the small exon 3 \[exon 2 in Mal7P1.175\] (Figure [1](#F1){ref-type="fig"}). This conserved domain has characteristic features of protein kinases, but can not be assigned to any known kinase family. The presence of all amino acids necessary for phosphotransfer except the canonical ATP fixation motif GxGxxG in all 20 paralogs \[see [Additional file 1](#S1){ref-type="supplementary-material"}\] suggests a novel protein kinase type. This is consistent with the conclusions of Ward et al, who independently identified this protein kinase family, that was called FIKK-kinase, using a whole genome search for protein kinases in *P. falciparum*\[[@B19]\]. Two members contain a repeat region inserted between subdomains III and IV consisting of 90 copies of the H K S/N D N/H/S N hexapeptide in PFD1175w and of 32 copies of the two amino acid motif NE/G in PFI0125. The other family members contain a stretch of up to 53 non-conserved and non-repeated amino acids in this region, as illustrated in Figure [1](#F1){ref-type="fig"}.
Contrasting with the conserved C-terminal kinase domain, the N-terminal region is unique to each paralog, with less than 20% identity and substantial size variation (ranging from 124 residues for PFI0100c to 471 residues for PF10\_0380). Homology and motif searches for each unique N-terminal region did not identify any known Pfam domain or a significant homology with any other *P. falciparum*protein. However, all twenty paralogs contain a stretch of hydrophobic amino acids within the first 70 amino acids, corresponding to a predicted trans-membrane or signal sequence for 14 members according to PlasmoDB. Analysis of the re-annotated paralogs using an improved signal peptide prediction algorithm \[[@B10]\] and manual curation identified a likely signal / anchor sequence in all but one paralogs (Figure [1](#F1){ref-type="fig"}). For PFI0100c, there was no signal / anchor sequence probability but there was a trans-membrane prediction. We considered here as a possible signal / anchor sequence that might have escaped detection by the algorithms used. The secretion motif RxSRILAExxx identified recently \[[@B20]\] was present in six paralogs (PFD1165w, PFI0105c, PF10\_0160, PFL0040c, MAL13P1.109, PF14\_0733+4). Interestingly however, the shorter *Plasmodium*export element (Pexel) identified independently by Marti et al \[[@B21]\] was detected in all paralogs downstream of the signal / anchor sequence (Figure [1](#F1){ref-type="fig"}) \[see [Additional file 2](#S2){ref-type="supplementary-material"}\]. The consensus (R,K) × (L,I) × (E,Q,D) pentamer was observed in 18 paralogs. A non-canonical amino acid in the fifth position was found in PFI0095c and PFI00105c (A and V, respectively). The Pexel motif is located 16--24 amino acids downstream of the start of the second exon, except in Mal7P1.175, where it is 11 residues downstream of the signal sequence. Pf11\_510 has two putative Pexel motifs located at positions 20--24 and 44--48.
R45-FIKK kinase orthologs in other Plasmodium species and in other Apicomplexans
--------------------------------------------------------------------------------
Orthologs were identified in most *Plasmodium*genome sequences available to date. A single copy gene was observed in the human species *P. vivax*(PvR45), the simian species *P. knowlesi*(PkR45), the murine species *P.y. yoelii*(PyR45) and *P. berghei*(PbR45), and the avian species *P. gallinacaeum*(PgR45). In contrast, six paralogs were identified in *P. reichenowi*(PrR45-1-6). No ortholog was identified in the murine species *P. chabaudi*, possibly reflecting the low coverage (2×) of the available genome sequence.
The deduced protein sequence alignment showed the FIKK-protein kinase domain to be well conserved within the genus \[see [Additional file 1](#S1){ref-type="supplementary-material"}\]. None of the non *P. falciparum*orthologs identified here contain any repeat region. Based on their kinase domain, the R45-FIKK kinase proteins from six *Plasmodium*species, namely PvR45, PkR45, PyR45, PbR45, PgR45 and PrR45-2, clustered perfectly in a phylogenetic tree with 82 -- 98 % identity (Figure [2](#F2){ref-type="fig"}), suggesting that they derive from the same common ancestor. Intriguingly, no *P. falciparum*R45-like protein was observed in this cluster. The best match, Pf10\_0160, showed a non-significant bootstrap of 44. The six R45-like orthologs of the core cluster are quite distinct from the *P. falciparum*R45-FIKK genes with regard to their gene structure and N-terminal region (Figure [3a](#F3){ref-type="fig"}) \[see [Additional file 3](#S3){ref-type="supplementary-material"}\]. They have a two exon structure. The short, highly conserved second exon is the ortholog of the conserved third *P. falciparum*exon. All genes lack an ortholog of the first *P. falciparum*exon and hence encode a protein without a signal sequence and a Pexel motif. In addition, the N-terminal region is much larger than for any *P. falciparum*paralog, ranging from 1060 residues in PkR45 to 759 residues in PrR45-2. Interestingly, analysis of the N-terminal region identified pairs of orthologs, with 56 % identity between PvR45 and PkR45, and 82 % identity between PyR45 and PbR45 (Figure [3a](#F3){ref-type="fig"}). These pairs form two quite separate homology groups, with less than 20 % overall inter-group identity. Despite of the low overall conservation, stretches of conserved amino acids are present within the 200 downstream residues of the N-terminal region \[see [Additional file 3](#S3){ref-type="supplementary-material"}\]. These motifs are not observed in any *P. falciparum*ortholog \[see [Additional file 2](#S2){ref-type="supplementary-material"}\].
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Phylogenetic distance of 32 R45-FIKK kinases as inferred from their C-terminal FIKK kinase domain. A consensus tree was built with the *T. gondii*sequence as outgroup. Numbers indicate bootstrap values, with 100 corresponding to a perfect cluster.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Schematic representation of R45-FIKK kinase genes in several *Plasmodium*species. a) Clustering of the R45-FIKK kinase genes into a single ortholog group. b) Orthology between *P. falciparum*and *P. reichenowi*. The degree of identity of the N-terminal domain and of the kinase domain between the orthologous pairs is indicated on the left and right hand side, respectively. NA = not applicable. Symbols are as in Figure 1.
:::
:::
*P. reichenowi*, like *P. falciparum*, has multiple R45-FIKK kinase paralogs. As indicated above, PrR45-2, clusters with the orthologs from the single copy species and does not have an ortholog in the 3D7 *P. falciparum*genome. Interestingly however, the five other *P. reichenowi*paralogs could each be attributed to a specific *P. falciparum*R45-FIKK kinase paralog (Figure [3b](#F3){ref-type="fig"}), forming orthologous pairs as follows: PrR45-1 -- PF10\_380, PrR45-3 -- PFI0110, PrR45-4 -- Mal7P1.144, PrR45-5 -- PFC0060, PrR45-6 -- PFI0125. For those *P. reichenowi*paralogs were enough sequence data was available, there was remarkable conservation of both the kinase domain (87 -- 97 % identity with the respective *P. falciparum*ortholog) and the N-terminal region (75--91 % idem). Interestingly, the two *P. reichenowi*orthologs with full length sequence available, PrR45-3 and PrR45-4, share the 3 exon structure with their *P. falciparum*orthologs, have an exon 1-encoded signal sequence and a conserved Pexel sequence close to the start of exon 2. Progress in the *P. reichenowi*genome should indicate whether or not the other orthologs present the same structure so far observed only in *P. falciparum*and *P. reichenowi*.
A single copy R45-like ortholog was identified in *T. gondii*(TgR45) as well as in *C. parvum*(CpR45), two parasite species that belong to a distinct class, namely *Coccidea*, within the phylum *Apicomplexa*. Their deduced amino acid sequence is quite divergent from the *Plasmodium spp*. family members ([Additional file 1](#S1){ref-type="supplementary-material"} and [2](#S2){ref-type="supplementary-material"}). The protein kinase domain of CpR45 and TgR45 shows 41 -- 42 % and 31 -- 33 % identity, respectively, with the cluster of orthologs. The lower value for TgR45 is explained by insertions of stretches of amino acids between some subdomains. The gene structure of TgR45 and CpR45 differs considerably from their orthologs in the various *Plasmodium spp*. TgR45 is encoded by a minimum of seven exons, with possible additional upstream exons. While the seventh exon corresponds to the third exon of the *P. falciparum*paralogs, the remaining kinase domain is encoded by six exons. A signal sequence is predicted within the first 21 amino acids of TgR45, but not followed by a Pexel motif. The significance of this finding is unclear, because the actual start of the TgR45coding region is not yet identified. In contrast, in line with the general paucity of introns in the *C. parvum*genome, CpR45 is encoded by a single exon \[[@B3]\]. The C-terminal amino acids encoded by a specific exon in all the other orthologs are present and well conserved in CpR45.
Chromosome localisation in *P. falciparum*
------------------------------------------
The 20 *P. falciparum*paralogs are distributed over 11 chromosomes. Interestingly, 17 of 20 R45-FIKK kinase paralogs are located within 150 kb from their telomer, all with the sense of transcription directed towards the telomer (Figure [4](#F4){ref-type="fig"}). Seven of these paralogs are single copy on their respective chromosome. A cluster of seven head to tail parologs is located on chromosome 9, and two paralogs are separated by a single gene on chromosome 4. Chromosomes 7 and 10 have two paralogs, one in the sub-telomeric region and one in the central region. Chromosome 13 has a single centrally located gene.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
Chromosomal localisation of R45-FIKK kinase genes and of some associated multigene families in the *P. falciparum*3D7 genome, as deduced from the maps published by Gardner et al \[1\] and the version 4.3 genome annotation from PlasmoDB. The genes transcribed leftwards (annotated \"c\" in the 3D7 genome) are placed above the chromosome. The genes transcribed rightwards (annotated \"w\" in the 3D7 genome) are placed beneath the chromosome. Total chromosome length as deduced from PlasmoDB (as per December 2004) is given in megabases on the right. Note that for chromosomes whose sequence is still incomplete, the actual chromosome size is slightly larger. The subtelomeric regions are not drawn to scale.
:::
:::
Interestingly, the subtelomeric R45-FIKK kinase paralogs are consistently associated with paralogs from other multigene families with a predominant subtelomeric localisation. The most frequent association (15 of 17 R45-FIKK kinase subtelomeric paralogs) is with the ortholog group named 775394, which contains proteins with a DNA J domain, in particular, the ring erythrocyte surface antigens (*resa*1--3). There seems to be a close association between both families as in most cases the 775394 paralog is the closest or the next closest gene to a R45-FIKK paralog. For instance, PFD1165w and PFD1175w, which are located 80 kb from the chromosome 4 telomer, are each flanked in 3\' by a 775394 paralog. Likewise, PF11\_0510 is located between the *resa2*and *resa3*paralogs on chromosome 11. PFA0130c is located further apart from *resa1*, but still is closely linked (less than 20 kb downstream of *resa1*). Approximately half of the 27 paralogs within the 775394 group are localised close to a R45-FIKK kinase gene. Additional subtelomeric gene families in the close vicinity of a R45-FIKK paralog include the 76615 (lysophospholipase), 779237 (erythrocyte binding antigens *eba*and *ebl*), 766147 (fatty acid CoA synthase) as well as several multigene families coding for hypothetical transmembrane proteins (see Figure [4](#F4){ref-type="fig"}). There are several examples of specific tandem arrangements such as *resa*, R45-like, *eba*(see Figure [4](#F4){ref-type="fig"}), suggesting an ancient higher order organisation.
*C. parvum*is the only other Apicomplexan for which chromosome maps have been established \[[@B3]\]. CpR45 (annotated cgd5\_4390) is located approx. 35 kb from the putative telomer on chromosome 5 \[[@B3]\], but does not seem to be associated with any specific subtelomeric family. In *P.y. yoelii*, the R45-FIKK ortholog is not present in a subtelomeric or telomeric contig \[[@B2]\], but the exact chromosomal localisation is unfortunately neither known for this species nor for other *Plasmodium*species, since at this stage of genome annotation and assembly, chromosome maps are quite incomplete. A partial synteny map has been established for some regions of the *P.y. yoelii*and *P. falciparum*chromosomes. Interestingly enough, many R45-FIKK kinase paralogs lie at the boundary of the subtelomeric region where both species diverge substantially and as such correspond to breakpoint in synteny. We looked systematically for *P.y. yoelii*orthologs in the vicinity of the various R45-FIKK kinase paralogs. We observed break of synteny close to all telomeric paralogs, confirming observations by other authors \[[@B2]\]. In particular the synteny reported for the *P.y. yoelii*chromosome 5 and *P. falciparum*4 stops close to the localisation of PFD1165w \[[@B2]\]. We also observed breaks of synteny close to internal R45-FIKK paralogs, in particular downstream of the internal gene MAL7P1.144, and upstream and downstream of PF10\_0160. We could not explore the vicinity of MAL13P1.109, which is still uncertain. Interestingly, the *P. vivax*contig that harbours PvR45 also contains a break in synteny *with P. falciparum*.
Discussion
==========
The family identified here on the basis of similarity with the non repeated regions from PFD1175w groups genes coding for a well conserved C-terminal protein kinase domain. Recently, Ward et al identified the same family using a whole genome mining for protein kinases. Based on a conserved motif from sub-domain II, they named this family \"FIKK-like kinase\" \[[@B19]\]. Since the kinase activity has not been formally demonstrated, and since the only paralog characterised in some detail so far is R45 (PFD1175w) \[[@B4]-[@B9]\], we prefer to name it the R45-FIKK kinase family.
The R45-FIKK family was observed only in protozoan parasites of the phylum *Apicomplexa*. A single copy gene was detected in most species except in *P. falciparum*and its closest relative *P. reichenowi*, where multiple copies were observed. As current genome coverage is ≥ 10× for *C. parvum*\[[@B3]\], *P. vivax*and *T. gondii*, 5× for *P. yoelii*\[[@B2]\] and *P. knowlesi*, the conclusion of a low (single) copy number is a fair statement. Due to incomplete coverage of the *P. berghei*, *P. gallinacaeum*and *P. chabaudi*genome sequence (2--3×), a definitive conclusion on the actual copy number in these species cannot be drawn, but a large copy number is highly unlikely. Conversely, copy number in *P. reichenowi*may be underestimated. Indeed, six copies have been detected in the available *P. reichenowi*sequence, currently only at 1.5× coverage. Additional copies might be uncovered upon completion of genome sequence.
The kinase domain is well conserved in all family members identified. It has many features of a catalytic protein kinase domain. The limited homology with well characterised kinase domains from other species precludes predicting a possible kinase type and phosphorylation target. This is in line with the observation that protozoan parasites such as *Plasmodium spp*., *T. gondii*and *C. parvum*have a quite different kinase repertoire as compared to yeast and metazoa \[[@B19],[@B22]\]. The fact that all residues essential for phosphotransfer are present in each kinase domain of the R45-FIKK kinase family suggests that these proteins function as protein kinases. These observations call for experimental investigation of the potential enzymatic activity of the R45-like proteins, the uniqueness of which could represent an attractive drug or vaccine target.
The N-terminal region is much less conserved than the kinase domain but interestingly, clear ortholog pairs with very good conservation were identified. Clustering of the two rodent species *P. berghei*and *P. yoelii*, of the human *P. vivax*and simian *P. knowlesi*and of *P. falciparum*and *P. reichenowi*is in line with the phylogeny inferred from rRNA and CSP gene sequences \[[@B23],[@B24]\]. The N-terminal domain of the R45-FIKK kinases may thus serve as a sensitive marker for genetic distance within the *Plasmodium*genus.
The conservation of the C-terminal FIKK-kinase domain and the restricted conservation of the N-terminal domain points to a hybrid gene, with individual components evolving at distinct rates. Evolution of the kinase domain may be constrained in order to retain enzymatic activity. In contrast, the high divergence of N-terminal regions among R45-like protein orthologs in distant species might reflect different (species- or host-specific) interaction partners and / or their possible implication in distinct biological processes or distinct stages of the life cycle. The sequence variation between the various *P. falciparum*and *P. reichenowi*paralogs is another sign of the rapid evolution of this gene family. The observation that none of the 20 *P. falciparum*R45-FIKK kinase genes clustered with the \"core\" group of orthologs from the other *Plasmodium*species may be related to this rapid divergence rate. Alternatively, it may merely reflect absence of this family member from the 3D7 genome, which presents several large sub-telomeric deletions and may lack the \"core\" ortholog. The ongoing sequencing of additional *P. falciparum*lines will clarify this issue.
PFD1175w, the trophozoite protein R45 that led us initiate this analysis, turned out to be a quite unusual family member in that it contains a large repeat region absent from all other R45-like proteins. PFD1175w clusters with the orthologous pair PrR45-1 -- PF10\_0380, suggesting that it originates from duplication of the latter after speciation, and has subsequently acquired the repeat region. This is the first evidence of such a process in malaria parasites. PFI0125c, which also contains a specific repeat region, clusters with a different *P. reichenowi*paralog, namely PrR45-6. Both PrR45-1 and PrR45-6 are devoid of repeats. Thus, in both cases, acquisition of repeats was *P. falciparum*-specific and posterior to the *P. reichenowi*/*P. falciparum*branching. Interestingly, the repeats are anchored within a gene region that shows substantial intra-family sequence polymorphism. This gene family thus represents a unique opportunity to analyse the genesis and possible evolution of these intriguing low complexity sequences in *P. falciparum*parasites. Indeed, we have observed considerable variation in the copy number of the PFD1175w hexapeptide repeats in field isolates (Schneider et al, unpublished).
*P. falciparum*and *P. reichenowi*present an expanded set of R45-FIKK kinase genes, which furthermore have unique structural features. The genes have a three exon structure and encode a protein with canonical export signatures (signal / anchor sequence preceding a Pexel motif). This was observed in all twenty *P. falciparum*paralogs and in the *P. reichenowi*paralogs for which full length coding sequence could be retrieved. No such export signature was observed in any of the other orthologs from the family. Presence of a signal sequence associated with a downstream Pexel motif is necessary and sufficient for exportation of the protein beyond the parasitophorous vacuole \[[@B21]\]. This suggests that all 20 *P. falciparum*paralogs have the capacity to be exported beyond the parasitophorous vacuole membrane. The available mRNA expression profiling indicates that all 20 paralogs are transcribed by blood stages and that seven are also transcribed in sporozoites \[[@B16],[@B17]\]. Evidence for protein expression was obtained for some paralogs by mass spectroscopy at the trophozoite and / or sporozoite stage \[[@B25]\]. Absence of protein detection must be interpreted with caution, because it may reflect low abundance of the protein in question at the developmental stage investigated. PFD1175w, the only paralog characterised to date is exported to the red blood cell membrane at the early trophozoite stage \[[@B5],[@B6]\]. Immunisation of *Saimiri*monkeys with a recombinant protein derived from its central repeat region resulted in either complete protection or in selection of parasite mutants that no longer expressed red blood cell surface variant antigens \[[@B6]\]. It is tempting to speculate that the R45 trophozoite antigen is involved in some sort of signalling cascade, and by extrapolation that the other family members are implicated in signalling from the extracellular / extravacuolar milieu as well.
Interestingly, most *P. falciparum*R45-FIKK kinase paralogs are located in the subtelomeric regions that harbour numerous multigene families that code for proteins destined to be exported to the host cellular membrane or host cytoplasm \[[@B1],[@B20],[@B21]\]. The telomeric and subtelomeric regions of *Plasmodium spp*. harbour several multigene families that code for products involved in antigenic variation \[[@B1],[@B2]\]. Remarkably enough, these multigene families differ markedly in *P. falciparum*as compared to *P.y. yoelii*\[[@B2]\]. Nine telomeric / subtelomeric multigene families have been identified in *P.y. yoelii*. Four of them, namely the *yir*or *vir*or the *pystb-d*families have no ortholog in the 3D7 genome, while the others are either expanded or contracted as compared to *P. falciparum*\[[@B2]\]. Similarly, the telomeric *var*, *rif*and *stevor P. falciparum*multigene families do not have orthologs in *P.y. yoelii*\[[@B2]\] or *P. vivax*\[[@B26]\]. Our observation of a markedly different copy number of R45-FIKK kinase genes in *P. falciparum*as compared to *P.y. yoelii*or *P. vivax*is in line with distinct sets of subtelomeric multigene families in different *Plasmodium*species. This is further substantiated by the observation that the *P. falciparum*R45-FIKK kinase genes are situated at or close to subtelomeric breakpoints of synteny between *P. falciparum*and *P. yoelii*\[[@B2]\]. The association of most R45-FIKK kinase telomeric paralogs with a DNA J / *resa*paralog and with additional paralogs from families coding for exported hypothetical membrane proteins or proteins involved in lipid metabolism suggests a structured complex interacting with and signalling from the host cell membrane. It is possible that different subtelomeric gene families execute this function in species such as such as *P. vivax, P. knowlesi*or *P.y. yoelli*that do not possess the specific adhesion properties associated with expression of erythrocyte variant antigens observed in *P. falciparum*.
Conclusion
==========
The new R45-FIKK kinase gene family codes for hybrid proteins with a variable N-terminal domain associated with a conserved, novel protein kinase domain. The presence of most key amino acids for phosphotransfer and the remarkable sequence conservation suggest an enzymatic function. The restricted occurrence of R45-like genes in Apicomplexan parasites and the distance from known kinase families makes a case for consideration of this family as a potential drug or vaccine target. The family shows considerable expansion in *P. falciparum*. All paralogs code for proteins with canonical export signals and most are subtelomeric, clustered with a subset of multigene families coding for exported proteins. This suggests that expansion of this family might contribute to *P. falciparum*virulence.
Methods
=======
Retrieval of homologous sequences in *P. falciparum*
----------------------------------------------------
The deduced protein sequence of PFD1175w was used for a tblastn search against the *P. falciparum*annotated proteins in the PlasmoDB database version 4, and verified for version 4.3 for which genome coverage is 18× <http://plasmodb.org>. (Note that PFD1175w has been annotated as R45 trophozoite-like antigen in the genome sequence, whereas PF10\_0160w, a paralog without repeats was annotated as trophozoite antigen R45 \[[@B1]\]. This is inappropriate, since the only gene that encodes the hexapeptidic repeats originally described for the gene named R45 by Bonnefoy et al \[[@B4]\] is PFD1175w. The PFD1175w repeat region alone did not yield any significant hit apart from PFD1175w either by blasting the whole repeat region or by scanning a single repeat motif against the database. Blasting the PFD1175w protein sequence without repeats, however, resulted in identifying 20 additional annotated proteins with high homology from amino acid 248 up to the end of the ORF, with E-values ranging from 6.8e-109 for PF10\_0160 to 2.6e-28 for PFC0060c, with the next best hit down to 7.3e-05.
The primary structure of the R45-like proteins was verified by comparing and aligning the prediction alternatives made by GlimmerM, Genefinder and FullPhat (PlasmoDB). Sequences were used as annotated in PlasmoDB for the following 8 proteins: PFD1165w, PFD1175w, PFE0045c, PFI0100c, PFI0110c, PFI0120c, PFI0125c, PF10\_0380. For PFI0105c, PFI0115c and MAL13P1.109, preference was given to the alternative sequence prediction chr.9 glm31, chr9.glm33 and chr13phat\_225, respectively, that includes the 3^rd^exon that is highly conserved in all other R45-like protein paralogs. For PF10\_0160, the alternative sequence prediction chr10gen334 was preferred, because the boundary between exon 2 and 3 resulted in a perfect alignment with other family members, whereas the annotated version aligned with gaps. For PFA0130c, PFC0060c, Mal7P1.175, PFI0095c, PF11\_0510 and PFL0040c, preference was given to the chr1.glm\_35, chr3.glm\_19, chr7.glm\_316 and chr9.glm\_29, chr11.glm\_542 and chr12.glm\_14 predictions respectively, because they include a signal sequence to the putative N-terminal domain. For Mal7P1.144, a 5\' exon was predicted manually resulting in a three exon protein with an N-terminal signal sequence. Mal7P1.175 (chr7.glm\_316) was manually curated by fusing the first two exons and considering the intron sequence as coding. This allowed a perfect alignment with its paralogs, although including one internal stop codon. Likewise, the two tandem chromosome 14 genes, annotated PF14\_0733 and PF14\_0734, were fused together, referred to as PF14\_0733+4, resulting in one full R45-like protein interrupted by one stop codon. The identical asexual stage mRNA expression pattern of both genes (DeRisi Lab Malaria Transcriptome Database <http://malaria.ucsf.edu/index.php>) strongly suggests that they indeed form a single transcription unit. The gene names for all paralogs but PF14\_0733+4 were kept as annotated by PlasmoDB, even when an alternative prediction was preferred, resulting in a total of 20 R45-like proteins in the 3D7 *P. falciparum*genome. Signal sequences were predicted using the signalP 3.0 server \[[@B10]\].
Note that MAL13P1.109 and the flanking two genes were no longer retrieved from the PlasmoDB version 4.3 released November 2004. Since the reason for this is unclear, we decided to keep including this gene within the family described here.
Retrieval of R45-like genes from other *Plasmodium*species, *T. gondii*and *C. parvum*
--------------------------------------------------------------------------------------
R45-like protein orthologs from other species were identified using a tblastn search of the protein kinase domain of PFD1175w against the genomic sequences of all *Plasmodium*species in PlasmoDB. This identified six highly matching contigs (E- value \<10 exp 40) as follows: Pg\_c000319933.Contig1 from *P. gallinacaeum*, Pk\_2154b11q1c from *P. knowlesi*, Pv\_4038 from *P. vivax*, chrPyl\_00951 from *P.y. yoelii*, Pb\_5155 from *P. berghei*and Pr\_4e04q1k from *P. reichenowi*. The same tblastn search was also repeated in the genomic databases from the Wellcome Trust Sanger Institute <http://www.sanger.ac.uk/Projects> for *P. berghei*, *P. knowlesi, P. gallinacaeum and P. reichenowi*sequences and the Institute for Genomic Research <http://www.tigr.org/tdb> for *P. vivax*and *P. yoelii*sequences. Contigs with the identical sequence information as those obtained from PlasmoDB were obtained for five species, however six highly matching hits were retrieved in the *P. reichenowi*genomic database. No hits were obtained by blasting against other protozoan parasites in the above mentioned resources. The *T. gondii*contig TGG994720 was retrieved from the Toxoplasma Database \[ToxoDB, release 3.0, <http://toxodb.org/>\] and the *C. parvum*contig AAEE01000010.1 was retrieved from the National Center for Biotechnology Information <http://www.ncbi.nlm.nih.gov>. BlastP analysis of PFD1175w against the NRprot database did not give a hit in any additional species.
R45-like orthologs were predicted as follows: the *C. parvum*ortholog annotated as EAK87586.1 \[[@B3]\] was named CpR45. The gene model proposed for the *P.y. yoelii*ortholog \[[@B2]\] was modified. The two genes annotated as Py03325 and Py03326 which are separated because of a one nucleotide frameshift were fused to one gene and named PyR45. The resulting two exon structure with most of the sequence encoded by exon 1 and the conserved exon 2 (corresponding to exon 3 in the *P. falciparum*orthologs) is supported by the EST data. Two exon orthologs in *P. vivax*, *P. knowlesi*, *P. berghei, P. gallinacaeum*(PvR45, PkR45, PbR45 and PgR45) were predicted accordingly.
Six R45-like proteins were identified in *P. reichenowi*(PrR45-1 to 6) from contigs 3502696.c000125078, c000024654, c000125213, c000126950, c000124557, c000128208, respectively. Their very close homology to specific orthologs in one or the other species was exploited to predict their exon-intron boundaries. The *T. gondii*R45-like protein, TgR45 was predicted using the overlapping ESTs TgEST\_95056840, TgEST\_100112765, TgEST\_95056367. The N-terminal exon(s) are not yet identified.
The databases were last checked for the presence of new hits by December 9, 2004. By then, the genome of *P.y. yoelii*was as published in 2002, namely with 5× coverage and short contigs \[[@B2]\], the genome of *C. parvum*was published with 13× coverage and 1--3 large contigs per chromosome \[[@B3]\]. Genome sequence of the other species was in progress, with mostly short contigs and the following coverage rates: *T. gondii*(10×), *P. vivax*(10×), *P. knowlesi*(5×)*, P. berghei*(3×), *P. gallinacaeum*(3×)*and P. reichenowi*(1.5×).
Analysis of chromosomal organisation in *P. falciparum*and possible synteny with other species
----------------------------------------------------------------------------------------------
Analysis of the genes localised within 40 kb on either side of the various R45-like paralogs was done using the published chromosomal maps \[[@B1]\] and the PlasmoDB version 4.3 annotation. Ortholog groups were systematically looked for each gene, and sequence alignments inspected. For each family, only paralogs with high homology along \>60% of the sequence were considered. In case a *P.y. yoelli*ortholog was present, the neighbouring genes were analysed for possible synteny groups with *P. falciparum*. The same was done for the *P. vivax*genes located within contig Pv\_4038.
Sequence alignment and phylogenetic tree
----------------------------------------
The conserved regions of all R45-like proteins were aligned by ClustalW \[[@B11]\] using the matrix PAM30 and adjusted manually where necessary. Four well characterised catalytic domains of protein kinases, namely protein kinase A (PkA), Jun-kinase 1 (Junk1), fibroblast growth factor receptor 1 (Figure [1](#F1){ref-type="fig"}) and extracellular regulated kinase 1 (ERK1) were used as a scaffold for the alignment of R45-like proteins to the 12 subdomains as provided in <http://pkr.sdsc.edu/html/classification.shtml>\[[@B12]\]. The *P. falciparum*protein kinase 5 (PfPk5), the only *Plasmodium*protein kinase which has been thus far structurally characterised, was included as an additional reference protein kinase \[[@B13]\]. For the phylogenetic tree, the kinase domains of 32 R45-FIKK proteins were taken and gaps between the subdomains were removed. Phylogenetic relationships were inferred by the neighbour joining method \[[@B14]\] with the *T. gondii*protein TgR45 as outgroup. The reliability of the trees was assessed by the bootstrap method \[[@B15]\].
Authors\' contributions
=======================
AS performed the database analysis for R45 paralogs and orthologs in the various species, reannotated the genes, performed the alignments and constructed the phylogenetic tree. OP searched for PEXEL export motifs, performed database mining for chromosomal localisation, analysed the conservation of the multigene families located in the vicinity of R45-FIKK like *P. falciparum*paralogs, generated the chromosomal map and investigated possible synteny in other species. OP and AS carried out data mining for expression profiles throughout the life cycle. AS and OP drafted the text of the manuscript. OP finalised the revised version.
Supplementary Material
======================
::: {.caption}
###### Additional File 1
Multiple sequence alignment of the kinase domain from all R45-FIKK kinase proteins identified.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 2
Multiple sequence alignment of the N-terminal region of R45-like proteins in *P. falciparum*and its orthologs in *P. reichenowi*.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 3
Multiple sequence alignment of the N-terminal region of R45-FIKK kinase orthologs from six *Plasmodium*species.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
We made use of the PlasmoDB database and wish to acknowledge our debt to the colleagues in charge of this database. We thank F. Tekaia (Institut Pasteur) for helpful advice in phylogenetic analysis, to S. Longacre (Institut Pasteur) for critical reading of the manuscript and to R. Stuart (Institut Pasteur) for interesting discussions. A. Schneider and this work was supported by the EU commission funded programme EUROMALVAC2 (QLK2-CT-2002-01197). We are grateful to C. Doerig and his colleagues as well as to A. Cowman and M. Marti for kindly communicating results before publication. We thank the reviewers of the first draft of this article for constructive remarks.
|
PubMed Central
|
2024-06-05T03:55:55.233630
|
2005-3-7
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079819/",
"journal": "BMC Genomics. 2005 Mar 7; 6:30",
"authors": [
{
"first": "Achim G",
"last": "Schneider"
},
{
"first": "Odile",
"last": "Mercereau-Puijalon"
}
]
}
|
PMC1079820
|
Background
==========
Large collections of ESTs enable the assembly of nucleotide sequence contigs and, if the genomic sequence is available, a means of mapping the species genome and ultimately assist in gene and pathway discovery. Multivariate statistical analysis of these collections is used for microarray design. The main feature of the Marine Genomics project is that all data is accessible to the public (non-curator viewers) as a curated clearing house for genomic and transcriptomic data of marine organisms. Accordingly, Marine Genomics includes a tool for automated EST submission to NCBI\'s GenBank to assist in integrating data and annotation results with a wider public resource. One of the primary goals of the clearing-house is to ensure the successful submission to NCBI, of all processed and curated data contained in the Marine Genomics databases. New species databases are welcomed in order to build a comprehensive marine repository. The conglomeration of multiple marine organisms with integrated analysis tools enables focusing on the comprehensive descriptions of transcriptomic responses to typical marine stresses such as water pollution and algal blooms, effects of climate change such as altered pH and increase in carbon dioxide levels \[[@B1],[@B2]\], as well as localized phenomena such as coral bleaching \[[@B3]\] and viral infections (crustacean and fisheries diseases \[[@B4]\]).
Construction and content
========================
Methodology
-----------
The marine genomics pipeline is a web-based software environment. The open-source public license for the scripting language PHP 4.2.2 is used at the front end for user interface development and runs on Apache web server 1.3.33. MATLAB version 6.5 <http://www.mathworks.com/products/matlab/>, a scientific engineering computational language, release 13 is employed for statistical data analysis and some of the more intensive computational processes. The open source database application PostgreSQL, version 7.2.3 <http://www.postgresql.org>, is used for all data storage. The entire system is developed and run on servers configured with the open source operating system Red Hat Linux version 7.3 <http://www.redhat.com>.
Site divisions and support functions
------------------------------------
The site functionally divides in two main upload functions for both EST and microarray data upload. These upload tools enable users to add EST and microarray data to their species PostgreSQL database where they are accessed and manipulated by various processing tools. The microarray upload also allows users to warehouse microarray MIAME and MAGE compliant data. Other site support functions include an EST annotative text search, a stand-alone BLAST \[[@B5]\] function as well as user authentication and curation control.
Species databases
-----------------
Marine Genomics currently contains 19 different marine species databases. Species currently included are: *Anas platyrhynchos*(mallard), *Crassostrea gigas*(Pacific oyster), *Callinectes sapidus*(blue crab), *Crassostrea virginica*(eastern oyster), *Eubalaena glacialis*(Northern Atlantic right whale), *Fundulus*species (killifish), *Homarus americanus*(American Atlantic lobster), *Karenia brevis*(red tide algae), *Leucoraja erinacea*(little skate), *Litopenaeus setiferus*(white shrimp), *Litopenaeus stylirostris*(blue shrimp), *Litopenaeus vannamei*(white shrimp), *Montastraea annularis*(lobed star coral), *Oculina varicosa*(stony coral), *Porites porites*(clubbed finger coral), *Palaemonetes pugio*(daggerblade grass shrimp), *Squalus acanthias*(spiny dogfish) and *Tursiops truncatus*(bottlenose dolphin). Each of these species databases undergoes a cross-BLAST for sequence similarities. New databases are added regularly upon user request.
Database Updates
----------------
In order to maintain the most current BLAST results both from GenBank and the internal Marine Genomics database, the databases will be BLAST on a quarterly basis.
Utility
=======
EST pipeline implementation
---------------------------
### EST Upload
Users can reference the Marine Genomics Process flow guide on the homepage of the website to get an overview of Marine Genomics EST and microarray processes. Currently the pipeline accepts both FASTA and text sequence files as well as electropherogram trace files. The user interface allows the user to upload sequences both as zipped batches and as individual uploads. The phred and the phd2fasta programs \[[@B6],[@B7]\] are used for converting the trace file into a readable text format. The files are then stored for back-up. Once submitted to the pipeline each sequence undergoes a number of QA/QC procedures and subsequently becomes available for curation and user initiated submission to NCBI\'s GenBank.
### EST quality control and sequence processing (QA/QC)
1\. Cross-match \[[@B6]\] is employed to mask vector content from the uploaded sequence files. Then this masked vector is automatically removed by a Marine Genomics trimming tool.
2\. The collars (user specified regions of the vector adapters) chosen by the user on file upload are used for a final vector screening in an attempt to ensure all vector is removed from the sequence before submission to the species databases. It allows the user some control in specifying the end of the vector sequence and thus adds an extra layer of vector screening and removal.
3\. Poly-A tail removal.
4\. Size control: Sequences shorter than 50 nucleotide bases are flagged.
5\. N-content control: Flagging of sequences with an N-content of greater than 3 bases in 10.
6\. Flagging of non-DNA sequences to prevent any possible file upload contamination.
### EST curation and submission to NCBI
1\. Sequence viewing: Public viewers have access to the sequence, BLAST and processing results such as when the sequence was updated and last modified as well as tissue and NCBI accession number.
2\. Sequence curation: Curators can review/edit trimmed ESTs as well as delete and submit the individual sequence to GenBank.
3\. Automated BLAST: Sequences automatically undergo a BLAST against NCBI\'s GenBank using the BLAST tool. GenBank databases currently used include the GenBank non-redundant CDS translations protein databases (nr) with BLASTx, the GenBank non-redundant nucleotide databases (nt) with BLASTn and the GenBank EST databases with BLASTn (dbEST). The sequences also undergo BLAST against all local Marine Genomics databases.
4\. Interactive sequence submission to GenBank: Once each curator-uploaded sequence batch has been appropriately run through the QA/QC process, the curator receives an automated email allowing them to submit their ESTs to GenBank\'s dbEST database. The email also lists which sequences will not be included in the submission due to problem-flagging. The curator has the option of reviewing sequences that may have been left out of the submission process, such that they can be edited appropriately and then later submitted.
5\. Contigs are assembled using CAP3 program \[[@B8]\] after ESTs have been curated. This enables the users to determine unique transcripts/genes. Currently curators can request the addition of this functionality for their particular species and the contig data is made web accessible then.
6\. Sequences are annotated with Gene Ontology terms by searching the sequences against the Gene Ontology database \[[@B9]\] using BLAST \[[@B10]\]. Species-specific Gene Ontologies (GO) are currently reported in a piechart on each species entry page and also within each EST sequence page. Also a cross species GO data summary is reported along side the listing of all current Marine Genomics species. The addition of GO annotation is currently on going.
Microarray pipeline implementation
----------------------------------
The microarray pipeline is a more recent addition to Marine Genomics than the EST pipeline and is still undergoing development particularly in the buildup of analysis tools. The MATLAB statistics and bioinformatic libraries are of critical importance for the fast development and deployment of advanced analysis procedures. Accordingly, a suite of multivariate statistical analysis tools, developed in MATLAB is being employed to assist in microarray design. An optimal cDNA microarray probe selection algorithm combining different clustering methods and contig information was implemented to assist microarray design \[[@B10]\]. The procedure can also be used for multiple species microarray design which critically benefits from the marine-specific nature of the EST databases maintained. EST probe selection for microarray design is available upon request and the clustering and probe selection output is available on the website on each species entry page.
A MIAME compliant \[[@B11]\] excel template is provided for user download <http://www.marinegenomics.org/~mckilldj/MIAME.xls> to ensure the data remains compliant. This file can be filled-in and exported from Microsoft Excel as a tab-delimited MIAME data file for upload alongside the corresponding MAGE data file. Currently Marine Genomics accepts MAGE data from in-house microarray experiments run at the Hollings Marine Laboratory in Charleston, South Carolina for storage and analysis.
### Microarray data upload and warehousing of experimental results data
1\. Data upload: The microarray pipeline accepts a text file (MIAME compliant) which contains information specific to the experimental information (lab notes etc.) It also accepts a text data file, output from the microarray laser scanner, (containing specific spot locations and intensities) which is parsed into MAGE compliant data for warehousing.
2\. Data warehousing: The uploaded tab-delimited files are parsed and stored in a relational PostgreSQL database. Currently warehoused microarray data is accessible to the public through specific species links.
3\. Data download: Microarray data can be accessed and downloaded as Web pages or as Excel compatible files. Marine Genomics also includes a MATLAB centric microarray access feature consisting of an m-file, mgma\_get.m. This function will automatically list all available microarray data and download complete microarray content directly into the user\'s MATLAB workspace. This makes an entire microarray dataset or selection of datasets available in a MATLAB Bioinformatics Toolbox specific format for analysis by the functions within the Bioinformatic Toolbox (e.g. maimage(mgma\_get(15), Ch1 Intensity)). This m-file is made freely available at: <http://www.marinegenomics.org/mgma_get.m>
Conclusion
==========
The Marine Genomics infrastructure was developed as a clearing house of functional genomics data for marine organisms. It currently includes tools to upload, preprocess, cross-reference, annotate, NCBI submit and store EST data. It also includes a corresponding microarray design, upload and storage tool with development of analysis tools underway.
The usage of the Marine Genomics infrastructure has been speedily increasing with more species databases being added and the numbers of sequences increasing even faster. Furthermore, Marine Genomics integrates the microarray entries with the MATLAB environment for which there are several commercial and public libraries (\"toolboxes\") for microarray analysis.
Current development goals include the added functionality of continued addition of ontology and contig information for all species. Another goal is to add the ability to parse and process multiple microarray platforms such that users can have the flexibility of uploading data output from their own individual microarray platforms. Finally, in particular special care will be given to speedily exporting expression data as MAGE-XML for incorporation in NCBI\'s GEO databases rather than having that data exclusively retained in Marine Genomics.
The ultimate purpose of Marine Genomics is indeed to assist in submitting quality data to the NCBI GenBank and GEO databases. For that purpose, the Marine Genomics pipeline and tools have been assembled to provide a medium for working with functional genomics in a marine biology environment.
Availability and contacts
-------------------------
The species databases are available for public viewing and new species additions can be made via the website (Fig. [1](#F1){ref-type="fig"}). The website URL is <http://www.marinegenomics.org>. Each species database is available for download in the form of a fasta file from each of the species data pages. Contact: David J. McKillen at <mckilldj@musc.edu>.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Marine Genomics homepage
:::
:::
Authors\' contributions
=======================
DJM is the lead developer on the pipeline construction and maintenance, YAC develops the bioinformatic algorithms, CC helps with server management, MJJ and JR advise on the biological aspects of pipeline development, HFT assists in the development of the pipeline, DCM co-developed the microarray database model, PSG and RWC are responsible for project overview and management, GWW and JSA proposed the underlying conceptual model and oversee the development. All authors read and approved the final manuscript.
Acknowledgements
================
This work was partially supported by the National Science Foundation (EPS0083102 & MCB0315393), the South Carolina Sea Grant Consortium (R/MT-6), and the South Carolina Department of Natural Resources. This is publication \#20 from the Marine Biomedicine and Environmental Sciences at the Medical University of South Carolina. YAC acknowledges the support by the South Carolina Sea Grant (NA16RG2250)
|
PubMed Central
|
2024-06-05T03:55:55.236749
|
2005-3-10
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079820/",
"journal": "BMC Genomics. 2005 Mar 10; 6:34",
"authors": [
{
"first": "David J",
"last": "McKillen"
},
{
"first": "Yian A",
"last": "Chen"
},
{
"first": "Chuming",
"last": "Chen"
},
{
"first": "Matthew J",
"last": "Jenny"
},
{
"first": "Harold F",
"last": "Trent"
},
{
"first": "Javier",
"last": "Robalino"
},
{
"first": "David C",
"last": "McLean"
},
{
"first": "Paul S",
"last": "Gross"
},
{
"first": "Robert W",
"last": "Chapman"
},
{
"first": "Gregory W",
"last": "Warr"
},
{
"first": "Jonas S",
"last": "Almeida"
}
]
}
|
PMC1079821
|
Background
==========
The use of cDNA and oligonucleotide microarray technology has revolutionized the fields of molecular biology, biochemistry and genetics. The ability to simultaneously evaluate gene expression across tens of thousands of genes gives researchers opportunities not previously afforded to them.
RNA extractions have proven to be of large concern for evaluating messenger RNA transcript levels by microarrays and other procedures such as RT-PCR, RNase protection assays and Northern blot analyses. Extraction procedures are still evolving and adapt to meet different needs, such as for pure cell populations \[[@B1]\]. Differences between two RNA extractions from the same source material have been shown to make a significant contribution to technical variance in microarray data \[[@B2]\].
Microarray technology utilizes various protocols based in part on reverse transcription and PCR technologies \[[@B3],[@B4]\]. Direct labeling protocols use modified deoxyribonucleotide phosphates (dNTPs) incorporated during a reverse transcription reaction, in which mRNA is copied into cDNA. One possible modification to the dNTPs is the addition of an amino-propagyl cyanine (Cy) fluorescent molecule at the 5-carbon of the pyrimidine base \[[@B5],[@B6]\]. For cDNA and oligonucleotide microarrays, Cy3 and Cy5 are commonly used fluorescent dyes that are excited by different wavelengths of light. Therefore they can be used in combination, one labeling a control or reference sample and the other labeling the treatment or test sample. After combining the two labeled cDNA samples and hybridizing to a microarray chip, gene expression can be extrapolated from the ratio of the two different cyanine dye fluorescences detected. The entire microarray process, especially the reverse transcription labeling procedure, is dependent upon the quality and integrity of the starting RNA.
The Agilent 2100 Bioanalyzer, first described for the investigation of DNA, employs microfluidic technology for the analysis of nucleic acids and proteins \[[@B7]\]. A total RNA assay determines a numerical value for the ratio of 28S ribosomal RNA (rRNA) subunit to 18S rRNA subunit, while an mRNA assay determines the percentage of rRNA contamination. To measure ribosomal subunit concentrations, RNA is combined with a sample buffer containing a fluorescent dye that intercalates into the RNA and is excited by an internal 635 nm diode laser. Data output is in the form of an electropherogram, which graphically depicts spikes in fluorescence over time; the larger the peak, the more intact ribosomal subunits are in the sample. If a sample is degraded, subunits will show a smaller degree of fluorescence that is spread out over a longer amount of time, indicating a greater variety of sizes in the sample. Auer and colleagues recently published the \"degradation factor\" utilizing data obtained using the RNA assay of the Bioanalyzer as a more quantitative approach \[[@B8]\]. This approach calculates a ratio between the 18S ribosomal peak area and the average of the peaks smaller than the 18S ribosomal peak that are indicative of degradation. The authors show that RNA isolations which exhibit similar degradation factors are more likely to give gene expression results that are more biologically relevant than comparing two RNA isolations with differing degrees of degradation.
The quality and integrity of RNA samples can be evaluated by gel electrophoresis, UV spectrophotometry and the Agilent Bioanalyzer. From one or a combination of these methods, assessments of how well an RNA sample may perform in a reverse transcription labeling reaction and subsequent microarray chip hybridization can be made. Unfortunately, the cDNA reverse-transcribed from an RNA of reasonable quality may not yield reliable products after the labeling steps and lead to poor hybridization results. Our lab has experienced this problem as a result of, among others, possible genomic DNA contamination of the RNA sample. This problem with the RNA is difficult to detect with an RNA assay on the Bioanalyzer, and the contamination will likely impact the quality of any labeling reaction the sample is used in, regardless of the fluorophore.
Agilent currently promotes the use of their instrument for measuring Cy-labeled cRNA obtained from an amplification protocol. They do not, however, currently have a protocol to measure the quality of Cy-labeled cDNA from a direct labeling protocol. To address and avoid the loss of time and money associated with a failed microarray analysis, we present a novel use for the Agilent 2100 Bioanalyzer in determining the relative quantity and quality of direct-labeled products obtained during the labeling reaction. By comparing with cDNA obtained from known high quality RNA, we can determine how well the cyanine dye was directly incorporated during reverse transcription, and thus if reliable microarray data can be obtained.
Results and discussion
======================
Using an Agilent Bioanalyzer, typical indications of high quality, intact total RNA samples are electropherograms with flat baselines and a relatively flat valley between the two strong fluorescent rRNA peaks (Figure [1A](#F1){ref-type="fig"}, No RNase). An important note is that within the context of a high quality total RNA sample, the mRNA fluorescence is below detection. mRNA normally accounts for only 1--5% of a total RNA sample \[[@B9]\] and the rRNA peaks dominate the fluorescence of these samples. The undegraded total RNA sample shows distinct 18S and 28S rRNA subunit spikes, with a ratio of 1.85 (\[28S\]: \[18S\]). This ratio, along with the contours of the electropherogram, led to the conclusion that this was an undegraded, high quality RNA sample. Treatment with 1 ng ml-1 RNase degrades essentially all RNA present, as evidenced by its lack of 18S and 28S rRNA peaks (Figure [1D](#F1){ref-type="fig"}, [1](#F1){ref-type="fig"} ng ml-1 RNase). Concentrations of RNase less than 1 ng ml-1 only partially degrade the RNA samples since 18S and 28S peaks are detectable but less than optimal (compare Figure [1A](#F1){ref-type="fig"} with [1B](#F1){ref-type="fig"} and [1C](#F1){ref-type="fig"}, No RNase with 100 ng ml-1 and 10 ng ml-1 RNase, respectively).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**RNA and Cy5-dUTP labeled cDNA measured by the Agilent 2100 Bioanalyzer.**Electropherogram images (one representative experiment) of RNAs treated with RNase as follows: No RNase (A, pink), 0.01 ng ml-1 (B, blue), 0.1 ng ml-1 (C, red), 1 ng ml-1 (D, green). Overlayed electropherogram image of Cy5-dUTP labeled cDNA (E, colors as described above, one representative experiment). The first peak (as shown by an arrow) from the left in all electropherograms is a 50 base pair marker present in the sample buffer (M). The second peak (as shown by an arrow) that is present only in Cy5 samples is free Cy5-dUTP (Cy5) and measures approximately 150 nucleotides (data not shown). Though it is a single nucleotide, the Cy5 modification is large, which could explain its delayed migration and detection.
:::
:::
Fully degraded, partially degraded and intact RNA, generated by titrating varying amounts of RNase, were used to prepare samples for measuring Cy5 incorporation directly into the cDNA by reverse transcription of RNA by the Agilent 2100 Bioanalyzer. It should be noted, however, that RNase treatment can be unpredictable, since even slight deviations in temperature and timing of incubation can result in varying degrees of degradation. Over triplicate experiments, individual treatments could vary over 100% from day to day, however the overall trends do not change.
Cy5-dUTP was directly incorporated by reverse transcription into the intact (No RNase) sample as well as the three RNase treated samples. Since the RNA is hydrolyzed after reverse transcription, only the cDNA detectable. Analysis of the Cy5 signal obtained from an intact RNA sample (Figure [1E](#F1){ref-type="fig"}, pink trace), as measured by the Bioanalyzer, reveals a distribution of different transcript sizes. Along with a wide distribution of sizes, the overall fluorescence is much higher for the undegraded sample as compared to the RNase-treated samples (blue, red and green traces for 0.01, 0.1 and 1 ng ml-1 RNase, respectively, Figure [1E](#F1){ref-type="fig"}). The Cy5-labeled cDNA concentration, calculated by the Bioanalyzer assay as the area under the curve in Figure [1E](#F1){ref-type="fig"}, decreases approximately 65% after treatment with 1 ng ml-1 RNase and 30% with 0.1 ng ml-1 RNase. Treatment with 0.01 ng ml-1 RNase also depleted Cy5-labeled cDNA concentration, though only by about 25%. These Bioanalyzer results show considerable reductions in Cy5-labeled cDNA signal obtained, due to the degraded nature of the RNA samples.
Gel electrophoresis, such as small scale poly-acrylamide analysis \[[@B10]\], has been used to determine characteristics of Cy5 incorporation. As shown in Figure [2](#F2){ref-type="fig"}, the phosphorimager scan of an agarose gel and the \"gel-like image\" obtained from the Bioanalyzer visually depict common trends. However, performing the gel electrophoresis was more sample- and time-consuming. Furthermore, quantitation was more subjective since measurement areas must be user-defined. The coefficients of variation, though intrinsically high because of the RNase treatment itself, were higher for the gel electrophoresis analysis.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Comparison of Bioanalyzer assay with conventional gel analysis.**A- cDNAs were separated based on size using a 1% agarose gel. Red fluorescence (650 long pass emission filter) was measured using a Storm phosphorimager and quantitated using IMAGEQuant. B- \"Gel-like\" images obtained from the Bioanalyzer are digital images rendered from conversion of electropherogram fluorescence traces using the Bioanalyzer software. Both images are the same sample from one representative experiment ran in parallel.
:::
:::
Further validation of this assay came from analyzing RNA samples that had previously been labeled and hybridized, but provided poor quality expression data. These RNAs, based on Bioanalyzer electropherograms similar to that of the intact RNA in Figure [1A](#F1){ref-type="fig"}, appeared to be of high quality with 18S:28S peak ratios in excess of 1.8. To test the new assay, the labeling reactions were performed again and the labeled products produced were assessed on the Bioanalyzer and hybridized to the NIEHS Human ToxChip.
As shown in Figure [3A](#F3){ref-type="fig"}, each of these test samples show a marked decrease in the total fluorescence measured by the Bioanalyzer as compared to the control samples, which were derived from RNA that had previously performed well in microarray labeling and hybridization procedures. Additionally, comparison of average array intensity for each of the test samples is dramatically reduced when compared with arrays hybridized with product from MCF7 intact RNAs (Figure [3B](#F3){ref-type="fig"}). We believe that there was possible genomic DNA contamination, as evidenced by non-migratory nucleotides in the wells of a formaldehyde gel (data not shown) It has been implicated that DNA contamination causes a decrease in efficiency of the labeling procedure \[[@B11]\], though other contaminants inherent to RNA extraction, such as phenol, or inherent to the RNA sample (high lipid or protein content) may also interfere. Though first thought to be of ideal quality for microarray, these human lung fibroblast test samples have been shown through the application of this novel assay to be less than optimal for use on a microarray chip. Had this assay been used prior to hybridization in previous experiments, the money and time spent in putting these direct labeled products on a chip would have been saved.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Practical application of the Cy5-cDNA Bioanalyzer assay. (A)**Overlayed electropherogram image of Cy5-dUTP labeled cDNA obtained from control RNAs (MCF7 and HACAT, red and blue traces, respectively) and from test samples (Test \#1 and Test \#2, brown and purple traces respectively). First and second sharp peaks are as described in legend for Figure 1. **(B)**Average microarray intensity for MCF7, Test 1 and Test 2. Error bar is standard error from three arrays. Test samples were hybridized to only one microarray each.
:::
:::
Conclusion
==========
From this novel Bioanalyzer assay, we are able to determine if a sample of questionable integrity is \"chip worthy\" or not. Selection criteria are subjective since the quality requirement will differ for different platforms. Determining the integrity of a sample could save an investigator a significant amount of capital resources (compare the tens of dollars that a Bioanalyzer chip costs to the hundreds of dollars in expenses for a microarray hybridization). The Agilent 2100 Bioanalyzer provides a cost-efficient way to determine an RNA, and the subsequent cDNA, sample\'s ability to provide high quality data (wide dynamic range) for a microarray chip. Though not meant as an absolute quantitative tool, trends in labeling efficiency can be assessed using this novel method. This assay can be used in place of conventional gel electrophoresis because it is more time and sample efficient, and it is considerably more cost effective than microarray hybridizations. In the future, we plan to use this assay to as a screening tool for direct labeled products prior to hybridization.
Methods
=======
Standard protocols can be accessed at <http://dir.niehs.nih.gov/microarray/methods.htm>.
RNA preparation and degradation
-------------------------------
Total RNA was prepared from logarithmically growing immortalized human mammary cells (MCF-7) using the RNeasy kit (Qiagen, Valencia, CA, USA) according to the manufacturer\'s instructions. RNA concentration and purity were estimated spectrophotometrically and the quality further assessed with the Agilent 2100 Bioanalyzer (Palo Alto, CA, USA). This intact RNA was split into aliquots and treated with varying concentrations of RNase A (0, 1, 0.1 and 0.01 ng ml-1). After incubation at 37°C for 10 minutes, RNA was analyzed for post-treatment quality assessment using the Bioanalyzer. The samples were loaded at 100 ng per well based on the original concentration values obtained by the spectrophotometer. The Agilent 2100 Bioanalyzer uses a 635 nm diode laser with an emission filter of 670--700 nm to detect fluorescence after dye intercalation into nucleic acids.
Total RNA was also prepared from human keratinocytes (Hacat) and human lung fibroblasts (16Lu) (American Type Culture Collection, Manassas, VA, USA) using the RNeasy kit (Qiagen, Valencia, CA, USA) according to the manufacturer\'s instructions. RNA concentration and purity were estimated spectrophotometrically and the quality further assessed with the Agilent 2100 Bioanalyzer (Palo Alto, CA, USA).
Direct Cy5-dUTP incorporation for RNase-treated samples
-------------------------------------------------------
Each total RNA sample (35 μg, post-RNase treatment) was labeled with direct incorporation of Cyanine 5 (Cy5)-conjugated dUTP (Amersham, Piscataway, NJ) by a reverse transcription reaction using the reverse transcriptase, SuperScript II (Invitrogen, Carlsbad, CA), and an oligo dT primer (Amersham, Piscataway, NJ). After reverse transcription, RNA was degraded by alkaline hydrolysis and excess dNTPs and enzymes from the reverse transcription reaction were removed using a modified version of Qiagen\'s QIAquick PCR Purification procedure (addition of a 35% guanidine HCl wash). This additional wash step occurs before the PE Buffer wash and uses the standard 750 μl volume for these columns. Following column washes, bound cDNA was eluted according to manufacturer\'s instructions.
Direct Cy-dUTP incorporation for assay validation samples
---------------------------------------------------------
Direct labelings (25 μg total RNA) by Cy3 and Cy5 were performed with two of the 16Lu RNAs being labeled with Cy3 and another two being labeled with Cy5. Additionally, MCF7-derived RNA as well as Hacat-derived RNA, both known to be of high quality, were labeled once per dye and are considered as controls. After the reverse transcription reactions were carried out, each Cy3/Cy5 pair was mixed together, cleaned and eluted using the Qiagen PCR Purification kit with modifications, as described above.
Cy5 detection by the Agilent 2100 Bioanalyzer
---------------------------------------------
For comparing the RNase-treated samples, 1 μl of the eluate was mixed with 5 μl 25% Sample Buffer (200 mM TAPS, 1 mM EDTA, pH 8.0) from the Agilent RNA 6000 Nano Kit, loaded onto an RNA LabChip (Agilent Technologies, Wilmington, DE) and detected with the Eukaryote Total RNA Nano Assay. Size distribution, concentration and overall amount of fluorescence were used for evaluation of dye incorporation. The detected fluorescence is the total combination of the intercalating dye and Cyanine 5. Comparisons are made between each of the RNase treated samples and the untreated control.
For the human lung fibroblast test samples, the Cy3 and Cy5 samples were mixed together in preparing for the hybridizations. We previously observed a lack of significant contribution to the overall fluorescence on the Bioanalyzer by Cy3-labeled samples (data not shown). Cy3 does not fluoresce in the range of the Bioanalyzer laser and any contribution to the outcome of the test would be minimal and yet consistent across all samples. One μl aliquots were analyzed as described above. Quality assessment was achieved by comparing the curves of the test samples to those of control samples.
Gel electrophoresis
-------------------
Five μl of each eluate was loaded onto a 1% agarose gel. The loading dye used in this experiment (25% glycerol, 0.1 M EDTA, 0.03% Bromophenol Blue) does not autofluoresce when measuring the red fluorescence and therefore did not interfere with measurement of the Cy5-labeled cDNA (data not shown). Red fluorescence (650 nm long pass emission filter) was detected using a Storm phosphorimager and quantified with IMAGEQuant (Amersham Biosciences, Piscataway, NJ).
Microarray hybridizations and analyses
--------------------------------------
Fluorescently labeled test sample and MCF7 cDNAs were mixed with an SSC/SDS hybridization solution and hybridized to the microarray overnight in a 65°C water bath. The MCF7 samples were hybridized on different days than the test samples. A cDNA Human ToxChip \[[@B12]\], developed in-house at NIEHS, was used for hybridization experiments. A complete listing of the genes on this chip is available at the following website: <http://dir.niehs.nih.gov/microarray/chips.htm>. cDNA microarray chips were prepared as previously described \[[@B4],[@B13]\].
The cDNA microarray chips were scanned with an Agilent Scanner (Agilent Technologies, Wilmington, DE) using laser excitation at 635 and 100% PMT sensitivity. The Agilent scanner is useful for these comparisons because of its ability to control laser variability, allowing data to be compared directly from day to day. The raw pixel intensity images were analyzed using the ArraySuite v2.0 extensions of the IPLab image processing software package (Scanalytics, Fairfax, VA). This program uses methods that were developed and previously described \[[@B14]\] to locate targets on the array, measure local background for each target and subtract it from the target intensity value. Data were calculated by taking the median of spot intensities over each chip
List of abbreviations used
==========================
RNA Ribonucleic Acid
rRNA Ribosomal Ribonucleic Acid
mRNA Messenger Ribonucleic Acid
cDNA Complementary Deoxyribonucleic Acid
PCR Polymerase Chain Reaction
RT-PCR Reverse Transcription -- Polymerase Chain Reaction
dNTPs deoxy-Nucleotide TriPhosphates
Authors\' contributions
=======================
SFG carried out the microarray analysis, performed most experiments, and wrote the paper
EKL provided scientific advice, performed traditional gel assay and recommended additional experiments
CJT conceived of the study, and participated in its design and coordination. All authors read and approved the final manuscript.
Acknowledgements
================
We would like to thank Dr. Douglas Bell and Mr. Jeffrey Reece for their critical reviews of this manuscript.
|
PubMed Central
|
2024-06-05T03:55:55.238162
|
2005-3-11
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079821/",
"journal": "BMC Genomics. 2005 Mar 11; 6:36",
"authors": [
{
"first": "Sherry F",
"last": "Grissom"
},
{
"first": "Edward K",
"last": "Lobenhofer"
},
{
"first": "Charles J",
"last": "Tucker"
}
]
}
|
PMC1079822
|
Background
==========
The nervous system is assembled from disparate classes of neurons that together define the overall properties of the network. The specific functions of these neurons are governed by genetic programs that control cell fate \[[@B1]\]. Thus, a key to understanding the molecular basis for neural function is to establish the gene expression blueprint that orchestrates neuronal differentiation. With its simple, well-defined nervous system and powerful genetics, the nematode *C. elegans*is a useful model system for addressing this issue. The *C. elegans*hermaphrodite nervous system is composed of exactly 302 neurons. The morphology and connectivity of each one of these neurons has been defined at high resolution \[[@B2]\]. In addition, the birth of each neuroblast is embedded in a lineage diagram of every cell division in *C. elegans*development \[[@B3],[@B4]\]. The *C. elegans*genome is fully sequenced and contains \~20,000 predicted genes \[[@B5]\]. At a fundamental level, the identity of a given class of neuron is defined by a unique combination of these genes. In principle, microarray-based strategies could be employed to establish these cell-specific patterns of gene expression. However, the small size of the nematode has limited access to individual cells for molecular analysis. Here we describe a strategy, MAPCeL (Micro-Array Profiling of *C. elegans*Cells) that overcomes these obstacles to generate neuron-specific gene expression profiles.
MAPCeL exploits recently developed methods of culturing *C. elegans*embryonic cells. GFP markers for specific classes of neurons and muscle cells are expressed *in vitro*and can be used to identify the corresponding differentiated cell types. We established that these GFP cells arise at a frequency predicted by their abundance in the intact embryo and display normal morphological, molecular, and physiological characteristics \[[@B6]\]. For example, a GFP reporter for the *unc-4*homeodomain transcription factor gene is expressed in 13 motor neurons out of a total 550 cells in the mature embryo (Figs. [2D](#F2){ref-type="fig"}, [5A](#F5){ref-type="fig"}) \[[@B7]\]. *In vitro*, we detected a comparable fraction (\~2%) of *unc-4*::GFP cells. Moreover, cultured *unc-4*::GFP cells adopt neuronal-like processes and express molecular markers also seen *in vivo*(Fig. [2D--E](#F2){ref-type="fig"}) \[[@B6]\]. On the basis of these results, we have profiled cultured *unc-4*::GFP neurons with the expectation that this approach will provide a comprehensive picture of genes expressed in these motor neurons *in vivo*.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**MAPCeL strategy for profiling *C. elegans*GFP neurons**. Embryos are isolated from gravid adults and treated with chitinase to degrade the egg shell. Embryonic cells are cultured for 24 hours and enriched by FACS. Amplified, labeled aRNA is hybridized to the Affymetrix *C. elegans*array.
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Isolation of *unc-4::GFP*neurons by FACS**. A. Fluorescence intensity profile of wildtype (non-GFP) cells. Boxed areas exclude autofluorescent7 cells (arrow). B. *unc-4*::GFP cells are gated to exclude propidium iodide-stained (non-viable) cells. C. Light scattering gate for GFP-positive cells (circle) to exclude cell clumps and debris. D. Combined fluorescence and DIC image of *unc-4*::GFP labeled motor neurons in L1 larva. (DA2 is not visible here.) Arrow points to embryo at stage (\< 400 min) prior to *unc-4*::GFP expression. E, F. Fluorescence and DIC images of 24 hr culture from *unc-4*::GFP embryos. G. *unc-4*::GFP neurons after enrichment by FACS. Arrow heads point to rare (\~10%) non-GFP cells. Scale bars are 5 microns.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Coefficients of Determination (R^2^) for individual hybridizations**. A. Scatter plot of normalized intensity values (log base 2) for representative hybridization (DMR30) from all cells (Reference) compared to the average intensity of four Reference hybridizations. B. Scatter plot of representative *unc-4*::GFP hybridization (DM39) compared to the average intensities for all three *unc-4*::GFP hybridizations. C. Results of single *unc-4*::GFP hybridization (DM39) (red) compared to average Reference intensities (green) to identify transcripts showing differential expression. The *unc-4*transcript (arrowhead) is highly enriched (\~13×) in *unc-4*::GFP neurons D. R^2^values for all pairwise combinations of *unc-4*::GFP hybridizations. E. R^2^values for all pairwise combinations of Reference (i.e. all cells) hybridizations.
:::
:::
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Comparison of Expressed Genes (EGs) in *unc-4*::GFP neurons vs all cells (N2 reference)**. 968 transcripts are detected exclusively in *unc-4*::GFP motor neurons and 3854 transcripts are detected exclusively in the N2 reference data set. 5249 transcripts are detected in both data sets.
:::
:::
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**GFP reporters validate UNC-4 motor neuron genes**. A. *unc-4*::GFP is expressed in 13 embryonically-derived motor neurons. Colored circles indicate approximate location of UNC-4 motor neuron soma in newly hatched L1 larva. Transgenic animals expressing GFP reporters for representative UNC-4 motor neuron genes. Anterior to left. B. *acr-14*. Confocal GFP/DIC projection. *acr-14*::GFP is expressed in RME in the head and in ventral nerve cord (VNC) motor neurons (arrows). C, D. *flp-13*. Confocal DIC/GFP image of head region. (C) and matched confocal GFP projection (D). Note DD1 commissure to dorsal side (white arrow). Arrowhead marks nerve ring. E, F. F29G6.2. Arrows point to VNC motor neurons (L2 Larva) (E). Posterior view of VNC showing F29G6.2::GFP expression in all VNC motor neurons. Asterisk marks gut autofluorescence (F). G, H. *tig-2*. DIC/GFP image of L2 larva. Note GFP expression in VNC motor neurons (arrows) and in head muscles (arrowhead) (G). Confocal projection of anterior VNC. *tig-2*::GFP is detected in A and B class motor neurons, pharyngeal muscle (arrow) and body wall muscle (bwm) (H). Scale bars are 10 microns in panels C, D, F, H and 20 microns in B, E, G.
:::
:::
We describe methods for isolating *unc-4*::GFP-labeled neurons by Fluorescence Activated Cell Sorting (FACS). mRNA from these cells is amplified, labeled and hybridized to the *C. elegans*Affymetrix Gene Chip. A comparison to microarray data derived from all embryonic cells reveals \~1000 genes with significantly higher levels of expression in *unc-4*::GFP neurons. The validity of these data is supported by the inclusion of genes known to be expressed in these neurons *in vivo*and by the generation of new GFP reporters from previously uncharacterized genes on this list. We conclude that MAPCeL offers a reliable strategy for profiling gene expression of a specific motor neuron class. Using this approach, we have provided, for the first time, a comprehensive picture of gene expression in a subset of *C. elegans*motor neurons. We expect that MAPCeL can now be applied to other *C. elegans*neurons and thereby link specific neuronal fates with unique combinations of differentially expressed genes.
Results
=======
Profiling strategy
------------------
*unc-4*::GFP is expressed in 13 embryonic motor neurons; (1) I5 (pharynx), (3) SAB (retrovesicular ganglion), and (9) DA (ventral nerve cord) (Figs [2D](#F2){ref-type="fig"}, [5A](#F5){ref-type="fig"}) \[[@B7]\]. Although each of the motor neuron classes is morphologically distinct, the DA and SAB motor neurons, which constitute the majority (12/13) of *unc-4*:GFP neurons, also share several characteristics including common presynaptic inputs, anteriorly directed axonal processes, cholinergic activity, and similar defects in *unc-4*mutants \[[@B2],[@B8]\]. It is therefore reasonable to assume that many of the same genes would be expressed in both of these motor neuron classes and that these could be revealed in microarray experiments.
A schematic of our approach to profile *unc-4::*GFP cells is presented in Fig. [1](#F1){ref-type="fig"}. *C. elegans*embryonic cells were cultured for 24 hr to allow differentiation of GFP-labeled motor neurons. *unc-4*::GFP cells are rarely observed in freshly dissociated preparations but constitute about 2% of all cells after 1 day in culture. The delayed appearance of *unc-4*::GFP cells in culture is consistent with the developmental timing of *unc-4*::GFP expression *in vivo*; *unc-4*::GFP motor neurons are normally generated after morphogenesis is initiated \[[@B7]\]. These older embryos are not dissociated by our methods \[[@B6]\]. Fluorescence Activated Cell Sorting (FACS) is used to isolate enriched (\~90%) populations of *unc-4*::GFP cells. RNA is extracted, amplified, and labeled for application to the *C. elegans*Affymetrix Gene Chip.
*unc-4*::GFP motor neurons are isolated by FACS
-----------------------------------------------
It is necessary to plate freshly dissociated embryonic cells on a solid substrate to promote differentiation and to prevent clumping. Although *C. elegans*neurons show extensive morphological differentiation on peanut lectin-coated glass they also adhere avidly and cannot be easily removed. We discovered that cells plated on poly-L-lysine coated surfaces also differentiate but can be readily dissociated from substrate by gentle trituration. A fluorescence profile was established for cells from the non-GFP wildtype strain (N2) to identify autofluorescent intestinal cells. Because these cells autofluoresce in both the Propidium Iodide (PI) and GFP channels, they are largely restricted to the diagonal axis of this scatter plot (Fig [2A](#F2){ref-type="fig"}). PI was added immediately prior to sorting to stain damaged cells (\~20%). Separate experiments with PI-stained wildtype cells and with cells from *unc-4*::GFP embryos were used to establish sorting gates for PI and GFP-labeled cells, respectively (Fig [2B](#F2){ref-type="fig"}). As shown in Fig [2C](#F2){ref-type="fig"}, viable *unc-4*::GFP neurons were simultaneously gated by light scattering parameters. This gate was established empirically to achieve \~90% enrichment of *unc-4*::GFP labeled cells (Fig [2G](#F2){ref-type="fig"}). We typically obtained about 40,000 *unc-4*::GFP neurons from each sort. RNA from the equivalent of 100,000 *unc-4*::GFP neurons was pooled for each separate microarray experiment. We will refer to microarray results from *unc-4*::GFP marked cells as the \"*unc-4*::GFP motor neuron\" data set. Reference RNA was extracted from all viable cells sorted from a 24 hr culture of wildtype embryonic cells. Microarray results with this \"Reference\" data set should reflect transcript levels in the average differentiated embryonic cell.
Microarray experiments yield reproducible profiles
--------------------------------------------------
Data obtained from successive hybridizations of two separate arrays with the same labeled probe yielded a coefficient of determination, R^2^= 0.99 (data not shown). This result indicates that potential differences between individual Affymetrix arrays or hybridization and scanning procedures are not significant sources of error. The overall concurrence of the experimental (*unc-4*::GFP motor neuron) and Reference data is illustrated graphically in the scatter plots shown in panels A and B of Fig [3](#F3){ref-type="fig"}. To assess the reproducibility of sample preparation methods (e.g. FACS isolation, RNA extraction, amplification, labeling, etc.), R^2^was calculated for each pairwise combination of independent samples. An average R^2^of 0.96 (n = 4) was calculated for the wildtype (N2) reference samples (Fig [3E](#F3){ref-type="fig"}); average R^2^was 0.92 (n = 3) for the *unc-4*::GFP motor neuron data set (Fig [3D](#F3){ref-type="fig"}). These values are indicative of highly similar samples and thereby show that our methods are reliable.
Detecting Expressed Genes (EGs)
-------------------------------
Differential hybridization to perfect match (PM) vs mismatch (MM) oligo probes on the Affymetrix chip was used to identify transcripts reliably detected as \"present\" in the Reference and *unc-4*::GFP motor neuron data sets (see Methods, Additional Files [4](#S4){ref-type="supplementary-material"}, [5](#S5){ref-type="supplementary-material"}). This list was adjusted in two ways for the *unc-4*::GFP motor neuron data set to arrive at a more accurate representation of Expressed Genes (EGs) (Additional Files [7](#S7){ref-type="supplementary-material"}, [17](#S17){ref-type="supplementary-material"}). In the first instance, transcripts that were statistically downregulated in *unc-4*::GFP motor neurons relative to the wildtype reference were removed from the \"present\" list as these are likely to be detected because they are actually highly enriched in contaminating the non-GFP cells (\~10%) ([Additional File 6](#S6){ref-type="supplementary-material"}). Conversely, we included transcripts that were considered enriched according to our statistical methods but originally scored as \"absent\" on the basis of PM vs MM signals used by Affymetrix MAS 5.0 software (see Methods, [Additional file 17](#S17){ref-type="supplementary-material"}). This second adjustment simply acknowledges that enriched transcripts are clearly expressed and therefore should be scored as \"present.\" We refer to the transcripts in these modified lists as EGs (Expressed Genes). A total of 9,103 EGs were detected in the Reference data set and 6,217 EGs in the *unc-4*::GFP motor neuron data set (Fig [4](#F4){ref-type="fig"}) (Additional Files [4](#S4){ref-type="supplementary-material"}, [7](#S7){ref-type="supplementary-material"}). Overall 10,071 unique transcripts were detected in these experiments or about 50% of all predicted *C. elegans*ORFs \[[@B9]\] ([Additional file 8](#S8){ref-type="supplementary-material"}). These results are comparable to microarray data from whole embryos that also detected about half of the predicted *C. elegans*genes \[[@B10]\]. Genes that are not detected may be expressed in a relatively small number of cells. This point is substantiated by our finding that 968 EGs in the *unc-4*::GFP motor neuron data set are not scored as present in the Reference data set ([Additional file 15](#S15){ref-type="supplementary-material"}, Fig [4](#F4){ref-type="fig"}). For example, the transcription factor UNC-3 is normally expressed in a small subset of embryonic neurons including the DAs \[[@B11]\]. The *unc-3*transcript is enriched in the *unc-4*::GFP motor neuron data set (Table [2](#T2){ref-type="table"}, [Additional file 9](#S9){ref-type="supplementary-material"}) but is not detected in the Reference ([Additional file 4](#S4){ref-type="supplementary-material"}). Thus, it seems likely that the overall number of EGs should increase as additional classes of embryonic cells are profiled (RMF, SEV, SJB, DMM, unpublished data).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Expression of promoter-GFP reporters for transcripts enriched in *unc-4*::GFP motor neuron data set. Reporters were examined for expression in DA, SAB and I5 neurons. 15/18 reporters showed GFP expression (bold type) in these cells. GFP reporters are listed according to statistical rank. All GFP-positive reporters were visible in embryos (data not shown) but were scored in larval animals to ease neuron identification.
:::
**Rank** **Cosmid** **Gene** **Protein** **UNC-4 neuron** **Other cells**
---------- ------------ ---------- ------------------- ------------------ ------------------------------------------------------------------------------
15 F33D4.3 *flp-13* neuropeptide **I5** ASE, ASG, ASK, BAG, DD, M3, M5, head neurons \[67\]
17 C11D2.6 *nca-1* Ca^++^channel **DA** DB, VA, VB, head/tail neurons
56 F09C3.2 phosphatase **DA** VA, VB, VD, DB, intestine, hypodermis
98 T19C4.5 novel no GFP
161 T23D8.2 *tsp-7* tetraspanin **DA** all VNC motor neurons, head/tail neurons, touch neurons
165 CC4.2 *nlp-15* neuropeptide DD, head/tail neurons, body muscles, pharyngeal muscle \[105\]
210 F29G6.2 novel **DA** DB, touch neurons, pharyngeal neurons, head neurons
215 F39G3.8 *tig-2* TGF-β **DA** VA, VB, DB, body wall muscle, touch neurons, pharyngeal muscle
233 F55C12.4 novel **DA** VB, DB, DD, AS, VD
234 E03D2.2 *nlp-9* neuropeptide VA, intestine, head neurons \[105\]
239 ZC21.2 *trp-1* Ca^++^channel **DA** DB, VA, VB
254 Y47D3B.2A *nlp-21* neuropeptide **DA** DB, VA, VB, AS, body muscle, head neurons, intestine \[72\]
329 F36A2.4 *twk-30* K^+^channel **DA** all VNC motor neurons \[106\]
377 C18H9.7 *rpy-1* rapsyn **DA** VD, AS, VB, DB, body muscles
593 F43C9.4a *mig-13* CUB domain **DA** DB, ant. VNC motor neurons, pharyngeal/intestinal valves, hypodermis \[107\]
782 T05C12.2 *acr-14* nAChR **DA** VB, AS, DB, DD, HSN, VC4 & 5, AIY, head neurons, muscles, intestine
788 T27A1.6 *mab-9* transcrip. factor **DA** DD, DB, VD, AS
877 K02E10.8 *syg-1* Ig Domain **DA** VA, HSN and other neurons
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Summary of genes with enriched transcripts in *unc-4::*GFP neurons. Genes are organized into categories according to molecular function (KOG or other description <http://www.wormbase.org>). Gene families or functional groups with potential functions in neurons are emphasized in this list. Statistical rank is indicated for each transcript.
:::
**Cosmid Name** **Common Name** **Rank** **KOG (Other description)**
--------------------------------------------- ----------------- ---------- -------------------------------------------------------------------------------------------
***Axon Guidance and Outgrowth***
B0273.4a *unc-5* 934 Netrin transmembrane receptor unc-5
T19B4.7 *unc-40* 188 Receptor mediating netrin-dependent axon guidance
F41C6.1 *unc-6* 616 Netrin, axonal chemotropic factor
F56D1.4a *clr-1* 754 Protein tyrosine phosphatase
M79.1a *abl-1* 882 Protein tyrosine kinase
F09B9.2 *unc-115* 696 Actin-binding LIM Zn-finger protein Limatin involved in axon guidance
B0350.2 *unc-44* 449 Ankyrin
C01G10.11a *unc-76* 182 Kinesin-associated fasciculation and elongation protein involved in axonal transport
K10D3.2 *unc-14* 763 (RUN domain protein required for axonogenesis and sex myoblast migration)
***Wingless Signaling***
K10B4.6 *cwn-1* 216 Wnt family of developmental regulators
Y71F9B.5a *lin-17* 899 Smoothened and related G-protein-coupled receptors (Frizzled Receptor)
***Acetylcholine Receptor Subunits***
K11G12.2 *acr-2* 66 Acetylcholine receptor
R01E6.4 *acr-12* 292 Acetylcholine receptor
F21F3.5 *unc-38* 399 Acetylcholine receptor
T05C12.2 *acr-14* 782 Acetylcholine receptor
Y110A7A.3 *unc-63* 800 Acetylcholine receptor
F21A3.7 212 Acetylcholine receptor
Y105E8A.7 *lev-10* 323 Cubilin, multiligand receptor mediating cobalamin absorption
T14A8.1 *ric-3* 237 Unnamed protein (Required for nAChR assembly/trafficking)
***Ligand-gated Ion Channel***
ZC196.7 *glr-5* 792 Glutamate-gated kainate-type ion channel receptor subunit GluR5 and related subunits
T27E9.9 41 Ligand-gated ion channel (glycine/GABA)
Y71D11A.5 502 Ligand-gated ion channel (glycine/GABA)
Y46G5A.30 *snf-5* 686 Sodium-neurotransmitter symporter
C09E8.1a 578 Sodium-neurotransmitter symporter
***G-proteins***
M01D7.7a *egl-30* 124 G protein subunit Galphaq/Galphay, small G protein superfamily
F08B6.2 *gpc-2* 469 G protein gamma subunit
F56H9.4 *gpa-9* 638 G-protein alpha subunit (small G protein superfamily)
***G-protein Pathway Components***
F28C1.2 *egl-10* 872 G protein signaling regulators
C05B5.7 *rgs-1* 486 G protein signaling regulators
F17C8.1 *acy-1* 884 Adenylyl cyclase
R07E4.6 *kin-2* 467 cAMP-dependent protein kinase types I and II, regulatory subunit
C17F4.6 *gcy-19* 213 Natriuretic peptide receptor, guanylate cyclase
C50H2.2 *egl-47* 432 (Gα~o~coupled receptor)
F57F5.5 *pkc-1* 747 Serine/threonine protein kinase
F39B2.8 928 Predicted membrane protein
C24A8.4 1010 STE20-like serine/threonine kinase MST
***Neuropeptides***
W07E11.3 *flp-2* 31 Unnamed protein (FMRF-like peptide)
C18D1.3 *flp-4* 691 (FMRF-like peptide)
C03G5.7 *flp-5* 562 (FMRF-like peptide)
F33D4.3 *flp-13* 15 Unnamed protein (FMRF-like peptide)
E03D2.2 *nlp-9* 234 Unnamed protein (Neuropeptide-like protein)
CC4.2 *nlp-15* 165 Unnamed protein (Neuropeptide-like protein)
Y47D3B.2a *nlp-21* 254 Unnamed protein (Neuropeptide-like protein)
F13B12.5 *ins-1* 200 (Insulin-like peptide)
T28B8.2 *ins-18* 767 Unnamed protein (Insulin-like peptide)
F56A11.5 928 Uncharacterized Fe-S protein
***Neuropeptide Processing and Secretion***
T03D8.3 613 Proprotein convertase (PC) 2 chaperone involved in secretion (neuroendocrine protein 7B2)
C32E8.7 *ric-19* 20 Secretory vesicle-associated protein ICA69, contains Arfaptin domain
ZK897.1 *unc-31* 588 Ca2+-dependent activator protein
***Neuropeptide Receptor***
T05A1.1 *npr-2* 362 7 transmembrane receptor (neuropeptide receptor family)
F59D12.1 797 7 transmembrane receptor (rhodopsin-like GPCR)
T07D4.1 77 7 transmembrane receptor (rhodopsin-like GPCR)
Y62E10A.4 275 7-transmembrane receptor (rhodopsin-like GPCR)
K07E8.5 452 Unnamed protein (FMRF receptor)
C35A11.1 814 Unnamed protein (rhodopsin-like GPCR)
ZC84.4 708 7 transmembrane receptor (rhodopsin-like GPCR)
F56B6.5 *uvt-6* 629 7 transmembrane receptor (somatostatin receptor)
F56A11.5 928 Uncharacterized Fe-S protein (Neuropeptide receptor activity)
:::
Selected *C. elegans*genes are enriched in UNC-4 motor neurons
--------------------------------------------------------------
A majority of transcripts in the Reference and *unc-4*::GFP motor neuron data sets show comparable levels of expression (Fig [2](#F2){ref-type="fig"}). Many of these transcripts are likely to encode core functions required in every cell. Other transcripts in this group could be limited to subsets of embryonic cells that include UNC-4 motor neurons. Genes that are widely expressed in neurons, for example, may not be detectably enriched in *unc-4*::GFP motor neurons in comparison to the Reference because neurons constitute a significant fraction (\~40%) of all cells in the embryo. To illustrate this point, we note that UNC-64 (Syntaxin), an integral component of the neurotransmitter release mechanism and therefore expressed in most neurons \[[@B12],[@B13]\], is detected in the *unc-4*::GFP motor neuron data set but is not enriched (Table [2](#T2){ref-type="table"}, Additional Files [7](#S7){ref-type="supplementary-material"}, [14](#S14){ref-type="supplementary-material"}).
As graphically illustrated in the scatter plot shown in Fig. [3C](#F3){ref-type="fig"}, subsets of genes in the *unc-4*::GFP motor neuron data set are differentially expressed relative to the average expression levels for all cells in the Reference data set (R^2^= 0.88). As expected for a gene that is selectively expressed in *unc-4*::GFP neurons, the hybridization signal for the *unc-4*transcript is highly elevated (13×) in comparison to all cells. Significant numbers of genes are also under-expressed in UNC-4 motor neurons relative to other embryonic cells. Transcripts showing ≥ 1.7× fold intensity difference in the *unc-4*::GFP motor neuron vs Reference data sets were defined using SAM statistics at a False Discovery Rate (FDR) of ≤ 1%. By these criteria 1012 genes are enriched (red) in UNC-4 motor neurons ([Additional file 9](#S9){ref-type="supplementary-material"}) whereas 1596 transcripts are depleted (green) ([Additional file 10](#S10){ref-type="supplementary-material"}). The threshold of ≥ 1.7× fold was defined empirically. At higher values (e.g. ≥ 2.0×) genes with known expression in these cells were excluded (e.g. *acr-2, unc-5*) \[[@B14],[@B15]\] ([Additional file 14](#S14){ref-type="supplementary-material"}) whereas, a lower threshold of 1.5× included significantly more false positives (e.g. muscle genes, *pat-3, sup-10*) \[[@B16],[@B17]\].
Confirmation of UNC-4 motor neuron genes
----------------------------------------
Information gleaned from published literature <http://www.textpresso.org> and from wormbase <http://www.wormbase.org>, identified 27 genes with known expression in embryonic motor neurons that also express *unc-4*::GFP (I5, SAB, DA) ([Additional file 11](#S11){ref-type="supplementary-material"}). We detect 21 (78%) of these genes as EGs of which 10 (37%) are enriched. In addition, a significant number of transcripts encoding core neuronal functions (e.g. axon guidance, neurotransmitter signaling, etc.) are detected in the *unc-4*::GFP data set (Table [2](#T2){ref-type="table"}, [Additional file 14](#S14){ref-type="supplementary-material"}). For example, in addition to UNC-64 (syntaxin or t-SNARE,) other components of the SNARE complex, SNB-1 (synaptobrevin or v-SNARE) and SNAP-25 (Y22F5A.3) are detected \[[@B18],[@B19]\]. We also examined the data set for potential false positives by considering transcripts that are known to be highly expressed in other tissues but not in UNC-4 motor neurons. For example, in the embryo, the homeodomain protein UNC-30 is exclusively detected in DD motor neurons. Expression of the GABA vesicular transporter, UNC-47, in DD motor neurons depends on *unc-30*function \[[@B20]\]. UNC-4 motor neurons are cholinergic and as expected neither of these GABA specific transcripts are present in the *unc-4*::GFP motor neuron data set ([Additional file 7](#S7){ref-type="supplementary-material"}).
The strong representation of \~80% of genes known to be expressed in I5, SAB, and DA motor neurons in the *unc-4*::GFP motor neuron dataset indicates that other previously uncharacterized transcripts in this list are also likely to be expressed in these cells *in vivo*. To test this idea, we evaluated GFP reporter lines for representative genes detected as enriched in the *unc-4*::GFP motor neuron data set (Fig. [5](#F5){ref-type="fig"}). As shown in Table [1](#T1){ref-type="table"}, 82% (15/18) of these promoter-GFP fusions show expression in UNC-4 motor neurons *in vivo*. Of the GFP reporters not detected in these neurons, one of them, T19C4.5, fails to express GFP in any cell. This finding could mean that the upstream sequence selected for this construct does not overlap the endogenous T19C4.5 gene regulatory region. In some cases, cell-specific expression of *C. elegans*genes depends on distal upstream regions, intronic sequences, or 3\' domains that would not be included in these 5\' promoter GFP fusions \[[@B21]\]. This explanation could also account for the apparent absence of GFP expression in the *unc-4*::GFP motor neurons of the *nlp-9*and *nlp-15*GFP reporters. The validity of this data set is further substantiated by the observation that GFP expression in DA motor neurons is detected even for lower ranking genes (e.g. *syg-1*::GFP, statistical rank = 877). Thus, we believe that the transcripts listed in the *unc-4*::GFP motor neuron data set are likely to constitute an accurate representation of genes normally expressed in these cells.
We note that the positive GFP reporters shown in Table [1](#T1){ref-type="table"} are not uniformly detected in UNC-4 neurons: all but one (*flp-13*) are expressed in the DAs, one in I5 and none in the SAB motor neurons. This bias reflects the relative abundance of DA motor neurons (\~70% or 9/13 of *unc-4*::GFP neurons *in vivo*) in the cells used to generate this data set and thus could indicate that most of the enriched transcripts are also expressed in the DAs. Therefore, results presented below are largely focused on potential gene functions in DA motor neurons.
Families of neuronal genes expressed in UNC-4 motor neurons
-----------------------------------------------------------
Here we describe transcripts detected in the *unc-4*::GFP dataset with an emphasis on genes that are enriched in these cells and therefore likely to encode proteins with important roles in the differentiation or function of UNC-4 motor neurons (Table [2](#T2){ref-type="table"}). A comprehensive discussion of gene families from this list can be found in [Additional file 16](#S16){ref-type="supplementary-material"}. Selected examples are presented here. Gene names for enriched trancripts discussed in this section are shown in bold and are listed in Table [2](#T2){ref-type="table"}. All EGs are listed in [Additional file 7](#S7){ref-type="supplementary-material"}.
Axon guidance and outgrowth
---------------------------
Growth cone steering and cell migration along the dorsal-ventral body axis in *C. elegans*depend on the **UNC-6/netrin**guidance cue. The **UNC-40/DCC**receptor mediates an attractive response to **UNC-6/netrin**whereas co-expression of **UNC-40/DCC**with a second **UNC-6**receptor, **UNC-5**, results in repulsion \[[@B15],[@B22]\]. The **UNC-6/netrin**signal is released from ventral ectoderm \[[@B23]\] to repel growth cones expressing both **UNC-40**and **UNC-5**; this interaction is required for normal outgrowth of DA motor neuron commissures to the dorsal nerve cord \[[@B15]\]. As expected, ***unc-5***and ***unc-40***transcripts are enriched in UNC-4 motor neurons. ***unc-6***, which is known to be expressed in the I5 pharyngeal neuron, is also elevated \[[@B23]\]. The **CLR-1**receptor protein tyrosine phosphatase (RPTP) is proposed to inhibit attractive **UNC-6/netrin**responses via interactions with **UNC-40**. In the DA motor neurons, **CLR-1**also promotes **UNC-6/netrin**repulsion by an **UNC-40**-independent mechanism \[[@B24]\]. As predicted by these models, the ***clr-1***transcript is elevated in UNC-4 motor neurons. Relevant to this point, we note that the *C. elegans*Abelson tyrosine kinase ortholog, ***abl-1***, is also enriched. In *Drosophila*, Abl tyrosine kinase antagonizes the axon guidance role of RPTPs in motor neurons \[[@B25]\]. It will be interesting to determine if **ABL-1**functions similarly in *C. elegans*and, in this case, if **ABL-1**works in opposition to **CLR-1**during DA motor axon outgrowth (Fig. [6](#F6){ref-type="fig"}).
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**Model of DA motor neuron axon guidance**. Ventrally derived UNC-6/Netrin guidance cues binds to the UNC-40/DCC and UNC-5 receptor to steer the DA motor axon toward the dorsal nerve cord. The receptor tyrosine phosphatase, CLR-1, promotes dorsal motor axon outgrowth via an UNC-40/DCC independent pathway \[24\]. The transcript encoding the *C. elegans*ortholog of Abelson tyrosine kinase (ABL-1) is enriched in the *unc-4*::GFP motor neuron data set and is proposed to antagonize CLR-1 activity.
:::
:::
We also detected axon guidance effectors ***unc-115***and *ced-10*in our microarray array dataset. Genetic approaches have shown that ***unc-115***(AbLIM, actin binding protein) and *ced-10*(Rac GTPase) are downstream effectors of **UNC-40**signaling and presumptive links to the cytoskeleton \[[@B26],[@B27]\].
Transcripts for genes with general roles in axon outgrowth are enriched in the *unc-4*::GFP motor neuron data set. These include ***unc-44***(ankyrin-like), ***unc-76***(novel) and ***unc-14***(RUN domain). All three of these genes are highly expressed in the *C. elegans*nervous system. ***unc-44***encodes multiple alternatively spliced transcripts with broad roles in axonal morphogenesis \[[@B28]\]. **UNC-76**and its vertebrate homologs define a new protein class of unknown biochemical function. In *C. elegans*, ***unc-76***mutants show axon outgrowth and fasciculation defects \[[@B29]\]. ***unc-14***and *unc-51*(serine/threonine kinase) mutants display similar neuronal defects with misplaced processes and enlarged abnormal varicosities \[[@B30]\]. UNC-51 (EG) has been proposed to phosphorylate **UNC-14**to regulate vesicular trafficking during axonal process outgrowth \[[@B31],[@B32]\].
Wingless signaling
------------------
Wingless (Wnt) signaling controls multiple developmental processes in the nervous system ranging from cell determination to axon guidance and synaptogenesis \[[@B33],[@B34]\]. The *C. elegans*genome contains 5 Wnt genes and 4 Wnt receptors or Frizzled homologs \[[@B35]\]. One of each, ***cwn-1***(Wnt) and ***lin-17***(Frizzled), are enriched. Transcripts for other components of the canonical (*mig-5*, *mom-5*, *cwn-2*, *dsh-1*, *dsh-2*, *Y73B6BL.21*) and noncanonical (*lit-1*, *mom-4*, *par-1*, *tap-1*) Wnt signaling pathways are detected as EGs. Thus, UNC-4 motor neurons are presumptively competent to send as well as respond to Wnt signals. Functions for Wnt signaling in the *C. elegans*motor neuron circuit have not been defined. Possibilities include the regulation of synaptogenesis as suggested by studies of *Drosophila*motor neurons which secrete Wnt to control both presynaptic and postsynaptic differentiation at the neuromuscular synapse \[[@B36]\]. A gradient of Wnt signaling controls cell migration along the AP axis in *C. elegans*\[[@B37]\]. Responsiveness to this graded Wnt signal could account for the anterior polarity of DA motor neurons in the dorsal nerve cord as suggested by the recent finding that commissural axonal polarity along the AP axis in the vertebrate spinal cord is dependent on Wnt signaling \[[@B38]\].
Nicotinic Acetylcholine Receptors (nAChRs)
------------------------------------------
The *C. elegans*genome encodes at least 27 distinct nAChR subunits \[[@B39]\]. Two of these, **ACR-2**and **UNC-63**are expressed in DA class motor neurons \[[@B14],[@B40]\] and are enriched in the *unc-4*::GFP motor neuron data set. Expression of *unc-29*\[[@B41]\] and *unc-38*(J.L. Bessereau, personal communication) in ventral cord motor neurons has been previously reported and these are also detected as EGs ([Additional file 7](#S7){ref-type="supplementary-material"}). *acr-12*::GFP is expressed in neurons (A. Gottschalk and W. Schafer, personal communication), and we have validated the enrichment of ***acr-14***with GFP reporters that confirm expression in DA motor neurons (Fig [5](#F5){ref-type="fig"}). In body muscle, **UNC-63**is an essential component of a levamisole-sensitive nACh receptor that also includes UNC-29, UNC-38, LEV-1 and LEV-8 \[[@B40],[@B42]\]. **ACR-12**may coassemble with **UNC-63**, UNC-29, and UNC-38 to generate a related nACh receptor in UNC-4 motor neurons (A. Gottschalk and W. Schafer, personal communication). Five additional sets of nAChR subunits are detected as EGs and a so-called \"orphan\" ligand gated ion channel (LGIC) subunit, **F21A3.7**, with significant similarity to the nAChR gene family, is enriched. Despite the diversity of nAChR subunits expressed in UNC-4 motor neurons and the potentially complex array of resultant receptors, no functions have been directly assigned to nAChRs in these cells \[[@B43]\]. Although loss-of-function mutations in nAChR subunits that are also expressed in muscle (i.e. *unc-29*, *unc-3*8, ***unc-63***) result in locomotory defects, gene knockouts of ***acr-2***(Y. Jin, personal communication) and ***acr-12***(data not shown), which are exclusively expressed in neurons, do not produce obvious effects on motility or behavior. Perhaps the surprisingly large number (12) of nAChR subunit genes detected in these cells results in overlapping functions that mask defects in single gene knockout mutants. Alternatively, these nAChRs may mediate subtle aspects of motor neuron activity. This idea is consistent with models in which nAChRs act presynaptically to modulate neurotransmitter release \[[@B44],[@B45]\]. Finally, we detect enrichment of transcripts for proteins **RIC-3**(novel) and **LEV-10**(CUB domain) that mediate nAChR localization \[[@B46],[@B47]\].
Ligand-Gated Ion Channels
-------------------------
UNC-4 motor neurons are potentially responsive to additional classes of neurotransmitters. Enrichment of ***glr-5***(kainate type ionotropic glutamate receptor subunit) is correlated with its known expression in the SAB motor neurons \[[@B48]\]. As members of the GABA/Glycine family of ligand-gated receptors, the presumptive anion channels encoded by **T27E9.9**and **Y71D11A.5**are predicted to hyperpolarize UNC-4 motor neurons and thus inhibit cholinergic activity \[[@B49]\]. It may be significant that a candidate sodium/chloride-dependent glycine transporter, ***snf-5***, is enriched. (**C09E8.1**, an outlier in the sodium/chloride-dependent transporter family is also enriched.) In mammalian cells, plasma membrane transporters GLYT1/GLYT2 remove glycine from the synaptic cleft, and in the case of GLYT2, thereby recycle glycine for reuptake into synaptic vesicles \[[@B50]\]. UNC-4 motor neurons do not express the GABA/Glycine vesicular transporter, UNC-47, however, and are therefore unlikely to release glycine presynaptically \[[@B51]\]. In this case, the physiological function of the **SNF-5**transporter could mirror that of GLYT1, which is believed to attenuate glycinergic signaling by pumping glycine into a non-glycinergic glial cell \[[@B52]\]. To date, the potential function of glycinergic signaling in *C. elegans*has not been explored.
G-protein signaling
-------------------
Cholinergic motor neuron activity in *C. elegans*is modulated by G-protein signaling pathways that respond to the neurotransmitters acetylcholine, serotonin (5-HT), and dopamine (Fig. [7](#F7){ref-type="fig"}) \[[@B53]-[@B55]\]. In each case, acetylcholine release is either promoted by **EGL-30**(Gα~q~) or inhibited by GOA-1 (Gα~o~). Input to these antagonistic pathways is provided by G-protein coupled receptors (GPCRs). Pharmacological evidence suggests that a muscarinic acetylcholine receptor activates **EGL-30**to enhance ACh release at the neuromuscular synapse \[[@B53],[@B56]\]. The enriched muscarinic AChRs, **GAR-2**and **GAR-3**could account for this effect \[[@B57],[@B58]\]. Similarly the enriched 5-HT receptor, **SER-4**, is a strong candidate for the GPCR mediating the inhibitory effect of serotonin on ACh release from ventral cord motor neurons \[[@B54]\]. Dopamine may either activate or inhibit ACh release within the same cholinergic motor neuron. Activation depends on **DOP-1**which is enriched in UNC-4 motor neurons. Inhibition is attributed to DOP-3. Expression of DOP-3 in cholinergic ventral cord motor neurons is reportedly weak and we do not detect the *dop-3*transcript in our data set \[[@B55]\]. UNC-4 motor neurons are also potentially responsive to GABA as a transcript (**Y41G9A.4**) encoding a metabotropic GABA type B1 receptor is enriched. GABA dependent effects on cholinergic motor neuron activity have not been previously reported in *C. elegans*.
::: {#F7 .fig}
Figure 7
::: {.caption}
######
**G-protein signaling pathways regulating neurotransmitter release in cholinergic motor neurons**. Components shown in bold are enriched in the *unc-4*::GFP motor neuron data set. All others are EGs with the exception of the *dop-3*transcript which is not detected (light gray text). See Table 3 for protein descriptions. Green denotes interactions that promote acetylcholine (ACh) release and red marks steps that inhibit synaptic vesicle fusion. Neurotransmitters are highlighted in gray boxes. This figure adapted from Reynolds et al. (2004) \[60\].
:::
:::
Genetic screens for mutations affecting neurotransmitter release have revealed a complex web of interacting components that couple G-protein signaling to synaptic vesicle fusion (Fig [7](#F7){ref-type="fig"}) \[[@B53],[@B56],[@B59],[@B60]\] (D. Sieburth and J. Kaplan, personal communication). With one exception (*dop-3*), transcripts for all of the known components of these pathways are either enriched (***acy-1***, ***egl-10***, ***gpc-2***, ***kin-2***, ***pkc-1***) or detected as EGs (*dgk-1*, *egl-8*, *egl-16*, *gpb-2*, *gsa-1*, *kin-2*, *ric-8*, *unc-13*). Lack of enrichment of some of these components is consistent with the widespread utilization of G-protein signaling pathways in *C. elegans*neurons and muscle cells \[[@B61],[@B62]\]. As noted above, these data have also revealed several additional enriched transcripts with potential roles in G-protein dependent locomotory behavior. ***egl-47***encodes an orphan GPCR and ***rgs-1***an RGS protein, both of which can regulate *goa-1*signaling in the egg laying circuit \[[@B63],[@B64]\]. RNAi of **F39B2.8**, which encodes a highly conserved but unusual protein with both serine/threonine kinase and 7-transmembrane domains, results in a locomotory defect \[[@B65]\] that could be indicative of a neuromodulatory function in DA motor neurons. A complete list of G-protein signaling components detected in this dataset can be found in Table [3](#T3){ref-type="table"}.
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Summary of G-protein signaling genes in *unc-4::*GFP neurons. Genes either known to function in *C. elegans*motor neuron G-protein signaling pathways or likely to have a role therein (e.g. SER-4) are listed here with KOG descriptions. Transcripts are listed as either enriched with statistical rank (top) or as EGs (bottom).
:::
**Cosmid Name** **Common Name** **Statistical Rank** **KOG Description**
--------------------- ----------------- ---------------------- ----------------------------------------------------------------------------------------------------------
***Enriched***
F47D12.1 *gar-2* 245 7 transmembrane receptor
Y40H4A.1 *gar-3* 421 Muscarinic acetylcholine receptor
F15A8.5d *dop-1* 683 7 transmembrane receptor
Y22D7AR.13 *ser-4* 680 7 transmembrane receptor
Y41G9A.4 769 GABA-B ion channel receptor subunit GABABR1 and related subunits, G-protein coupled receptor superfamily
M01D7.7a *egl-30* 124 G protein subunit Galphaq/Galphay, small G protein superfamily
F28C1.2 *egl-10* 872 G protein signaling regulators
F57F5.5 *pkc-1* 743 Serine/threonine protein kinase
F17C8.1 *acy-1* 884 Adenylyl cyclase
R07E4.6 *kin-2* 467 cAMP-dependent protein kinase types I and II, regulatory subunit
***Present (EGs)***
Y69A2AR.1 *ric-8* Signaling protein RIC-8/synembryn (regulates neurotransmitter secretion)
C16C2.2a *eat-16* G protein signaling regulators
B0348.4a *egl-8* Phospholipase C
F52A8.2 *gpb-2* G-protein beta subunit
C26C6.2 *goa-1* G-protein alpha subunit (small G protein superfamily)
ZK542.2a *unc-13* Neurotransmitter release regulator, UNC-13
C09E10.2a *dgk-1* Diacylglycerol kinase
R06A10.2 *gsa-1* G protein subunit Galphas, small G protein superfamily
:::
Neuropeptide signaling
----------------------
The *C. elegans*genome includes a large and diverse array of genes encoding potential neuroactive peptides. GFP reporter studies indicate that these genes are predominantly expressed in neurons. 23 \"flp\" genes encoding FMRFamide and related peptides (FaRPs) have been described. FaRPs have been shown to modulate a wide array of invertebrate neural functions \[[@B66]\]. Previously reported expression of ***flp-2, flp-4, flp-13***in the pharyngeal I5 neuron \[[@B67]\] (Fig [5](#F5){ref-type="fig"}) is confirmed by their enrichment in the *unc-4*::GFP motor neuron data set (Table [I](#T1){ref-type="table"}). ***flp-5***is also elevated in these cells and 8 additional flps are detected as EGs (Additional Files [7](#S7){ref-type="supplementary-material"}, [14](#S14){ref-type="supplementary-material"}). Specific FaRPs modulate cell excitability (***flp-13***), locomotion (*flp-1*) and feeding behavior (*flp-21*) in *C. elegans*\[[@B68],[@B69]\]. The inhibitory action of the **FLP-13**peptide on pharyngeal muscle activity is consistent with its expression in I5 \[[@B70]\].
The *C. elegans*genome contains 37 genes encoding predicted insulin-like peptides \[[@B71]\]. Transcripts for two of these, ***ins-1***and ***ins-18***, are enriched; *ins-17*, *ins-24*and *ins-30*are present but not significantly elevated relative to other cells. ***ins-1***and ***ins-18***have been implicated in the DAF-2 insulin receptor dependent pathways regulating growth, metabolism and lifespan \[[@B71]\].
A total of 32 genes encoding other potential classes of neuropeptides have also been identified in the *C. elegans*genome. Three of these neuropeptide-like protein (nlp) genes, ***nlp-9***, ***nlp-15***, and ***nlp-21***, are enriched in UNC-4 motor neurons (Table [2](#T2){ref-type="table"}). An additional group of 11 nlp transcripts are detected as EGs (Additional File [7](#S7){ref-type="supplementary-material"}, [14](#S14){ref-type="supplementary-material"}). To date, no functions have been directly assigned to nlp genes in *C. elegans*\[[@B72]\].
Our studies have revealed that a surprisingly large number of neuropeptide genes are transcribed in UNC-4 motor neurons. These results indicate that UNC-4 motor neurons are likely to exhibit significant neurosecretory activity. This conclusion is consistent with our finding that proteases required for neuropeptide processing and activation \[**T03D8.3 (**Proprotein convertase (PC) 2 chaperone), *egl-3*(zinc carboxypeptidase) and *egl-21*(subtilisin-like proprotein convertase)\] are also expressed in these cells \[[@B73]-[@B75]\]. Other genes with important roles in neurosecretion are also detected. ***ric-19***encodes a novel arfaptin-related protein that is believed to function in the Golgi in the generation of neurosecretory granules and may through this activity and subsequent neuropeptide signaling exert an indirect effect on ACh release from conventional synaptic vesicles \[[@B76],[@B77]\]. Our finding that ***ric-19***is highly enriched in cholinergic motor neurons could be indicative of autocrine neuropeptide modulation of ACh secretion at the neuromuscular synapse. Consistent with this idea is our finding that *ida-1*, a conserved membrane component of the dense core vesicles in which neuropeptides are typically packaged, is an EG \[[@B78]\]. Finally, **UNC-31**(CAPS), a known facilitator of dense core vesicular release, is enriched \[[@B78]\]. Plasma membrane fusion of both dense core vesicles and the small, clear vesicles in which classical neurotransmitters are packaged, depend on a common set of calcium-activated components \[[@B79]\] most of which are either enriched or present in these cells (see Table [2](#T2){ref-type="table"} and Additional Files [14](#S14){ref-type="supplementary-material"}, [16](#S16){ref-type="supplementary-material"}).
In addition to secreting neuropeptides, UNC-4 motor neurons are also likely to respond to neuropeptide signaling. Transcripts for nine putative neuropeptide receptors are enriched. (Table [2](#T2){ref-type="table"}). RNAi of two of these, ***npr-2***and **F59D12.1**, results in locomotory defects that could be indicative of specific functions in DA motor neurons \[[@B65]\]. ***npr-2***is a close relative of *npr-1*(not detected) which has been shown to control social feeding behavior in response to the FLP-21 (not detected) peptide \[[@B69]\]. **F59D12.1**is most closely related to melatonin receptors but its *in vivo*ligand is unknown. Neuropeptides are believed to modulate secretion of classical neurotransmitters \[[@B79]\]. Neuropeptide specific effects on excitatory motor neuron activity in the *Ascaris*ventral nerve cord have been reported \[[@B80]\]. Genetic evidence in *C. elegans*indicates that acetylcholine release at the neuromuscular junction is enhanced by neuropeptide activity \[[@B73]\] and that this pathway depends on the EGL-30 G~q~α protein \[[@B68]\] (Table [2](#T2){ref-type="table"}). These neuropeptides may be released from neurons and also as a retrograde signal from muscle cells \[[@B73],[@B81]\].
Other classes of enriched transcripts are discussed in [Additional file 16](#S16){ref-type="supplementary-material"} (Transcription factors, Cell Adhesion Molecules, Synapse-Associated Proteins, Neurotransmitter Vesicular Release Components, TGF-β Signaling Proteins, Serpentine Receptors, Calcium Channels, Calcium Ion Binding Proteins, Potassium Channels, Innexins, DEG/ENaC Channels, and Stomatins).
Discussion
==========
We have described MAPCeL, a microarray-based strategy for fingerprinting specific *C. elegans*neurons, and provide evidence that this approach can reveal a comprehensive picture of gene expression in these cells *in vivo*. *unc-4*::GFP-marked neurons were isolated by FACS from primary cultures of embryonic cells and profiled on the *C. elegans*Affymetrix gene chip. Because these *unc-4*::GFP neurons differentiate *in vitro*, it was important to establish that our microarray data provide an accurate representation of gene expression in the intact animal. This conclusion is supported by three observations: (1) A majority (21/27) of genes with known expression in *unc-4*::GFP neurons are detected in our microarray data set; (2) \~80% (15/18) of GFP reporters constructed for transcripts enriched in UNC-4 motor neurons are expressed in these cells *in vivo*(Table [1](#T1){ref-type="table"}, Fig. [5](#F5){ref-type="fig"}); (3) Transcripts known to encode proteins with key roles in *unc-4*::GFP motor neuron differentiation (e.g. axon guidance and outgrowth, synaptogenesis) and function (e.g., neurotransmitter vesicle release, G-protein signaling pathways) are highly represented in our data sets. These findings parallel earlier studies showing that cultured *C. elegans*neurons and muscle cells adopt apparently normal morphological and physiological characteristics \[[@B6]\] and are consistent with evidence favoring a cell autonomous mode of differentiation for *C. elegans*embryonic cells after an initial phase of inductive signaling events \[[@B4],[@B82]\]. We have now generated comparable microarray profiles of other motor neuron classes and muscle cells that also show strong congruence with known patterns of gene expression (RMF, SEV, SJB, DMM, unpublished data). We therefore conclude that our approach of profiling GFP marked neurons isolated from primary culture can now be widely applied to fingerprint specific *C. elegans*embryonic cells. In some cases, however, differentiation of a given neuron is likely to depend on specific intercellular signals that primary cultures will not provide. Thus, in every instance, it will be necessary to confirm microarray data by independent methods as described here.
Methods for profiling specific *C. elegans*cells
------------------------------------------------
Previous studies have described other methods for cataloging transcripts from specific *C. elegans*cells. Comparisons of microarray data from mutant animals with either supernumerary or absent sensory neurons in the male tail, have revealed genes that are preferentially expressed in these cells \[[@B83]\]. However this approach is limited to cell types that can be manipulated by specific mutants. In addition, this method may be insufficiently sensitive to detect changes in smaller subsets of cells due to high background mRNA from cells that are not affected by the mutation (SEV, DMM, unpublished data). This limitation can be overcome by enriching for mRNA from target cells. To this end, Zhang et al. (2002) used an approach similar to the strategy outlined in this paper to identify downstream genes of the MEC-3 transcription factor in *C. elegans*touch neurons \[[@B84]\]. However, this work did not provide a comprehensive cell-specific gene expression profile as we have here perhaps due to the limited enrichment (\~50%) of GFP-labeled touch neurons. We have now optimized the application of nematode embryonic cell culture and FACS technology to obtain \~90% enrichment of GFP-marked neurons and muscle cells (Fig. [2](#F2){ref-type="fig"}) (RMF, SEV, SJB, DMM, unpublished data). These methods have now been successfully applied to profile other classes of *C. elegans*embryonic cells \[[@B85],[@B86]\].
MAPCeL cannot be used for postembryonic cells because these apparently do not arise in culture \[[@B6]\]. Microarray profiles of specific larval cells have been obtained, however, by mRNA tagging. In this approach, an epitope-labeled polyA binding protein (FLAG-PAB-1) is expressed transgenically under the control of a cell-specific promoter and mRNAs isolated by co-immunoprecipitation with anti-FLAG. This method has been used for microarray analysis of *C. elegans*body muscle cells and ciliated sensory neurons \[[@B87],[@B88]\]. We have now successfully used the mRNA tagging strategy to profile specific subsets of motor neurons from *C. elegans*larvae (SEV, RMF, J. Watson, S. Kim, P. Roy, DMM, unpublished data). Thus, in principle, it should now be possible to obtain an accurate gene expression profile for virtually any *C. elegans*cell throughout development.
UNC-4 motor neurons are sensitive to a wide range of neurotransmitters and peptidergic signals
----------------------------------------------------------------------------------------------
Acetylcholine (ACh) release at the DA neuromuscular junction is presumptively triggered by excitatory input from command interneurons. The strength of the DA cholinergic signal, however, may be strongly modulated by other cells that release neurotransmitters from distal locations. For example, dopamine is produced by 8 neurons, none of which are presynaptic to DA motor neurons \[[@B89]\]. Dopamine, however, is a potent regulator of cholinergic secretory activity in the ventral motor circuit. The dopamine effect is mediated in part by DOP-1, a G-protein coupled receptor (GPCR) \[[@B55]\]. We have confirmed enrichment of the *dop-1*transcript and also detected elevated levels of transcripts encoding GPCRs for acetylcholine and serotonin, additional neurotransmitters known to modulate cholinergic motor neuron activity via G-protein signaling pathways \[[@B18],[@B55]\]. Enrichment of a GABA metabotropic receptor transcript offers yet another mechanism for exogenous adjustment of neurotransmitter vesicular fusion in DA motor neurons. Indirect evidence indicates that acetylcholine release from ventral cord motor neurons may also be sensitive to neuropeptide signals from other neurons or muscle cells \[[@B73],[@B81]\]. We have established that *unc-4*::GFP motor neurons express elevated transcript levels for nine different GPCRs with significant homology to insect or mammalian neuropeptide receptors. This signaling complexity is further compounded by the enrichment of transcripts for 18 members of the serpentine GPCR-like family in *unc-4*::GFP neurons (Additional Files [9](#S9){ref-type="supplementary-material"}, [16](#S16){ref-type="supplementary-material"}). Ligands for this outlier group of GPCRs are unknown \[[@B90]\]. The picture emerging from these data is of a motor neuron festooned with multiple G-protein linked receptors each responding to a different class of neurotransmitter or peptidergic signal (Fig [8](#F8){ref-type="fig"}). In effect, these motor neurons are also functioning as a kind of sensory neuron in which disparate inputs are internally assessed to fine-tune output in concert with temporal requirements for locomotory activity.
::: {#F8 .fig}
Figure 8
::: {.caption}
######
Signaling components detected in *unc-4*::GFP motor neurons.
:::
:::
The microarray data also reveal multiple additional classes of receptors and ion channels through which the differentiation and function of *unc-4*::GFP motor neurons could be modulated by extracellular signals (Fig [8](#F8){ref-type="fig"}). Finally, we have detected enrichment of transcripts encoding TGF-β, wingless, and several classes of neuropeptides (Table [2](#T2){ref-type="table"}, Additional Files [9](#S9){ref-type="supplementary-material"}, [16](#S16){ref-type="supplementary-material"}). Thus, in addition to responding to a wide range of stimuli, *unc-4*::GFP motor neurons are also potentially capable of regulating the activities of other cells with a variety of different signals. If an organism as simple as *C. elegans*builds motor neurons with such sophisticated signaling and response mechanisms, it is tempting to speculate that neurons in other, more advanced species may have evolved even more complex pathways. It is likely, however, that the core signaling systems described here are also conserved. This prediction is underscored by our finding that approximately half of the enriched transcripts (537/1012) and 2/3 of EGs (4050/6217) detected in *unc-4*::GFP neurons have human homologs (BLAST ≤ e^-10^) (Additional Files [12](#S12){ref-type="supplementary-material"}, [13](#S13){ref-type="supplementary-material"}).
Applications of MAPCeL
----------------------
In addition to confirming expression of genes with known roles in *unc-4*::GFP motor neuron differentiation and function, the microarray data also uncovered strong candidates for new genes governing these events. For example, DA motor axons grow dorsally in response to a ventrally provided repulsive UNC-6/netrin guidance cue \[[@B15]\]. Recent work has shown that the receptor protein tyrosine phosphatase (RPTP), CLR-1, positively enhances this response \[[@B24]\]. As expected, we found that the *clr-1*transcript is enriched in the *unc-4*::GFP motor neuron data set. We also noted enrichment of *abl-1*, the *C. elegans*homolog of Abelson tyrosine kinase. By analogy to findings in *Drosophila*in which Abelson tyrosine kinase functions in opposition to RPTP-dependent axon guidance \[[@B91]\], we propose that ABL-1 antagonizes CLR-1 activity (Fig [6](#F6){ref-type="fig"}). This model predicts that either genetic ablation or RNAi of *abl-1*will suppress the DA motor axon guidance defects of *clr-1*mutants.
Another application of this strategy includes the identification of transcription factor target genes. A comparison of expression fingerprints of wildtype cells vs cells that are mutant for a specific transcription factor could reveal downstream genes \[[@B84]\]. For example, the UNC-4 transcription factor regulates axon morphology and synaptic strength in embryonically derived *unc-4*::GFP neurons \[[@B8]\]. UNC-4 also defines the specificity of synaptic inputs to postembryonically-derived VA motor neurons \[[@B7],[@B92],[@B93]\]. We have now used a combination of MAPCeL and mRNA tagging strategies to identify candidate genes regulated by UNC-4 in these cells (RMF, SEV, DMM, unpublished data). Gene regulatory motifs to which transcription factors bind may also be revealed as common cis-acting sequences in cohorts of co-regulated genes \[[@B94]\].
The *C. elegans*nervous system is composed of exactly 302 neurons. The morphology and connectivity for every neuron has been defined by serial section electron microscopy to generate a detailed wiring diagram for the entire network \[[@B2]\]. The *C. elegans*genome is similarly well defined. All 6 chromosomes are completely sequenced and the structure of over 20,000 genes described \[[@B5]\]. Unique combinations of these genes are likely to specify different classes of neurons and their differentiated traits. The problem now is to link the gene map with the wiring diagram. We believe that MAPCeL offers a powerful approach toward achieving this goal.
Conclusion
==========
We have described a new method, Micro-Array Profiling of *C. elegans*cells, or MAPCeL, for generating gene expression fingerprints of subsets of *C. elegans*neurons. Embryonic motor neurons marked with a reporter gene, *unc-4*::GFP, were isolated by FACS from primary cultures and profiled on the *C. elegans*Affymetrix array. We confirmed that microarray data generated by this approach reliably identify genes expressed by these motor neurons *in vivo*. We propose that MAPCeL can now be used to generate a gene expression map for the *C. elegans*nervous system.
Methods
=======
Cell Culture
------------
Embryonic cells were obtained using methods previously described \[[@B6]\]. Briefly, embryos were isolated from gravid adults following lysis in a hypochlorite solution. Intact embryos were separated from debris by flotation on 30% sucrose. Eggshells were removed by incubation in 0.5 ml chitinase (0.5 U/ml in egg buffer) for 45 minutes. Following resuspension in L-15 medium supplemented with 10% FBS (L15-10) and antibiotics, the embryos were dissociated by passage through a 5μm syringe filter (Durapore). Cells were plated on poly-L-lysine (0.01%, Sigma) coated single-well chambered coverglasses (Nalge Nunc International) at a density of \~10 million cells/ml and maintained in L15-10 media. Cells were incubated at 25°C in a humidified chamber. Wildtype (N2) cells were isolated and treated similarly.
FACS analysis
-------------
Sorting experiments were performed on a FACStar Plus flow cytometer (Becton Dickinson, San Jose, CA) equipped with a 488 nm argon laser. Emission filters were 530 ± 30 nm for GFP fluorescence and 585 ± 22 nm for PI fluorescence. The machine was flushed with egg buffer \[[@B6]\] prior to sorting to enhance viability. 2 μm fluorescent beads were used to calibrate light scattering parameters for the relatively small size of *C. elegans*embryonic cells. Cells were sorted at a rate of 4000--5000 cells per second through a 70 μm nozzle.
Immediately prior to sorting, supernatant from the 24 hour cultures was removed and discarded. 1 ml of egg buffer was added to the chamber coverglass. Cells are loosely adherent to poly-L-lysine and can be easily dislodged with gentle pipetting. 3 ml of egg buffer + cells were drawn into a 3 cc syringe and the suspension filtered with a 5 μm Durapore syringe filter. Propidium Iodide (PI) was added to the cell suspension at a final concentration of 5 μg/ml prior to sorting. Autofluorescence levels were established by flow cytometry of cells isolated from wildtype (i.e. non-GFP) embryos (See Fig [2A](#F2){ref-type="fig"}). Next, wildtype cells stained with PI were used to define the sorting gate for damaged cells. GFP+ cells containing no PI were sorted to establish the intensity range of GFP fluorescence. Finally, *unc-4*::GFP cells stained with PI were gated (Fig [2B](#F2){ref-type="fig"}) using the parameters established above. The sorting gate for size and granularity (Fig [2C](#F2){ref-type="fig"}) was empirically adjusted to exclude cell clumps and debris and to achieve \~90% enrichment for GFP-labeled cells. *unc-4*::GFP cells were collected in a 15 ml conical tube containing 1 ml of L15-10 media. Cells were pelleted using low-speed cenrifugation (300 × g) and either plated on peanut lectin-coated slides for visualization \[[@B6]\] or used for RNA isolation (see below). Reference cells were obtained from 1 day old cultures of embryonic blastomeres isolated from the non-GFP wildtype strain (N2). In this case, all viable cells (i.e. non-PI stained) were collected by FACS for RNA isolation.
RNA isolation, amplification, and hybridization
-----------------------------------------------
RNA was prepared from FACS-isolated *unc-4*::GFP cells for comparison to RNA from the wildtype reference strain (N2). Cells were pelleted using low-speed centrifugation (300 × g). The supernatant was removed and RNA was extracted with a micro-RNA isolation kit (Stratagene) using the recommended volumes for 1 million cells. Typical yields were 1 pg total RNA/cell. 100 ng of total RNA was subjected to 2 rounds of amplification, as described in the Affymetrix GeneChip Eukaryotic Small Sample Target Labelling Protocol, with the following modifications. 100 ng (5 pmol) of T7-dT primer (5\'-GGCCAGTGAATTGTAATAC GACTCACTATAGGGAGGCGG-(dT)~24~-3\') was used as opposed to the recommended 100 pmol. RNA cleanup was achieved using the RNeasy mini kit (Qiagen); 300 μl of 100% ethanol (final concentration = 40% ethanol) was added to the sample prior to absorption to the column matrix. Eluate was passed through the column 2× prior to washing to improve yields. The BioArray High Yield RNA Transcript Labeling Kit (Enzo) was used to biotinylate the sample in the second round of amplification. 10--15 μg of labeled aRNA (amplified RNA) was fragmented and hybridized to the Affymetrix *C. elegans*chip according to the Affymetrix Expression Analysis Technical Manual. The Agilent Bioanalyzer was used to assess RNA quality prior to labeling and to confirm fragmentation (\<200 bp) before hybridization.
Data analysis
-------------
The commercially available *C. elegans*Affymetrix array was used for all experiments. This chip was designed using the December 2000 genome sequence. All probe set information is available at <http://www.affymetrix.com> as well as <http://www.wormbase.org>. *unc-4*::GFP neurons were profiled in triplicate; baseline data (all cells) were obtained from four independent experiments with wildtype embryonic cells. Hybridization intensities for each experiment were scaled in comparison to a global average signal from the same array (A complete list of Affy normalized values can be found in [Additional file 2](#S2){ref-type="supplementary-material"}) \[[@B95]\]. Expressed transcripts were initially identified on the basis of a \"Present\" call in a majority of experiments (2/3 for *unc-4*::GFP and 3/4 for wildtype cells) as determined by Affymetrix MAS 5.0 (see below) (Additional Files [4](#S4){ref-type="supplementary-material"}, [5](#S5){ref-type="supplementary-material"}). In this approach, a Mismatch (MM) value for each feature is compared to a Perfect Match (PM) value to estimate non-specific binding. This strategy, however, tends to arbitrarily exclude low intensity signals in which PM and MM values may be comparable \[[@B96],[@B97]\]. To avoid this bias in the detection of transcripts that might be differentially elevated in the *unc-4*::GFP data set, intensity values were normalized using RMA (Robust Multi-Array Analysis) available through GeneTraffic (Iobion) in which the MM values are not considered ([Additional file 3](#S3){ref-type="supplementary-material"}) \[[@B96],[@B97]\]. Comparisons of RMA normalized intensities for *unc-4*::GFP vs reference cells were statistically analyzed using Significance Analysis of Microarrays software (SAM, Stanford) \[[@B98],[@B99]\]. A two-class unpaired analysis of the data was performed to identify genes that differ by ≥ 1.7-fold from the wildtype reference at a False Discovery Rate (FDR) of ≤ 1%. These genes were considered significantly enriched ([Additional file 9](#S9){ref-type="supplementary-material"}). This analysis also identified \~1600 transcripts that are depleted (1.7×, ≤ 1% FDR) in *unc-4*::GFP cells vs the wildtype reference ([Additional file 10](#S10){ref-type="supplementary-material"}). Although 729 of these transcripts are also scored as \"present\" in the *unc-4*::GFP motor neuron dataset, we attribute their detection to high expression in the small fraction (\~10%) of non-GFP cells contaminating this preparation (see above). Therefore, we excluded all 729 of these wildtype-enriched transcripts from the list of present calls in the *unc-4*::GFP motor neuron data set ([Additional file 6](#S6){ref-type="supplementary-material"}). Finally, to compute the overall sum of Expressed Genes (EGs) in the *unc-4*::GFP data set we restored 118 *unc-4::*GFP-enriched genes that were initially excluded from the present list due to high mismatch signals. These considerations produce a final list of 6,217 genes that are detected in *unc-4*::GFP motor neurons ([Additional file 7](#S7){ref-type="supplementary-material"}). (see Logic Tree, [Additional file 17](#S17){ref-type="supplementary-material"}).
Annotation of datasets
----------------------
A Wormbase mirror was established by downloading code and databases from <http://www.wormbase.org>. Using the acedb perl module, an annotation script was generated that queries the wormbase mirror. Affymetrix IDs have been mapped to specific transcripts in wormbase. Text files containing Affy IDs (one per line) and cosmid names are input into the script which then searches the wormbase mirror and matches Affy ID/cosmid name to a specific transcript. Cosmid names are used for this search when Affy IDs have not been mapped in wormbase. This information is used to acquire other linked annotations (i.e. KOG, common name, RNAi phenotype, Expression data, Kim mountain data and Gene Ontology, etc.).
*In litero*analysis
-------------------
An extensive literature search was performed using Textpresso <http://www.textpresso.org>. The keywords \"DA motor neurons\" generated a list of 68 citations, a similar search was conducted using the keywords \"I5 pharyngeal neuron\" and \"SAB neurons\" that detected an additional 21 citations. Expression patterns on wormbase were also searched using the \"Cell identity\" function to identify genes with documented expression in DA, SAB or I5. A list of 27 genes with documented expression in DA motor neurons, the I5 pharyngeal neuron and the SAB neurons was compiled from this information ([Additional file 11](#S11){ref-type="supplementary-material"}).
Strains
-------
Nematode strains were maintained at 20--25°C using standard culture methods \[[@B100]\]. The wildtype strain was N2. Transgenic lines carrying promoter GFP fusions are listed in [Additional file 1](#S1){ref-type="supplementary-material"}.
Generating transgenic promoter GFP strains
------------------------------------------
*twk-30*::GFP (25 ng/ul) was microinjected with the *myo-3*::dsRed2 marker (25 ng/ul) \[[@B101]\]. Other transgenics were generated by biolistic transformation with promoter::GFP constructs from the Promoterome project ([Additional file 1](#S1){ref-type="supplementary-material"}). Primer sequences for \"promoterome\" constructs can be found at <http://vidal.dfci.harvard.edu/promoteromedb>\[[@B102]\]. Microparticle bombardment was conducted \[[@B103]\] in a BioRad Biolistic PDS-1000/He equipped with the Hepta Adapter. Gold beads (1 micron) were coated with DNA at 1 ug/ul. 100 mm NGM plates were seeded with a monolayer of \~100,000 L4/adult *unc-119 (ed3)*animals. For each construct, 1 \'shot\' was performed using a 1550 psi rupture disk at 28 inches of Hg vacuum. After a 1 hr recovery period, animals were washed from the plates with 7 ml M9 buffer and transferred to 7 NGM plates (1 ml/plate). Animals were grown at 20°C for 1 week. To pick transgenic animals, one-half of the plate was \'chunked\' and added to a new 100 mm NGM plate; animals with wildtype movement were picked to 60 mm NGM plates and allowed to self. Worms derived from separate plates were considered independent lines; at least 2 lines were obtained for each construct.
Microscopy
----------
Transgenic animals and cultured cells were visualized by differential interference contrast (DIC), or epifluorescence microscopy using either a Zeiss Axioplan or Axiovert compound microscopes. Images were recorded with CCD cameras (ORCA I, ORCA ER, Hamamatsu Corporation, Bridgewater, NJ). Some images were recorded on a Zeiss 510 META confocal microscope.
Authors\' contributions
=======================
RMF developed the FACS protocol, RNA amplification methods, performed *unc-4*::GFP microarray experiments, analyzed microarray data, generated/scored GFP reporters, and helped draft the manuscript. SEV helped develop RNA amplification and bombardment protocols, generated/scored GFP reporters and helped draft the manuscript. SJB generated the reference dataset. CS and KO wrote Perl scripts used to annotate datasets, make comparisons, and identify Present genes. JM offered advice on statistical methods. DD and MV provided unpublished GFP reporter constructs as part of the Promoterome project. DMM collected images of GFP lines in the confocal microscope, oversaw all aspects of the project and helped draft the manuscript.
Supplementary Material
======================
::: {.caption}
###### Additional File 1
***promoter*::GFP fusions**Strain names, source and references for GFP reporters used to validate microarray data.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 2
**Intensity Values Following Global Normalization (Affymetrix MAS 5.0)**Raw intensity values were scaled against a global average intensity value calculated for each chip using Affymetrix MAS 5.0. Column A lists Affymetrix IDs for 22,548 features on the chip. These are indexed by rows with normalized intensity values for each independent hybridization and its Present/Absent (P/A) call. Columns labeled DMR 28, 30, 32, 34 include intensity values for hybridizations using RNA from all N2 (wildtype) cells. Columns labeled DM39, 51, 59 include the experimental hybridizations using RNA from FACS-isolated *unc-4*::GFP neurons.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 3
**Intensity values after RMA normalization.**Raw data from [Additional file 2](#S2){ref-type="supplementary-material"} were normalized with Robust Multi-array Analysis (RMA) (GeneTraffic, version 2.8) (Iobion). Normalized values are listed for the 22,548 features on the Affymetrix *C. elegans*chip. Affy IDs are listed in column A. Columns C-I show RMA normalized values for individual replicates. DMR 28, 30, 32, 34 are the baseline hybridizations using RNA from all N2 (wildtype) cells. DM39, 51, 59 are the experimental hybridizations using RNA from FACS-isolated *unc-4*::GFP neurons.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 4
**N2 (wildtype) Expressed Genes (EGs)**9103 genes were called \"Present\" by MAS 5.0 in 3 out of 4 N2 hybridizations and are listed here as EGs. Affymetrix ID, Cosmid Name and common names (Columns A-C respectively) are given for each gene. Column D contains KOG descriptions and Column E indicates the dataset in which an individual gene is present.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 5
***unc-4*::GFP Present Genes**6828 genes were called \"Present\" by MAS 5.0 in 2 out of 3 *unc-4*::GFP hybridizations and are listed here.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 6
***unc-4*::GFP Present Genes minus N2 Enriched Genes**The list of *unc-4*::GFP \"present\" calls in [Additional file 5](#S5){ref-type="supplementary-material"} is likely to include transcripts that are highly expressed in the small fraction of contaminating non-GFP cells (\~10%). These transcripts were removed from the list of *unc-4*::GFP Present calls ([Additional file 5](#S5){ref-type="supplementary-material"}) by subtracting genes that are highly enriched in N2 cells ([Additional file 10](#S10){ref-type="supplementary-material"}). The balance of 6099 genes is listed here with Affy ID, Cosmid Name, Common Name and KOG description.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 7
***unc-4*::GFP Expressed Genes (EGs)**RMA normalization ignores the MAS 5.0 P/A calls and therefore does not discard some of the \"absent\" genes that are excluded from [Additional file 6](#S6){ref-type="supplementary-material"} (*unc-4*::GFP Present Genes). As a result, the *unc-4*::GFP enriched genes ([Additional file 9](#S9){ref-type="supplementary-material"}) identified by SAM statistics actually includes 118 genes that are not listed in Additional Files [5](#S5){ref-type="supplementary-material"} and [6](#S6){ref-type="supplementary-material"}. These 118 additional *unc-4*::GFP enriched genes were therefore restored to the list of *unc-4*::GFP Present Genes in [Additional file 6](#S6){ref-type="supplementary-material"} to provide a more accurate list of genes (6217) that are actually expressed in *unc-4*::GFP cells. These *unc-4*::GFP Expressed Genes (EGs) are listed with Affy ID, Cosmid Name, Common Name and KOG description.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 8
**Expressed Genes (EGs)**EGs from N2 ([Additional file 4](#S4){ref-type="supplementary-material"}) and from *unc-4*::GFP ([Additional file 7](#S7){ref-type="supplementary-material"}) cells were combined to detect a total of 10,071 unique EGs.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 9
***unc-4*::GFP Enriched Genes**Transcripts elevated 1.7 fold above baseline at a False Discovery Rate (FDR) ≤1% were considered enriched. 1012 transcripts met these criteria and are listed according to statistical rank (Column A). Annotation includes Affy ID, Cosmid Name, Common Name and KOG description (Columns B-E); SAM score, Fold Change and q-value are listed (Columns F-H). We attribute the high ranking (2) of *dpy-20*(T22B3.1) to its use as a coselectable marker in the generation of the *unc-4*::GFP transgenic line \[[@B104]\].
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 10
**N2 Enriched Genes**1586 genes are enriched 1.7 fold (FDR ≤1%) in N2 cells compared to *unc-4*::GFP neurons. These are listed according to statistical rank (Column A). Annotation includes Affy ID, Cosmid Name, Common Name and KOG description (Columns B-E); SAM score, Fold Change and q-value are listed (Columns F-H).
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 11
**Genes previously known to be expressed in *unc-4*::GFP neurons.**Published literature and wormbase <http://www.wormbase.org> were searched to identify 27 genes that are expressed in embryonic neurons that also express *unc-4*::GFP (I5, DA, SAB).
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 12
***unc-4*::GFP motor neuron enriched genes with human homologs.**537 Enriched genes ([Additional file 9](#S9){ref-type="supplementary-material"}) and human genes listed here have BLAST scores ≤ e^-10^.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 13
***unc-4*::GFP EGs with human homologs.**4050 Expressed genes ([Additional file 8](#S8){ref-type="supplementary-material"}) and human genes listed here have BLAST scores ≤ e^-10^.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 14
**Summary of *unc-4*::GFP enriched and expressed (EGs) transcripts in selected categories.**This file is an expanded version of Table [2](#T2){ref-type="table"} (see text) in which EGs are added to the list of enriched genes. Genes are organized as in Table [2](#T2){ref-type="table"} according to molecular function (KOG description, other description <http://www.wormbase.org>). Statistical rank is indicated for each enriched transcript and \"EG\" designates transcripts that are EGs.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 15
**Genes Expressed only in *unc-4*::GFP neurons.**Affy ID, Cosmid Name, Common Name and KOG description for 968 genes that are expressed in *unc-4*::GFP neurons. These genes are not detected in the reference (N2) dataset.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 16
**Gene families represented in *unc-4*::GFP neurons.**A comprehensive description of neuronal transcripts organized according to gene family.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 17
**Logic tree of data analysis methods.**
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
We thank Anne Hart, Chris Li, Mike Nonet, Larry Salkoff, Cori Bargmann, Harald Hutter, Cynthia Kenyon, Kevin Hamming, Colin Thacker, Terry Snutch, Roger Pocock, and Allison Woollard for GFP reporter plasmids and transgenic lines; Cathy Alford and Jim Price of the VU Flow Cytometry Special Resource Center; Joseph Watson and Jennifer Ross for assistance with scoring transgenic animals, and Millet Treinin, David Greenstein and members of the Miller lab for comments on the text. Confocal images were obtained in the Cell Imaging Shared Resource (CA68485, DK20593, DK58404, HD15052 and EY08126). Microarray experiments were performed in the Vanderbilt Microarray Shared Resource which is supported by the Vanderbilt Ingram Cancer Center (P30 CA68485), the Vanderbilt Diabetes Research and Training Center (P60 DK20593), the Vanderbilt Digestive Disease Center (P30 DK58404) the Genomics of Inflammation Program Project Grant (1 P01 HL6744-01), and the Vanderbilt Vision Center (P30 EY08126). Additional support for microarray experiments was provided by DK58749 (PI, A. George). Supported by NIH grants to D.M.M. (R01 NS26115 and P01 DK58212), R.M.F. (F31 NS046923), S.E.V. (F31 NS043068), and M.V. (R33 CA097516).
|
PubMed Central
|
2024-06-05T03:55:55.240241
|
2005-3-21
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079822/",
"journal": "BMC Genomics. 2005 Mar 21; 6:42",
"authors": [
{
"first": "Rebecca M",
"last": "Fox"
},
{
"first": "Stephen E",
"last": "Von Stetina"
},
{
"first": "Susan J",
"last": "Barlow"
},
{
"first": "Christian",
"last": "Shaffer"
},
{
"first": "Kellen L",
"last": "Olszewski"
},
{
"first": "Jason H",
"last": "Moore"
},
{
"first": "Denis",
"last": "Dupuy"
},
{
"first": "Marc",
"last": "Vidal"
},
{
"first": "David M",
"last": "Miller"
}
]
}
|
PMC1079823
|
Background
==========
Flowering plants have developed from a common ancestor with mosses, liverworts, ferns, and gymnosperms over the last 450 million years \[[@B1]\]. Most recent angiosperms do not closely resemble their ancestors, as known from the fossil record. Quite a few gymnosperms (like *Ginkgo*or *Cycas*) still resemble the plants known from the fossil record, and this is even more true for \"lower\" land plants, namely mosses, liverworts and ferns \[[@B2],[@B3]\]. In addition, mosses seem to evolve with a slow molecular clock \[[@B4]\]. So, if these plants appear to be more ancient than modern flowering plants as measured by morphological means and mutation rate, does this also hold true for how they employ their genetic system?
A lot of comparative studies on protein encoding genes have already been carried out within and between the two major groups of flowering plants, mono- and dicotyledons, with rice (*Oryza sativa*) and *Arabidopsis thaliana*as the most prominent examples. Currently more than two million *Liliopsida*(monocotyledons) EST are publicly available, the corresponding number for the *Magnoliophyta*(dicotyledons) even exceeds four million sequences. However, sequence information for other plant phyla is still scarce. There are only about 160,000 EST sequences available of both *Coniferophyta*(part of the gymnosperms) and *Chlorophyta*(green algae), 130,000 from *Bryophyta*(mosses) and 3,700 from *Filicophyta*(ferns) (all numbers from Genbank). For the moss *Physcomitrella patens*, more than 102,000 nucleic acid sequences (mainly EST) are publicly available to date. This \"ancient\" land plant therefore is an ideal candidate to unravel some details about how simple plants encode proteins and whether they do so in a different manner from \"modern\" plants, as represented by the monocotyledon rice and the dicotyledon *Arabidopsis*in this study.
*Physcomitrella*is increasingly being used as a model plant because of its unrivalled capability among plants to include ectopic DNA into its genome by means of homologous recombination (see e.g. \[[@B5]-[@B7]\]), thus enabling gene replacement in a straight forward manner. As in all mosses, the haploid gametophyte is the dominant generation in the heteromorphic life cycle. In this respect the moss is different from seed plants (gymnosperms and flowering plants), in which the polyploid sporophyte dominates the life cycle. It has been argued before \[[@B8],[@B9]\] that the set of genes of the respective dominant generation is equivalent, while a large proportion of moss transcripts cannot currently be assigned a putative function. These \"orphan\" genes might encode functions that are specific to mosses and are not present in other taxonomic groups. Besides species-specific orphan genes, mosses might also possess retained genes, that have been lost in seed plants during evolution. Both types of genes are candidates to encode functions that make mosses unique in terms of physiology and metabolism. For example, *Physcomitrella*exhibits increased tolerance towards abiotic stresses \[[@B10],[@B11]\], uses proteins derived from the same gene in different cellular compartments by dual targeting \[[@B12],[@B13]\] and displays secondary metabolite pathways not known in seed plants \[[@B14]-[@B16]\]. In this study, following up on the initial analyses by Nishiyama et al. \[[@B8]\], we aimed to increase our knowledge of the moss transcriptome.
Results and discussion
======================
Comparative BLAST searches
--------------------------
Around 30% of the *Physcomitrella*ORF have homologs in both rice and *Arabidopsis*transcriptome whereas 80% of the *Arabidopsis*genes have a homolog in rice and 40% of the rice genes in *Arabidopsis*(Fig. [1](#F1){ref-type="fig"}). Although these numbers are lower than the actual amount of sequence homologs because of filtering (see below), they demonstrate that *Physcomitrella*contains a lot of as yet unknown protein encoding genes that might be specific for mosses. A homology search against the taxprot dataset (Table [1](#T1){ref-type="table"}, Fig. [2](#F2){ref-type="fig"}) reveals that 45.8% of the predicted moss ORF find a query in plants (E-value threshold 1E-4), after rigorous filtering 28.1% remain and 21.7% are non-redundant, i.e. do not match multiple subject sequences. The rigorous filtering (see methods for details) for true homologs thus necessarily decreases the set of available sequences, so that false conclusions are not made based on comparison of non-homologous sequences.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Comparative BLAST searches between *Arabidopsis*, rice and moss**. Comparative BLAST searches of the *Arabidopsis*(At, yellow), rice (Os, cyan) and *Physcomitrella*(Pp, green) transcriptomes. Each search was done with the respective sets once as query and once as search space (subject). The area of the circles represents the percentage of the query/subject sequence space that yielded filtered hits.
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Taxonomic constitution of the taxprot dataset
:::
**taxonomic group** **txid** **\# of sequences**^1^
---------------------------------------- ---------- ------------------------
Metazoa 33208 862,420
Fungi 4751 184,282
Viridiplantae (plants and green algae) 33090 293,156
Non-green algae^2^ 21,889
Other Eukaryotes^3^ 49,732
Eubacteria (without Cyanobacteria) 2 1,386,089
Cyanobacteria 1117 94,920
Archaea 2157 122,394
Viruses 10239 331,246
**Total** **3,346,128**
^1^Genbank amino acid sequences as of 2004--04--07, NCBI taxon ids are shown under \"txid\", all taxonomic crown groups with at least 100 sequence members were used; ^2^Cercozoa \[136419\], Cryptophyta \[3027\], Euglenozoa \[33682\], Glaucocystophyceae \[38254\], Haptophyceae \[2830\], Rhodophyta \[2763\], Stramenopiles \[33634\]; ^3^Acanthamoebidae \[33677\], Alveolata \[33630\], Diplomonadida \[207245\], Entamoebidae \[33084\], Heterolobosea \[5752\], Jakobidae \[143015\], Mycetozoa \[142796\], Parabasalidea \[5719\]
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**BLAST hits of *Physcomitrella*protein genes against the taxprot dataset**. a) Absolute number of hits against different taxonomic groups. b) Amount of non-redundant hits as percentage of the respective sequence space.
:::
:::
Full-length transcripts
-----------------------
The total number of clusters after EST clustering do not equal the number of protein encoding genes. This is mainly due to partial (as opposed to full-length) transcripts, i.e. a single gene is represented by more than one sequence because they do not overlap. How many of the clustered public (PPP) transcripts represent full-length coding sequences? Of those sequences that yield a filtered hit against plant mRNAs, 7.9% are putatively full-length. Of those, 53.9% start with Methionine, of the latter, 32.4% contain no X (X represents an indeterminable codon, which can be included by the ORF prediction).
Orthologs, paralogs and mapping
-------------------------------
The filtered hits against the *Arabidopsis*transcriptome (1,994 in total) were divided into non-redundant orthologs (722) and paralogs (1,015). As *Arabidopsis*orthologs, we defined all those sequences for which the initial subject matches the query in the reverse search (reciprocal hit). Paralogs were defined as those sequences for which this rule does not apply. This method of detecting potential orthologs has been used previously for cross-species comparisons (e.g. \[[@B17],[@B18]\]). The three sequence sets were mapped against the *Arabidopsis*chromosomes using BLAST (Fig. [3](#F3){ref-type="fig"}). The distribution pattern clearly reveals the centromeric regions but otherwise does not display significant differences. Although there are some chromosome and sequence set-specific differences in the rate of hits per Mbp, these are not significant as measured by absolute average deviation.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Mapping of *Physcomitrella*transcripts to the *Arabidopsis*chromosomes**. Mapping of filtered BLAST hits (grey), paralogs (red) and orthologs (green) against the five *Arabidopsis*chromosomes (left to right / top to bottom). a) Hits per Mbp; error bars: average absolute deviation (AAD); column 6: mean values. b) Graphical representation using a finer granularity (100 kbp), each vertical step represents one hit.
:::
:::
Taxonomic distribution and retained genes
-----------------------------------------
The highest total number of non-redundant, filtered BLAST hits is (as expected) derived from the plant subset of the taxprot dataset (Table [1](#T1){ref-type="table"}, Fig. [2a](#F2){ref-type="fig"}), followed by the animal and fungi subsets. When looking at the hits as percentage of the search space size (Fig. [2b](#F2){ref-type="fig"}), it becomes evident that quite a proportion of the sequence space of lower eukaryotes (including non-green algae) is covered. The comparatively high coverage of this \"ancient\" gene space suggests that mosses share many specialized genes with unicellular organisms.
134 *Physcomitrella*ORF have their best BLAST hit not among plants (Fig. [4a](#F4){ref-type="fig"}). Consequently, these are candidates for horizontal gene transfer or, more likely, retained genes that were lost in seed plants during evolution. We had a closer look at those 57 transcripts which are specific to a single taxonomic group, namely bacteria, cyanobacteria, animals or fungi (unique hits). For 25 of those, a putative function could be assigned manually (Fig. [4b](#F4){ref-type="fig"}, Table [2](#T2){ref-type="table"}). The broad functional categories of these taxon-specific retained genes are to some extent unevenly distributed. Whereas transport associated proteins are found solely among fungi, signal transduction gene products are found in both bacteria and animals. Transport and metabolism associated gene products support the wealth of secondary pathways found in moss (e.g., \[[@B14],[@B15],[@B19]-[@B22]\]), whereas the signal transduction genes also separate the moss from seed plants in this regard. Of special interest are two other functional categories among these candidate retained genes: cytotoxicity and nucleic acid modification. A broad range of cytotoxic abilities might explain why mosses can survive in moist environments mainly unplagued by microbial parasites, without the protection of a cuticula. Furthermore, it is, up until now, puzzling why *Physcomitrella*is able to integrate ectopic DNA into the genome by homologous recombination with an extraordinarily high rate \[[@B23],[@B24]\] so far only found in bacteria and yeast, but in no other plant or any animal. Hints to unravel this mystery might be found in the presence of genes involved in DNA repair, binding and modification, as we discovered during this research.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Retained genes in moss: taxonomic distribution and functional categories**. a) *Physcomitrella*transcripts which have their best BLAST hit not among plants, divided by taxonomic category, further subdivided into specific hits (unique to a single taxonomic group -- yellow) and those that could be assigned a putative function by means of homology searches (green). b) Distribution of functional categories among those taxonomic groups that yielded unique hits.
:::
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Functional annotation of retained genes into broad functional categories, assembled transcripts can be retrieved via <http://www.cosmoss.org>.
:::
**Pp transcript (putative retained gene)** **taxonomic group** **homolog** **broad functional category (potential)** **functional category** **details**
-------------------------------------------- --------------------- ------------------------------------------------------ ------------------------------------------- --------------------------- -----------------------------------------------------------------------------------
PPP\_2925\_C1 bacteria Membrane-bound lytic murein transglycosylase B cytotoxicity murein degradation murein-degrading enzyme, may play a role in recycling of muropeptides during cell
BJ203770 bacteria putative protease cytotoxicity protease
PPP\_4234\_C1 metazoa cytolysin I cytotoxicity cytotoxicity involved in pore-formation
PPP\_3510\_C1 cyano RTX toxins and related Ca2+-binding proteins cytotoxicity cytotoxicity
PPP\_1172\_C1 bacteria Enoyl-CoA hydratase/carnithine racemase metabolism fatty acid metabolism
PPP\_6629\_C1 bacteria mannosylglycerate synthase metabolism sugar metabolism
PPP\_5746\_C1 metazoa L-kynurenine 3-monooxygenase Fpk metabolism amino acid metabolism
PPP\_8479\_C1 metazoa COMMD2 metabolism copper metabolism COMM (copper metabolism MURR1) domain containing 2
BJ173412 metazoa ubiquitin metabolism protein metabolism ribosomal protein in C. elegans dehydrogenases
PPP\_3987\_C1 fungi MNN9 metabolism N-glycosylation
PPP\_6514\_C1 cyano oxidoreductase metabolism energy metabolism related to aryl-alcohol dehydrogenases
PPP\_11394\_C1 bacteria homolog of eukaryotic DNA ligase III nucleic acid binding / modification DNA repair
BJ191550 bacteria formamidopyrimidine-DNA glycosylase nucleic acid binding / modification DNA repair
BJ160862 metazoa Osa1 nuclear protein nucleic acid binding / modification DNA binding chromatin regulation
BJ582496 cyano SAM-dependent methyltransferase nucleic acid binding / modification nucleic acid modification
PPP\_2586\_C1 bacteria CarD protein signal transduction DNA binding leucine zipper transcription factor, light- and starvation-induced response
PPP\_3689\_C1 bacteria serine/threonine protein kinase signal transduction signal transduction
BJ172132 bacteria serine/threonine protein kinase signal transduction signal transduction
PPP\_460\_C1 metazoa HLA-B-associated transcript signal transduction signal transduction
PPP\_1041\_C1 metazoa calcium/calmodulin-dependent protein kinase II delta signal transduction signal transduction
PPP\_6326\_C1 metazoa tumor suppressor tout-velu signal transduction signal transduction involved in diffusion of hedgehog
PPP\_11399\_C1 metazoa dual-specificity tyrosine phosphatase YVH1 signal transduction signal transduction Non-receptor class dual specificity subfamily
PPP\_184\_C1 fungi high-affinity iron permease transport transport high affinity iron uptake
PPP\_7115\_C2 fungi uric acid-xanthine permease transport transport belongs to the Xanthine/Uracil oermeases family
PPP\_11191\_C1 fungi inorganic phosphate transporter transport transport probable inorganic phosphate transporter; yeast pho99 homologue
:::
Gene structure and splice sites
-------------------------------
The average rate of introns per gene (\~5) is the same in *Physcomitrella*, *Arabidopsis*and human \[[@B25]\]. The average *Physcomitrella*intron (252 bp) is longer than those of *Arabidopsis*(146 bp) and shorter than the typical human intron (740 bp). Furthermore, the *Physcomitrella*intron is longer than the exon, whereas in *Arabidopsis*it is the other way round. While the size distribution of *Arabidopsis*introns is centered around 70 bp, the longer moss introns are mainly clustered around 180 bp (data not shown). This fits the weak correlation of intron length and genome size generally found in eukaryotic organisms \[[@B26]\]. Intron positions of close homologs between *Physcomitrella*and *Arabidopsis*are generally conserved (e.g., \[[@B27]\]).
The *Physcomitrella*G/C content of 40% in the intron and 50% in the exon differs significantly from that of *Arabidopsis*; 33% and 44%, respectively. Generally, *Physcomitrella*introns contain more thymine (T) than the exons. In terms of mononucleotide composition, T is overrepresented in the intron and C is underrepresented in the exon. In terms of dinucleotides, there is a significant overrepresentation of TT in the introns. Outstanding trinucleotide usage are the overrepresented TTT in the intron and the stop codon TGA in the exon, while the other two stop codons TAA and TAG are underrepresented in the moss.
A visualisation of the *Physcomitrella*donor and acceptor sites is shown in figure [5a](#F5){ref-type="fig"}. Comparison of the *Arabidopsis*-trained Netplantgene \[[@B25]\] and the *Physcomitrella*-trained svmsplice (Fig. [5b](#F5){ref-type="fig"}) reveals a better overall performance of the support vector machine. Although Netplantgene exhibits a high recall, precision is low, which is due to the large amount of false positive predictions. Svmsplice predicts a lower rate of true positives (thus lower recall), however, precision is much better. The mean values of recall and precision for both donor and acceptor site are higher for svmsplice and thus make it the method of choice for accurate prediction of *Physcomitrella*splice sites.
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Splice site sequence logos and efficiency of splice site prediction**. a) Sequence logos of *Physcomitrella*donor and acceptor sites. b) Prediction performance of Netplantgene and svmsplice for *Physcomitrella*splice sites. TP = true positive, FN = false negative, FP = false positive, measured on the lefthand (%) axis. Recall (sensitivity) = tp/(tp+fn), precision = tp/(tp+fp), measured on the righthand axis.
:::
:::
Composition of coding sequences and codon usage
-----------------------------------------------
Significant differences in codon fraction usage for the three above mentioned sequence subsets (*Arabidopsis*orthologs and paralogs, retained genes) when compared with the averaged codon usage in *Physcomitrella*and *Arabidopsis*are shown in Fig. [6a](#F6){ref-type="fig"}. The Average G/C content of the *Arabidopsis*CDS is \~43%, whereas it is \~50% for *Physcomitrella*(Table [3](#T3){ref-type="table"}). It might be argued that the EST-based estimation of G/C content in *Physcomitrella*is too high because of potential decay of AT-rich sequences \[[@B28]\]. However, when calculating the G/C content for all available 399 full-length CDS from Genbank, the percentage value is also \~50% (50.67%). This rate is also found in the retained genes and the *Arabidopsis*paralogs (Table [3](#T3){ref-type="table"}), whereas the ortholog fraction has a significantly lower G/C content of \~49%, i.e. towards the *Arabidopsis*nucleotide composition. Codon bias in *Physcomitrella*is positively correlated with gene expression level and G/C content of the CDS \[[@B29]\]. It was argued that weak natural selection for translational efficiency is the driving force behind codon bias in the moss rather than mutational bias. Given the G/C rate of 50% in the CDS, a mutational bias indeed seems unlikely.
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**Trinucleotide frequencies and codon usage**. a) The averaged *Physcomitrella*codon fraction usage measured as percentage of the total amount of counted codons is shown as grey diamonds, including a margin of 2× average absolute deviation (AAD, error bars), in comparison with *Arabidopsis*(yellow circles). Significantly deviating codons of the sequence subsets are presented as colored circles, namely retained genes (blue), paralogs (red) and orthologs (green). b) The effective number of codons (enc) for *Physcomitrella*(green) and *Arabidopsis*(yellow) as a range distribution scatter plot (y axis: % of analysed genes) and as averaged values (horizontal bar chart; error bars: standard deviation).
:::
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Codon usage of *Physcomitrella*retained genes, orthologs and paralogs
:::
**sequence set** **\# bases** **mean \# of each triplet** **G/C content** **\# significant codon usage changes** **codon usage towards At** **codon usage away from At** **codon over represented** **codon under represented** **significant changes per aa**
------------------ -------------- ----------------------------- ----------------- ---------------------------------------- ---------------------------- ------------------------------ ---------------------------- ----------------------------- --------------------------------
At mRNAs 10,755,859 56,020 43.32 n.a. n.a. n.a. n.a. n.a. n.a.
Pp ORFs 7,638,122 39,782 49.94 n.a. n.a. n.a. n.a. n.a. n.a.
retained genes 77,998 406 50.30 7 1 **6** 2 5 Phe under represented
paralogs 1,115,937 5,812 50.07 3 1 2 1 2 none
orthologs 953,293 4,965 **49.04** 10 **8** 2 4 6 Pro under reprensented
sum 10 10 7 13
The predicted *Physcomitrella*ORF were used as background to check for significant changes in percentage codon fraction usage in the orthologs, paralogs and retained genes (best BLAST hit not among plants). In case of significant deviation (two times average absolute deviation -- AAD) from the total set, the direction of the change relative to the *Arabidopsis*codon usage was checked. Significant deviations are shown enlarged, At = *Arabidopsis thaliana*, Pp = *Physcomitrella patens*.
:::
In retained genes, Phenylalanine codons are underrepresented, in the orthologs this is the case for Proline codons. As can also be seen from the G/C content drift mentioned above, the majority of deviating codons in the orthologs changed in the direction of the *Arabidopsis*percentage usage. For retained genes, it is the direct opposite: the significantly deviating codons in these genes point away from the *Arabidopsis*codon fraction usage. Orthologs are thought to be functionally equivalent across taxonomic groups. The common ancestor of land plants might have had a G/C content similar to mosses, i.e. around 50%. In order to preserve efficient functioning of orthologs it might have been necessary to evolve a slightly different codon usage for these genes in mosses, as is e.g. the case in *Arabidopsis*. The retained genes, on the other hand, are not found in seed plants and do not reflect the codon usage found there.
The average number of synonymous codons that is used in *Physcomitrella*and *Arabidopsis*CDS is not significantly different (Fig. [6b](#F6){ref-type="fig"}, bar chart). However, the percentage distribution of synonymous codon usage, as measured by the effective number of codons (enc), is surprisingly dissimilar (Fig. [6b](#F6){ref-type="fig"}, scatter plot). Most *Arabidopsis*coding sequences use a lot of synonymous codons (enc 45--59), whereas *Physcomitrella*displays a linear percentage increase from low to high values. Interestingly, around 15% of the moss genes contain no codon bias at all (enc = 61).
Conclusion
==========
The genome of the ancient land plant *Physcomitrella patens*, a moss, harbours genes of which at least 30% have a detectable homolog in seed plants. EST clustering yielded a database that covers a large proportion of the transcriptome, approximately 8% of the virtual transcripts contain full-length CDS.
Transcripts that are clear homologs of *Arabidopsis*genes were mapped against the *Arabidopsis*chromosomes, along with the set of paralogs and orthologs between the two organisms. All three sequence sets could be mapped evenly across the chromosomes, revealing neither hot nor cold spots (despite centromeric regions) nor differences in gene density.
While moss genes resemble those of *Arabidopsis*, there are significant differences. Introns are larger than those of the seed plant and are also longer than exons within moss, which is not the case in *Arabidopsis*. The G/C content of exons equals the A/T content. This might reflect a certain tenacity of the mosses\' genetic system and its slow mutational rate. These might be necessary characteristics, as due to haploidy in the dominant gametophyte, the chance for the propagation of a disadvantageous change is higher than in a polyploid organism.
Whereas orthologs display a codon fraction usage drift towards *Arabidopsis*, the contrary is the case for retained genes. Thus, evolution of codon usage seems to be correlated with evolutionary history of protein genes. Mutational bias does not seem to play a role in the evolution of moss coding sequences. While the majority of *Physcomitrella*CDS displays codon bias, there is a significant fraction (\~15%) of genes that is not biased at all, possibly representing a more ancient nucleotide composition than oberved in *Arabidopsis*. Splice sites in the moss resemble those in *Arabidopsis*, however, species-specific prediction models, like the one presented here, are necessary in order to avoid false positives. The same is true for the prediction of ORF based on EST data.
A high proportion of the sequence space of unicellular eukaryotes is covered by moss homologs, which apparently have not been lost since the days of the last common ancestor. The majority of moss genes find their best scoring homolog in plants. However, there are 134 putative retained genes that have their best BLAST hit among other taxonomic groups. Of those, 57 genes are specific to a single taxonomic group, putative functional annotation could be carried out for 25 of these proteins. The functional annotation revealed deviations in the taxonomic distributions: certain sets of genes seem to be shared with specific taxonomic groups, for example, transport proteins with fungi or signal transduction genes with bacteria and animals. Of special interest are genes that are possibly involved in cytotoxicity, metabolism and nucleic acid repair. These genes might be the reasons for some of the extraordinary capabilities of mosses, namely resistance against microbial pathogens, additional secondary pathways (as compared with seed plants) and a high rate of homologous recombination.
Methods
=======
Clustering of EST data
----------------------
All publicly available protein encoding DNA sequences of *Physcomitrella*were retrieved using Entrez \[[@B30]\] and divided into 399 \"seeds\" (full length mRNA sequences) as well as 102,535 EST and other sequences. This dataset is called the *Physcomitrella patens*public set, or PPP.
A set of 17 moss-specific repetitive elements, detected mainly in the untranslated regions of *Physcomitrella*genes \[[@B31]\] and used for filtering (see below) is available via cosmoss.org \[[@B32]\]. Filtering, clustering and assembly of EST data were done using the Paracel transcript assembler, PTA \[[@B33]\]. A species-specific parameter set has been developed and is available upon request.
For sequences where electropherograms were available, base-calling was carried out using phred \[[@B34]\]. Base quality values of EST sequences without available sequencer raw data was set arbitrarily to a low value, 10%, and in the case of seed sequences to a higher confidence value of 50%. Filtering included steps for removal of synthetic (vector/linker) and low quality sequences as well as of contaminants (homologs of *E. coli*as well as *Physcomitrella*mitochondrial, rRNA and chloroplast genes). Low-complexity regions were annotated together with poly-A tails, untranslated regions (UTR, UTRdb see \[[@B35]\]) and repetitive elements (repeats, repbase see \[[@B36]\]), in order not to disturb clustering and assembly. In a final step, sequences containing less than 150 bases of sense characters were removed. For PPP, a total of 100,079 sequences went into the clustering.
Prior to clustering, homologs of the seed sequences were pulled out of the sequence pool and assembled independently. Where possible, sequences were placed into 5\' and 3\' partitions based on detected poly-A tails and inherent annotated information. Both during clustering and assembly, putative chimeras (cloning artefacts) were detected and tagged. During assembly, contigs were built within clusters and putative splice variants detected. After clustering and assembly, the PPP set contained a total of 26,131 sequences. By using only the longest sequence in each cluster, a non-redundant set of 22,218 sequences was produced. The PP dataset contained 63,685 sequences in the complete and 48,961 sequences in the non-redundant set.
Splice site prediction
----------------------
For the splice site prediction, all publicly available pairs of genomic and cDNA/mRNA sequences were retrieved (40 genes). Together with 29 unpublished sequences, these sequences were aligned using MGAlign 1.3.6 \[[@B37]\] in order to determine the splice sites. The procedure yielded a total of 438 exons and 368 introns. The complete dataset is available via cosmoss.org \[[@B32]\].
The sequence logos (Fig. [5a](#F5){ref-type="fig"}) were created via the web interface at \[[@B38]\] using 10 nucleotides up- and downstream of the donor / acceptor sites.
Suppor vector machine: The software used for training and classification was SVMlight \[[@B39]\], libsvm \[[@B40]\] and svmsplice \[[@B41]\]. The complete set of splice sites was divided into training/testing sets of sizes 10--90%, for each set three samples were drawn. The set containing 90% of the sites for training proved to yield the best results. Optimization of parameters was done by 10-fold cross-validation, plotting precision vs. recall and chosing the best curve. The best performing model could be constructed using 50 nucleotides up- and downstream of the splice sites as context with the basepairing feature set of svmsplice and a polynomial kernel function of 4^th^order.
BLAST searches and filtering
----------------------------
BLAST searches were carried out using Paracel BLAST \[[@B33]\], a parallelized version of BLAST 2 \[[@B42]\], on amino acid level whenever applicable. In order to exclude random hits which are not based on true sequence homology, alignments had to contain at least 30% identical positions and a minimum length of 100 amino acid characters. This rigorous filtering excludes some true positive hits but removes almost all false positives \[[@B43]\]. Putative full-length CDS had to pass the same filtering. In addition, in this case only those hits were counted that covered at least 90% of the subjects length. For the determination of identical sequences, BLASTN was performed and hits were filtered to be at least 95% identical and 300 nucleotides long. Non-redundant hits were counted by removing all subjects that were present more than once in the search result.
Additional sequence datasets
----------------------------
The predicted coding sequences of rice (56,056 sequences) and *Arabidopsis*(28,581 sequences) genes were taken from release 1.0 and 4.0 of the TIGR database \[[@B44]\], respectively. The taxprot dataset (3,346,100 sequences, see Table [1](#T1){ref-type="table"} for details) was created by downloading the respective sequences from Genbank \[[@B30]\] using appropriate Entrez queries. All three datasets consist of amino acid sequences. The *Arabidopsis thaliana*chromosome sequences were retrieved from Genbank \[[@B45]\].
ORF prediction
--------------
ESTScan 2.0 \[[@B46]\] was used to predict open reading frames. The species-specific model for *Physcomitrella*was built by using the 399 public full length seed sequences (complete mRNAs) mentioned above. ORF were predicted from the clustered EST data (non-redundant datasets). For the PPP set, 19,081 ORF were predicted; 34,981 for the PP set. Predictions were done using both the *Arabidopsis*and the *Physcomitrella*model for comparison. Manual inspection of several known CDS revealed that the *Arabidopsis*-based prediction contained false-positive stretches, which was not the case for the *Physcomitrella*-based prediction. Although the *Physcomitrella*model predicted a lower number of ORF, it was used in order to keep false-positives to a minimum.
Codon usage
-----------
Four different sets of coding sequences were used (see table [3](#T3){ref-type="table"}). A set of 7,765 well annotated *Arabidopsis*mRNAs was retrieved using Entrez. The *Physcomitrella*datasets contained the predicted ORF for the complete PPP set (19,081 sequences), the *Arabidopsis*paralogs (1,659) and orthologs (1,476) described above as well as the putatively retained genes not found in higher plants (134). The smallest set contained 77,998 nucleotides and thus a theoretical average of 406 instances of each triplet, which allows significant analyses.
Nucleotide frequencies were calculated with the GCG 10.3 \[[@B47]\] software composition. Codon usage fractions for individual datasets were calculated as percentage of the respective total amount of counted codons. Absolute deviations in comparison to the full *Physcomitrella*ORF set were calculated for the three subsets (retained genes, *Arabidopsis*orthologs and paralogs). The computed mean value over all sets (average absolute deviation) was 0.069. Codon fraction usage deviation was counted as significant only if it differed at least twice as much (+/- 0.138%) from the full set.
The effective number of codons (enc) was calculated using CodonW (J. Peden, \[[@B48]\]). The enc values range from 20 (maximum bias, i.e. only one synonymous codon is used per amino acid) to 61 (no bias, all synonymous codons are being used).
Abbreviations
=============
CDS = coding sequence(s), EST = expressed sequence tag(s), ORF = open reading frame(s)
Authors\' contributions
=======================
SAR carried out most of the analyses, drafted the manuscript and designed the work. DF carried out the splice site prediction. DL participated in the analyses and generated the databases and the web interface. RR participated in drafting and conception of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
================
We would like to thank Sven Degroeve (University of Gent, Belgium) for providing of and assistance with svmsplice as well as Hans Stenøien, Colette Matthewman and several anonymous reviewers for helpful comments on the manuscript.
|
PubMed Central
|
2024-06-05T03:55:55.246288
|
2005-3-22
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079823/",
"journal": "BMC Genomics. 2005 Mar 22; 6:43",
"authors": [
{
"first": "Stefan A",
"last": "Rensing"
},
{
"first": "Dana",
"last": "Fritzowsky"
},
{
"first": "Daniel",
"last": "Lang"
},
{
"first": "Ralf",
"last": "Reski"
}
]
}
|
PMC1079824
|
Background
==========
It is well known that histone acetylation and deacetylation are involved in the transcription process. First, histone modifications can change the affinity of histone proteins to DNA sequences. The disrupted higher order folding of chromatin in turn can affect the transcription process \[[@B1],[@B2]\] because chromosome structural changes can affect accessibility of transcription factors to their target sequences \[[@B3],[@B4]\]. Second, some acetylated lysine sites of histone proteins can function as binding sites for transcription factors \[[@B5]-[@B7]\]. Histone deacetylation is also related to aberrant gene expressions in human cancer. For example, in cancer cells, cyclin-dependent kinase inhibitor *p21*^*WAF*1^is silenced by the promoter histone hypoacetylation caused by recruited histone deacetylases \[[@B8],[@B9]\]. *p21*^*WAF*1^is considered as a *bona fide*tumor-suppressor gene, and its activity can be restored by histone deacetylase inhibitors, which are related to the promoter histone hyperacetylation \[[@B9]\]. Therefore, there is great interest and potential in studying histone acetylation and deacetylation.
The organization of chromatin can be classified into two distinct domains: heterochromatin and euchromatin. The transcription of genes in heterochromatin is usually repressed whereas genes in euchromatin are usually active. Recent studies have shown that the boundaries of heterochromatin are surrounded by high-level histone acetylation \[[@B10]\]. It has been a long-term puzzle whether chromatin can be further divided into distinct gene expression domains, with reports showing that co-expressed genes are clustered into chromatin domains \[[@B11]-[@B14]\], which suggests that euchromatin can be further divided into separate domains. The experiments that showed that changing genes\' positions may cause aberrant gene regulation also support this hypothesis \[[@B15],[@B16]\]. Recent studies on boundaries and insulators also suggest that chromatin can be partitioned into gene expression sub-domains, because insulators and boundary elements can block the effects of nearby enhancers or nearby heterochromatin \[[@B17],[@B18]\]. Consequently, genes within the blocked domains may be co-regulated through unknown mechanisms. For example, the chicken β-globin HS4 is an insulator in the chromatin region with transitions of histone acetylation levels: from silent chromatin regions with hypoacetylated and Lys-9 methylated histones to active chromatin region with acetylated histones and active genes in erythroid cells \[[@B10]\].
It is well known that there is correlation between histone acetylation level and gene expression \[[@B19]\]. But the detailed relationship between histone acetylation patterns and gene expression domains is less clear. Hypoacetylated histones are associated with silenced chromatin. Histone hyperacetylation may prevent the folding of nucleosomal arrays into more condensed chromatin structures \[[@B1],[@B2]\]. Conversely, histone deacetylation might stabilize higher order chromatin folding. We hypothesize that histone acetylation sites may serve as boundaries of gene expression domains. Although neighboring genes tend to be co-regulated by some enhancers because of their close proximity, higher order chromatin folding caused by histone deacetylation may reduce this neighborhood effect by preventing the distal enhancers from accessing the promoter.
Results
=======
Adjacent genes tend to be co-expressed
--------------------------------------
Although many reports have shown that nearby genes tend to have similar expression profiles \[[@B11]-[@B14]\], such observed co-expression may be caused by the spatial arrangement of probes for these genes on microarrays \[[@B20]\]. To minimize this potential spatial effect in the study of co-expression, we used 269 microarray datasets under different conditions because the chip designs differed among different datasets. In order to assess the significance of co-expression between neighboring genes, we performed the hypergeometric test for the pair-wise correlations between adjacent genes and those among all the genes. For every correlation threshold, we counted the number of pair-wise correlations above the threshold for adjacent genes and that for all gene pairs regardless of their physical locations. The number of adjacent genes above a threshold has a hypergeometric distribution under the null hypothesis. The results are summarized in Table [1](#T1){ref-type="table"}. As the p-values in this table indicate, the co-expression of neighboring genes is significant for every threshold considered (from 0.5 to 0.9).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
The hypergeometric test for the co-expression effect among neighboring genes. For every correlation threshold, the number of pair-wise correlations for the adjacent genes that are above the threshold was counted. The probability of randomly drawing such high numbers of gene pairs above the threshold from all of the gene pairs is determined by a hypergeometric distribution, as indicated by the p-values.
:::
Total gene pairs Neighboring gene pairs Threshold for correlations Number of total gene pairs above the threshold Number of neighboring gene pairs above the threshold P-values based on the hypergeometric tests
------------------ ------------------------ ---------------------------- ------------------------------------------------ ------------------------------------------------------ --------------------------------------------
18,797,646 6,116 0.5 616,074 752 2.00e-008
0.6 251,505 401 4.84e-009
0.7 89,955 196 1.04e-008
0.8 21,513 68 1.92e-008
0.9 1,171 13 4.07e-009
:::
The deacetylation sites reduce correlations among adjacent genes
----------------------------------------------------------------
In order to test the hypothesis that chromatin deacetylation sites may reduce expression correlations among neighboring genes, we again performed a hypergeometric test. For every correlation threshold, we counted the number of pairwise correlations above the threshold for all adjacent genes and that for all adjacent genes separated by a deacetylation site. The number of adjacent genes with deacetylaiton sites between them above the threshold is hypergeometrically distributed under the null hypothesis. The results in Table [2](#T2){ref-type="table"} suggest that the partition effect is significant at significance level 0.05 for thresholds from 0.5 to 0.9. When a one-sided t-test was performed for correlations between adjacent genes separated by a deacetylation site and those without a deacetylation site, the p-value was 0.0467. This result suggests that the expression correlations between adjacent genes are significantly reduced by the presence of deacetylation sites. In addition, in Figure [1](#F1){ref-type="fig"}, we plotted the expression correlations versus the physical distances at log scale for the adjacent genes without a deacetylation site. The correlation between the expression similarities and the log of physical distances is -0.140 with the 95% confidence interval (-0.168, -0.111), meanwhile in Figure [2](#F2){ref-type="fig"} the correlation for adjacent genes separated by a deacetylation site is -0.074 with the 95% confidence interval (-0.123, -0.024). It indicates that if two neighboring genes are close to each other physically, they are more likely to be co-expressed. However the deacetylation sites can reduce this correlation between the physical distance and the expression similarity.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
The hypergeometric test for the partition effect of the deacetylation sites. For every correlation threshold, the number of pair-wise correlations for the adjacent genes separated by deacetylation sites that are above the threshold was counted. The probability of randomly drawing such low numbers of gene pairs above the threshold from all of the adjacent gene pairs is determined by a hypergeometric distribution.
:::
Total neighboring gene pairs Neighboring gene pairs separated by deacetylation sites Threshold for correlations Number of neighboring gene pairs above the threshold Number of neighboring gene pairs separated by deacetylation sites above the threshold P-values based on the hypergeometric tests
------------------------------ --------------------------------------------------------- ---------------------------- ------------------------------------------------------ --------------------------------------------------------------------------------------- --------------------------------------------
6,116 1,558 0.5 752 161 0.00322
0.6 401 73 0.000231
0.7 196 26 0.0000167
0.8 68 5 0.000120
0.9 13 0 0.0218
:::
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Scatter plot for the log of physical distances and the expression correlations for neighboring genes without a deacetylation site.**The physical distance is measured by the difference of the start positions (in bp) of the two ORFs. The correlation between the log of physical distances and the expression correlations is -0.140 with the 95% confidence interval (-0.168, -0.111).
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Scatter plot for the log of physical distances and the expression correlations for neighboring genes with a deacetylation site.**The physical distance is measured by the difference of the start positions (in bp) of the two ORFs. The correlation between the log of physical distances and the expression correlations is -0.074 with the 95% confidence interval (-0.123, -0.024).
:::
:::
Tests on assembled co-regulation groups from YPD and ChIP-chip experiments
--------------------------------------------------------------------------
Because both transcription factors and chromatin status can affect gene expression levels, we further studied the effects of chromatin status on genes sharing common regulators. We assembled 294 co-regulation groups from YPD (Yeast Proteome Database as of March 2004 \[[@B21]\]) and calculated the pair-wise correlations within each group. The co-regulation group was defined by the target genes of a regulator. The deacetylation sites through the genome were used to partition the genes into different domains. There are 24,833 gene pairs sharing one common regulator. Among them, 141 pairs of genes are in the same deacetylation partition group, and 46 pairs of genes belong to neighboring deacetylation partition groups. We performed a hypergeometric test for the 141-pair group versus all of the pair-wise genes, and the p-values were significant (Table [3](#T3){ref-type="table"}). We also performed a hypergeometirc test for the 46-pair group and all of the pair-wise genes and the p-values were not significant compared with the 141-pair group (Table [3](#T3){ref-type="table"}). The results show that genes sharing common regulators are more co-expressed if they are in the same deacetylation partition groups. We also performed a one-sided t-test for correlations between the genes sharing regulators and deacetylation partition groups and those sharing regulators but in neighboring deacetylation partition groups, and the p-value was 3 × 10^-4^. This suggests that there is significant evidence that the correlations between genes sharing regulators and deacetylation partition groups are greater than those between genes sharing regulators but in neighboring groups. In other words, in addition to sharing common regulators, belonging to the same deacetylation partition group may be another factor contributing to the co-expression of two genes.
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
The hypergeometric tests for the partition effect of the deacetylation sites for the co-regulated (YPD) gene pairs. For every correlation threshold, the number of pair-wise correlations for the genes with a common regulator and in the same partition group that are above the threshold was counted. The number of pair-wise correlations for the genes with same regulator and in the neighboring partition groups that are above the threshold was also counted. The probability of randomly drawing such high numbers of gene pairs above the threshold from all of the gene pairs with same regulator is determined by a hypergeometric distribution. The p-values are indicated in the table.
:::
Total gene pairs sharing one regulator Gene pairs sharing one regulator and both are in the same deacetylation partition group Threshold for correlations Number of gene pairs sharing one regulator above the threshold Number of gene pairs sharing one regulator and both are in the same deacetylation partition group above the threshold P-values based on the hypergeometric tests
---------------------------------------- ------------------------------------------------------------------------------------------------ ---------------------------- ---------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------ --------------------------------------------
24,833 141 0.5 4,676 80 3.43e-011
0.6 2,925 54 2.37e-011
0.7 1,519 45 2.24e-011
0.8 422 20 2.92e-011
0.9 30 3 0.0000243
Total gene pairs sharing one regulator Gene pairs sharing one regulator and they belong to neighboring deacetylation partition groups Threshold for correlations Number of gene pairs sharing one regulator above the threshold Number of gene pairs sharing one regulator and they belong to neighboring deacetylation partition groups above the threshold P-values based on the hypergeometric tests
24,833 46 0.5 4,676 12 0.0784
0.6 2,925 8 0.0853
0.7 1,519 8 0.00165
0.8 422 2 0.0432
0.9 30 0 0.0541
:::
In addition to YPD data, we also used ChIP-chip data \[[@B22]\] to define regulatory targets of transcription factors. A p-value threshold of 0.005 was used to infer the binding of regulators for each gene. There are 476,649 gene pairs sharing at least one regulator based on this threshold, and we calculated the pair-wise correlations for these pairs. Among all the pairs, 1,811 pairs belong to the same deacetylation partition group, and 1,019 pairs are in the neighboring partition groups. The comparisons among different groups are summarized in Table [4](#T4){ref-type="table"}. For the genes in the same deacetylation partition group, the proportion of the number of pairs having correlations above each threshold is significantly higher than that based on all gene pairs. But for the genes in the neighboring deacetylation partition groups, the difference is not statistically significant.
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
The hypergeometric test for the partition effect of the deacetylation sites for the co-regulated (ChIP-chip experiments) gene pairs. For every correlation threshold, the number of pair-wise correlations for the genes with the same regulator and being in the same partition group that are above the threshold was counted. The number of pair-wise correlations for the genes with the same regulator and being in the neighboring partition groups that are above the threshold was also counted. The probability of randomly drawing such numbers of gene pairs above the threshold from all of the gene pairs with the same regulator is determined by a hypergeometric distribution.
:::
Total gene pairs sharing one regulator Gene pairs sharing one regulator and both are in the same deacetylation partition group Threshold for correlations Number of gene pairs sharing one regulator above the threshold Number of gene pairs sharing one regulator and both are in the same deacetylation partition group above the threshold P-values based on the hypergeometric tests
---------------------------------------- ------------------------------------------------------------------------------------------------ ---------------------------- ---------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------ --------------------------------------------
476,649 1,811 0.6 14,738 134 \<4.75e-011
0.7 8,419 68 5.71e-009
0.8 4,556 29 0.00329
0.9 621 7 0.00293
Total gene pairs sharing one regulator Gene pairs sharing one regulator and they belong to neighboring deacetylation partition groups Threshold for correlations Number of gene pairs sharing one regulator above the threshold Number of gene pairs sharing one regulator and they belong to neighboring deacetylation partition groups above the threshold P-values based on the hypergeometric tests
476,649 1,019 0.6 14,738 34 0.287
0.7 8,419 18 0.438
0.8 4,556 8 0.638
0.9 621 1 0.383
:::
When we performed a one-sided t-test for the correlations for the 1,811-pair sample and the 1,019-pairs sample, the p-value was 4 × 10^-8^. In Figure [3](#F3){ref-type="fig"}, we compared the empirical cumulative distributions of the two samples. It can be seen that the distributions of the correlations are different, and there are more pairs having higher pair-wise correlations for the 1,811 pair group belonging to the same deacetylation partition than that for the group belonging to the neighboring partitions. A one-sided two-sample Kolmogorov-Smirnov test was performed between these two samples, and there is statistically significant evidence suggesting that the two distributions are different (p-value \< 0.0001). The Kolmogorov-Smirnov test is a test of whether the two data samples come from the same distribution, without making assumption about the distribution of the data.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Comparison of the empirical cumulative distributions of the two samples.**The x-axis indicates the pair-wise correlation. The y-axis indicates the empirical cumulative distribution of the pair-wise correlation. The solid line is for the correlations between genes sharing one regulator and being in the same deacetylation partition group (the 1,811-pair sample). The dotted line is for the correlations between genes sharing one regulator and belonging to neighboring deacetylation partition groups (the 1,019-pair sample).
:::
:::
Study on gene expression domains
--------------------------------
In addition to examine neighboring genes, we also studied all genes in the same partition group simultaneously. For each partition group, we calculated the average correlation, the average of absolute correlations, and the maximum correlation. To test the significance of the correlations, we sampled groups of genes from the genome for each group size at random 1,000,000 times. For each sampled group, the average correlation, the average of absolute correlations, and the maximum correlation were calculated. The p-values can be estimated based on the correlations for these simulated groups. The results summarized in Figure [4](#F4){ref-type="fig"} clearly demonstrate the statistical significance for the high correlations for the genes grouped by deacetylation sites.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**The histograms of the p-values for the correlations in partition groups.**(A) The histogram of the p-values for the average of pair-wise correlations in partition groups. (B) The histogram of the p-values for the average of absolute pair-wise correlations in partition groups. (C) The histogram of the p-values for the maximum pair-wise correlations in partition groups. The p-values are based on randomly sampling from the whole genome 1,000,000 times for each partition group size.
:::
:::
Because the presence of such clusters may be due to reasons other than histone deacetylated sites, e.g. spatial proximity among neighbouring genes, we also performed a different type of simulations to assess the significance of the maximum correlation within each group. We permuted the order of the groups on the whole genome 900,000 times. For each permuted sample, the number of groups that had the maximum correlation larger than 0.6, 0.7, 0.8, or 0.9 was counted. We called these groups with the maximum correlation above a given threshold as qualified groups. We counted the number of permuted samples with more qualified groups than that based on the observed histone deacetylation partition. The statistical significance of the observed partition is estimated as the proportion of the permuted samples having more qualified groups than that based on the observed histone deacetylation partition. As shown in Table [5](#T5){ref-type="table"}, the results are statistically significant for the threshold of 0.6, 0.7, and 0.8. But the results for the average correlation and the average of absolute correlations were not statistically significant (data not shown). One possible reason is that many other regulators can affect the transcription process within each partition, resulting in less significant results.
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
The significance of the number of qualified groups for histone deacetylation partition. The statistical significance of the histone deacetylation partition is estimated by the proportion of permutations that have more groups with the maximum correlation larger than the threshold than the observed histone deacetylation partition.
:::
Threshold Observed qualified groups for maximum correlations Number of times of having more qualified clusters for simulations P-values
----------- ---------------------------------------------------- ------------------------------------------------------------------- ----------
0.6 352 27,795 0.0309
0.7 198 26,285 0.0292
0.8 77 6,536 0.00726
0.9 17 61,006 0.0678
:::
In order to test whether there is function enrichment for the deacetylation partition groups, the following procedure was performed. There are 1,113 deacetylation partition groups with more than one gene in total. For each of these deacetylation partition groups, the Yeast GO Slim terms were mapped to the genes using SGD Gene Ontology Slim Mapper \[[@B23]\]. The p-value was calculated using the hypergeometric distribution as the probability of n or more out of m genes having a specific GO Slim term, given that N out of M genes have that GO slim term in the whole genome. If there was at least one GO Slim term with p-value less than or equal to 0.0001 in a deacetylation partition group, we called it a significant group. There are 386 significant groups out of the 1,113 deacetylation partition groups, whereas only 4 out of 100 simulations have more than 386 significant groups. The simulation was performed by the group order permutation as described in the above paragraph. If we pool all of the p-values from the simulations together, the 5% percentile rank is 3.02 × 10^-4^and 493 out of 1,113 groups have at least one GO Slim term with p-value less than or equal to 3.02 × 10^-4^. These results indicate that the deacetylation partition groups are enriched for genes of specific GO Slim terms.
Discussion
==========
The effect of chromatin status on gene transcription regulation is well known, but the detailed mechanisms remain elusive. In this paper, we tested the hypothesis that histone deacetylation takes part in the chromatin expression domain formation through combined analyses of different data types, including microarray data for deacetylation sites, gene expression, ChIP-chip data, and YPD database. These analyses have revealed significant effects of histone deacetylation on gene expression patterns. Although we only presented results using 1.95 as the threshold for defining the deacetylation sites (which was used in the ref \[[@B24]\]), different thresholds (e.g. 1.8 and 2.1) yield similar results.
As an example for the effect of deacetylation sites, we considered two genes, YLR329W and YLR328W, which are separated by a deacetylation site in our analysis. YLR329W (REC102) is a protein involved in early stages of meiotic recombination. The early meiosis-specific genes are repressed by histone deacetylation in the mitotic cell cycle. They are activated by histone acetylation in the meiotic stage \[[@B25]\]. The correlation between these two genes expression profiles is 0.1965. The correlation increases to 0.7499 in the sporulation time course dataset \[[@B26]\] which consists of meiosis. This example illustrates that these two genes are co-expressed when there is no deacetylation between them, while the correlation is decreased by the presence of a deacetylation site.
The observed co-expression among neighboring genes in the yeast cell cycle datasets are still debatable because the co-expression may be due to spatial arrangement of the probes on microarrays \[[@B20]\]. In our study, we used 269 microarray datasets under different conditions. Because the chip designs differed among different datasets, the spatial effect is significantly minimized when combining a large number of datasets of diverse sources. More importantly, the results that compare the genes in the same deacetylation partition and those in neighboring partitions demonstrate that the spatial effect is not the reason for the observed difference. This is because the spatial effect should have similar impact under both conditions. This is further supported by the permutation results of the partition groups.
In addition to histone acetylation and deacetylation, DNA methylation is another key factor for the epigenetic effects on gene expression. Because *Saccharomyces cerevisiae*has no detectable DNA methylation, this organism serves as a good model to examine the relationship between deacetylation patterns and gene expression domains.
Since histone deacetylation was measured in a heterogeneous population, the domain partitions may not be an accurate representation for a specific condition. In addition, other histone modifications that can affect gene expressions may be ignored and need further studies.
Conclusion
==========
The fact that histone modification may affect gene transcription has been recognized for many years. Thanks to the high-throughput gene expression microarray data and gene acetylation microarray data, we are able to study the effects of histone acetylation at the whole genome scale. In this article, we confirmed the previous findings on the co-expression effect of neighboring genes. Furthermore, we have tested the hypothesis that histone deacetylation sites serve as possible partitions for chromatin domains of gene expression through several approaches. We hypothesize that histone deacetylation can lead to dense higher order chromatin structure and reduce the accessibility of the transcription factors to their target sequences. The detailed mechanisms need to be further studied through biological experiments.
Methods
=======
Histone deacetylation sites and gene expression data
----------------------------------------------------
In order to test our hypothesis that deacetylation sites function as a boundary complex and partition the neighborhood gene expression domains, we correlated histone deacetylated sites with gene expression profiles as follows. The histone deacetylation sites were inferred through specific histone acetylation microarrays \[[@B24]\], where chromatin immunoprecipitation and DNA microarrays are combined to determine the histone acetylation levels. The increased histone acetylation sites in the absence of histone deactylases were considered as the putative histone deacetylation sites (1.95 fold change as cutoff) in the wild type strain. The data for the histone deacetylases RPD3, HDA1, HOS1, HOS2, HOS3 and SIR2 were combined for the intergenic regions. These data are normalized average acetylation fold changes. The intergenic region was considered as a putative deacetylation site if one dataset for any of the histone deacetylases is larger than 1.95. In total, there are 1,747 putative deacetylation sites. The putative sites throughout the genome were used to define gene cluster domains. We used 269 yeast microarray datasets (downloaded from NCBI GEO database \[[@B27]\], there are 6,132 genes with \<=25% missing data across all of the experiments) for gene expression analysis. In total, there are 1574 deacetylation groups. Among them, 461 deacetylation groups only contain one gene, and 315, 183, 149 and 121 groups contain 2, 3, 4, and 5 genes. The detailed group information is summarized in [Additional File 1](#S1){ref-type="supplementary-material"}.
Statistical analyses
--------------------
Pearson\'s correlation coefficients were calculated for gene expression data. The hypergeometric tests were performed to assess the significance of the co-regulation effects between neighboring genes, the significance of the partition effect of the chromatin deacetylation sites, and the effects of chromatin status on genes sharing common regulators. One-sided t-tests were performed to compare the genes in the same deacetylation partition groups and those in neighboring partition groups. The one-sided two-sample Kolmogorov-Smirnov test was performed to test the difference between the distribution of correlations for genes sharing the same regulators and deacetylation partition groups and that for genes sharing the same regulators but are in neighboring groups.
Authors\' contributions
=======================
LC carried out the data analyses and was involved in the preparation of the manuscript. HZ supervised the study and was involved in the preparation of the manuscript. Both authors read and approved the final manuscript.
Supplementary Material
======================
::: {.caption}
###### Additional File 1
**Deacetylation partition groups information.**Groups.els is an excel spreadsheet which includes the gene names and the chromosome number for each deacetylation partition group.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
This work was supported in part by NSF grant DMS 0241160.
|
PubMed Central
|
2024-06-05T03:55:55.249759
|
2005-3-23
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079824/",
"journal": "BMC Genomics. 2005 Mar 23; 6:44",
"authors": [
{
"first": "Liang",
"last": "Chen"
},
{
"first": "Hongyu",
"last": "Zhao"
}
]
}
|
PMC1079825
|
Background
==========
Sperm adhesion molecule 1 (SPAM1, also known as PH-20) is a member of a family of at least six mammalian hyaluronidases. The genes encoding these enzymes cluster in two groups of three -- *SPAM1*, *HYAL4*and *HYALP1*(a pseudogene) on human chromosome 7q31, and *HYAL1*, *HYAL2*and *HYAL3*on human chromosome 3p21 \[[@B1],[@B2]\]. The orthologous mouse genes form similar clusters at syntenic chromosomal locations \[[@B1]\]. This suggests that two single-gene duplications, followed by a small segmental duplication, occurred before the divergence of human and mouse approximately 80 million years ago.
HYAL4 exclusively degrades chondroitin. In contrast, HYAL1, HYAL2, HYAL3 and SPAM1 hydrolyze hyaluronic acid, with different substrate size preferences and tissue specificities \[[@B1]-[@B3]\]. Expression of *SPAM1*has been unanimously reported in the testis in various species (reviewed in \[[@B1],[@B4]\]). Expression has also been detected in the human epididymis, vas deferens, prostate and placenta \[[@B2],[@B5]\] and the murine epididymis, kidney, uterus, vagina and oviduct \[[@B6]-[@B8]\]. Expression of *SPAM1*has not been detected in the human female reproductive tract \[[@B2],[@B9]\].
SPAM1 has various functions in fertilization. A catalytic domain has been shown to degrade hyaluronic acid \[[@B10],[@B11]\]. This molecule is a major extracellular matrix component of the cumulus cell layer that surrounds the ovum, and SPAM1 has been shown to remove this cumulus layer *in vitro*\[[@B12]\]. SPAM1 has hyaluronic acid and zona pellucida binding regions that are distinct from its catalytic domain \[[@B13],[@B14]\] and is also involved in an intracellular signalling pathway in sperm cells upon binding to the zona pellucida \[[@B4],[@B15],[@B16]\].
The role of murine SPAM1 in fertilization has been investigated using a knockout mouse line. Sperm from *Spam1*^-/-^mice showed a delay in the removal of the cumulus cell layer and fertilization *in vitro*. Surprisingly, however, *Spam1*^-/-^males showed normal *in vivo*fertility rates and sired normal-sized litters \[[@B17]\]. Sperm from *Spam1*^-/-^mice maintained 40% of the wild-type level of hyaluronidase activity, while protein expression assays indicated the presence of a second hyaluronidase in these cells \[[@B17]\]. This was unexpected, as SPAM1 was thought to be the only testicular hyaluronidase. These results may be explained by recent evidence that the murine orthologue of the human *HYALP1*pseudogene has an intact ORF and is expressed in mouse testis \[[@B1],[@B3],[@B18]\], and that a seventh hyaluronidase, *Hyal5*, may exist in mouse, but not in human \[[@B3],[@B18]\]. There may therefore be some redundancy among murine testicular hyaluronidases that explains the fertility of *Spam1*^-/-^mice. In this case, it remains likely that SPAM1 is an essential protein in human fertilization.
Little is known about the transcriptional regulation of *SPAM1*. A non-consensus cAMP response element (CRE) in the murine *Spam1*promoter bound the testis-specific CRE modulator (CREM) protein and was involved in activation of *Spam1*transcription *in vitro*\[[@B19]\]. In addition, *Spam1*expression was abolished in CREM-deficient mice \[[@B19]\]. Various other putative transcription factor binding sites have been identified in the human, mouse and rat *SPAM1*promoters \[[@B5]-[@B7],[@B19],[@B20]\]; however, the sites are generally non-consensus and have not yet been shown to be functional. The restricted developmental and spatial expression of SPAM1 \[[@B5],[@B7],[@B21]\], as well as the unique transcriptional mechanisms employed during spermatogenesis (reviewed in \[[@B22]\]), may render *SPAM1*unamenable to traditional methods of transcription and promoter analysis.
In a previous study by our group, *SPAM1*was identified as one of the many human transcripts that contain transposable element (TE) sequence \[[@B23]\]. TEs include long and short interspersed nuclear elements (LINEs and SINEs), DNA transposons, and endogenous retroviruses (ERVs). TEs are extremely common in the human and mouse genomes, and together contribute 45% and 40% of the total sequence, respectively \[[@B24],[@B25]\]. Many human and mouse gene transcripts contain TE sequences in their untranslated regions (UTRs) \[[@B23],[@B26],[@B27]\]. TEs also contribute to the transcriptional regulation of many genes. The antisense LINE1 promoter and the long terminal repeat (LTR) promoters of ERVs are known to participate in the tissue-specific expression of various host genes \[[@B28]-[@B30]\]. Through bioinformatic analysis, human and mouse *SPAM1*transcripts were predicted to initiate within an antisense ERV \[[@B23]\], indicating that this gene may represent another example of transcriptional regulation by a TE.
In this study, we show that the first exons and proximal promoter regions of the human and rodent *SPAM1*genes are derived from an ERV1 *pol*coding region, and identify novel alternative promoters and splicing variants of the gene. We show that the human and mouse ERV-derived promoters are largely testis-specific, and discuss the implications of ERV insertion on the evolution of transcriptional regulation at this locus.
Results
=======
The human *SPAM1*gene initiates within an ERV1 *pol*region
----------------------------------------------------------
A recent study by our group used bioinformatic methods to investigate the contribution of TEs to human and mouse gene transcripts \[[@B23]\]. That study determined that 3.1% of human RefSeq genes initiate within a TE sequence, indicating that these genes are candidates for transcriptional regulation by TEs. One example identified in this way was the *SPAM1*gene, where the 5\'-terminus was found to map within an antisense ERV element. We have now investigated this locus in more detail.
We used the University of California at Santa Cruz (UCSC) genome browser at <http://www.genome.ucsc.edu> to more closely examine the genomic region surrounding the first exon of *SPAM1*. The region bears the hallmarks of multiple TE insertions into older, pre-existing repetitive elements, resulting in a \"patchwork\" effect of fragmented TEs from different families (Figure [1B](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}). The previously described *SPAM1*transcriptional start site (nucleotide +40 in exon 1A) mapped within an antisense ERV1 element of the MER34 family (Figure [1B](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}). Surprisingly, the ERV sequence was derived not from an LTR, but rather from an internal retroviral region. This is counter to all previously described examples of transcription initiation within an ERV. To confirm this RepeatMasker annotation, we performed a BLAST homology search of the translated retroviral sequence against a protein database. This analysis confirmed that *SPAM1*transcripts initiate within a fragment of the ERV1 *pol*gene.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Genomic structure of human *SPAM1*.**(A) Overview of the *SPAM1*locus. Exons are boxed and numbered, with the size in bp shown above. Intron sizes in kb are shown below. The approximate positions of the *SPAM1*ORF start and stop codons are indicated. The four *SPAM1*splicing variants are shown schematically below. ERV-derived sequences are shown in green, with other sequences in blue. The diagram is not to scale. A black bar indicates the region shown in more detail in Figure 1B and 2. (B) The positions of *SPAM1*alternative first exons 1A and 1B with respect to TE sequences. Exons are represented by solid black boxes. Bent arrows indicate the position and relative usage of the 5\'-most transcription start site for each exon. TEs are represented by the colored boxes below; arrows indicate the orientation of each TE.
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**The sequence and position of *SPAM1*exons 1A and 1B with respect to TEs.**The sequence shown corresponds to human chromosome 7 co-ordinates 123158485 to 123160464 in the UCSC genome browser (May 2004 release). Solid lines define the boundaries between different TE sequences. The class and orientation (F, forward; R, reverse) of each TE are given on the right hand side. Exon sequences are shown in upper case reverse type. The dotted white line frames the portion of exon 1A included in splicing variant 1A1, but not variant 1A2. The transcription start sites identified by 5\'-RACE are underlined, and splice donor sites are shown in bold italic type. A non-consensus CRE is shown in bold underlined type.
:::
:::
The alternative promoters and splicing variants of *SPAM1*
----------------------------------------------------------
We performed 5\'-rapid amplification of cDNA ends (RACE) to confirm the position of the *SPAM1*transcriptional start site. Since expression of *SPAM1*is confined largely to the testis, we used human testis RNA for this analysis. Sequencing of 5\'-RACE clones identified two alternative first exons of *SPAM1*(Figure [1](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}). We have designated the upstream, previously-identified first exon as exon 1A, and the novel downstream first exon as exon 1B. Exon 1A is wholly derived from the antisense ERV1 *pol*region. Exon 1B initiates within a different fragment of the same *pol*gene, but terminates within a sense orientation LTR of the ERV1 MER4C family (Figure [1B](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}). Transcripts containing both alternative first exons spliced into the same downstream exons; the *SPAM1*ORF begins in exon 4, and is therefore not affected by alternative promoter usage (Figure [1A](#F1){ref-type="fig"}).
Multiple transcription start sites were identified within exon 1A, at position +1, +6, +20 and +51 (Figure [2](#F2){ref-type="fig"}). We also identified a splicing variant of exon 1A, with variant 1A2 using a splice donor site at position +118. Use of this alternative splice site resulted in a truncated 117 bp first exon, as opposed to a full-length size of 296 bp for variant 1A1 (Figure [1A](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}). In contrast, a single transcription start site and no splicing variants were observed for exon 1B. However, some transcripts initiating within exon 1B contained a novel alternatively-spliced 85 bp exon (Figure [1A](#F1){ref-type="fig"}). The sequences of all human and murine *SPAM1*splicing variants have been deposited in GenBank with accession numbers AY920278 -- AY920283.
Both ERV-derived promoters are male-specific
--------------------------------------------
To verify the expression patterns of the *SPAM1*alternative promoter and splicing variants, we performed non-quantitative RT-PCR on a panel of RNAs derived from normal human tissues. As shown in Figure [3](#F3){ref-type="fig"}, expression of transcripts containing *SPAM1*ORF sequence was detected in the testis, as well as the heart, small intestine, prostate, muscle and placenta. Primers designed to amplify both exon 1A splicing variants detected transcripts only in the testis and prostate, while exon 1B-specific transcripts were detected in the testis, prostate, and to a lesser degree in the placenta. The smaller of the two splicing variants was predominant for both promoters; this may be due to an amplification bias introduced by the PCR.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Detection of *SPAM1*transcripts by RT-PCR.**Primer pairs specific for *GAPDH*, the *SPAM1*ORF, and *SPAM1*transcripts initiating within exon 1A or 1B were used in RT-PCR assays. Assays were carried out on cDNAs derived from a range of normal human tissues: 1, brain; 2, heart; 3, kidney; 4, liver; 5, lung; 6, bone marrow; 7, colon; 8, small intestine; 9, spleen; 10, stomach; 11, thymus; 12, mammary gland; 13, prostate; 14, muscle; 15, testis; 16, uterus; 17, spinal cord; 18, placenta. Approximate molecular weights are indicated on the left.
:::
:::
We next used real-time RT-PCR to quantify the level of *SPAM1*transcripts and the contribution of each alternative promoter to total gene expression. Primers annealing to exon 4 and exon 5, common to all *SPAM1*transcripts (Figure [1A](#F1){ref-type="fig"}), were used to determine the level of total gene expression. This value was normalized to the level of *GAPDH*transcripts and expressed relative to that obtained for the heart, which showed a low level of *SPAM1*expression (Figure [4](#F4){ref-type="fig"}). As expected, *SPAM1*was highly expressed in the testis, although low levels of expression were also detected in the prostate and some other tissues.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Quantitative analysis of *SPAM1*expression in normal human tissues.**Primers were used in real-time RT-PCR assays to amplify transcripts specific for *GAPDH*and total *SPAM1*transcripts, and, in testis only, for *SPAM1*transcripts initiating in exon 1A and 1B. Green bars represent the relative abundance of total *SPAM1*transcripts normalized to *GAPDH*. Solid and hatched blue bars represent the contribution of exon 1A and 1B transcripts, respectively, to total *SPAM1*expression. All bars represent the mean of four independent assays ± standard deviation.
:::
:::
On the basis of the results obtained with the ORF-specific primers, we decided to quantify the expression of exon 1A and 1B transcripts only in the testis. To avoid amplification of different-sized products from alternatively-spliced *SPAM1*transcripts, we designed forward primers that spanned two exons. The forward primer for exon 1A transcripts contained 14 bases that anneal to the 3\'-end of truncated exon 1A2, and 5 bases that anneal to the 5\'-end of exon 3. Similarly, the forward primer for exon 1B transcripts contained 18 bases that anneal to the 3\'-end of exon 1B, and 5 bases that anneal to the 5\'-end of exon 3. Each forward primer was used with a reverse primer specific to the 3\'-end of exon 3. In this way, only the smaller of the two splicing variants originating in each first exon was amplified (see Figure [1A](#F1){ref-type="fig"} and Table [1](#T1){ref-type="table"}). The levels of transcripts detected with these primer pairs were used to calculate the contribution of each ERV-derived promoter to total gene expression. As shown in Figure [4](#F4){ref-type="fig"}, exon 1A transcripts were approximately 10-fold more abundant than those initiating in exon 1B, accounting for 78% of total *SPAM1*expression in the testis compared to 7.6% for exon 1B transcripts. The 14% of *SPAM1*ORF transcripts not accounted for by these primer pairs most likely correspond to splicing variants 1A1 and 1B1, which were excluded from this analysis.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Primer positions and sequences
:::
Primer name^a^ Exon Sequence (5\'-3\')
---------------- ------ --------------------------
HGF1 6 CATGAGAAGTATGACAACAGCCTC
HGF2 8 TGGTCTCCTCTGACTTCAAC
HGR1 9 GTTGCTGTAGCCAAATTCGTTGTC
HGR2 9 CTGTAGCCAAATTCGTTGTC
HSF1 4 CTACACTCTATGTGCGCAATCG
HSF2 1A TAGCAGTGTAGGTGGTTAGCAG
HSF3 1B GGGTCTAATTGACTTGAGAATGTG
HSF4 4 TTTTTGCATATACCCGCATAG
HSF5 1A-3 TCCTTCCTAGCAAGGGATG
HSF6 1B-3 AAAAAATTAAGCTGAAATGGATG
HSR1 6 TTTGGCTGCTAGTGTGACGTTG
HSR2 4 CTGATGCAAAGTATGAGCACAG
HSR3 4 CATTCCAGGCCCAGAGGAAAG
HSR4 5 CCCATATTACAATTCCAGAAG
HSR5 3 AAGTCTGCTTTCAAAATCCAG
MGF 3 GTGGAGTCTACTGGTGTCTTC
MGR 5 GTGGCAGTGATGGCATGGAC
MSF1 4 GATGCTATGAGTTTAGCACAACG
MSF2 1 ATGATGGAGATGCGAGTGGTAG
MSR1 5 CATCAGATGTCTCCTTACATGTC
MSR2 3 TGTGGTCTGTTTAGTATTAGATGC
MSR3 3 TTCCTTCTTACACACTGTGGTC
a\) H, human; M, mouse; G, GAPDH; S, SPAM1; F, forward; R, reverse.
:::
The results obtained by non-quantitative RT-PCR (Figure [3](#F3){ref-type="fig"}) and quantitative, real-time RT-PCR (Figure [4](#F4){ref-type="fig"}) were generally similar. However, transcripts containing *SPAM1*ORF sequence were detected in the small intestine and muscle by the former method, but not the latter. The bands amplified from these tissues by non-quantitative RT-PCR were sequenced and were confirmed to correspond to the predicted *SPAM1*ORF transcript. 5\'-RACE analysis performed on human muscle total RNA identified a low level of transcripts initiating within promoter 1B, but no other *SPAM1*-specific transcripts (data not shown). These results suggest that the 35 cycles used for non-quantitative RT-PCR analysis amplified transcripts present at levels too low to be detected by real-time RT-PCR.
ERV1-derived promoter 1A is conserved in the mouse genome
---------------------------------------------------------
Initial analysis of *SPAM1*revealed that transcripts of the mouse orthologue, *Spam1*, also initiate within an ERV \[[@B23]\]. Examination of the mouse genomic sequence revealed that, as in human, the published 5\'-ends of the *Spam1*first exon (nucleotides +12, +21, +48 and +68 in Figure [5B](#F5){ref-type="fig"}\[[@B6],[@B19]\]) map within an antisense ERV1 MER34 *pol*region (Figure [5B](#F5){ref-type="fig"}). A similar overlap between *Spam1*transcripts and an antisense ERV1 *pol*region was observed in the rat genome (data not shown). This ERV1 element therefore inserted into the ancestral genome before the divergence of humans and rodents, approximately 80 million years ago.
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Genomic structure of murine *Spam1*.**(A) Overview of the *Spam1*locus. Exons are boxed and numbered, with the size in bp shown above. Intron sizes in kb are given below. The approximate positions of the *Spam1*ORF start and stop codons are indicated. The two *Spam1*splicing variants are shown schematically below. ERV-derived sequences are shown in green, with other sequences in blue. The diagram is not to scale. A black bar indicates the region shown in more detail in Figure 5B. (B) Sequence and position of *Spam1*exons 1 and 2 with respect to TE sequence. The sequence shown corresponds to mouse chromosome 6 co-ordinates 24623802 to 24624821 in the UCSC genome browser (May 2004 release). The solid line frames the sequence annotated as ERV1 MER34 *pol*sequence in the RepeatMasker database. *Spam1*exon sequences are shown in upper case reverse type. Splicing variant 2 sequences are shown in bold type. The transcription start sites identified by 5\'-RACE are underlined; splice donor and acceptor sites are italicized. A non-consensus CRE is shown in bold underlined type.
:::
:::
The RepeatMasker track of the UCSC genome browser annotated only a 342 bp region of the murine *Spam1*locus as ERV1 *pol*sequence; the analogous *pol*fragment containing human *SPAM1*exon 1A is considerably larger (compare Figure [2](#F2){ref-type="fig"} and [5B](#F5){ref-type="fig"}). Due to the higher neutral mutation rate in mouse \[[@B24]\], murine TEs that inserted prior to the primate-rodent split are roughly twice as diverged as the orthologous human element, making detection by repeat-finding programs difficult. Older elements in rodents are therefore often excluded from annotation as TEs in the RepeatMasker database \[[@B24],[@B31]\]. We used the DOTTER sequence comparison program to determine whether this was the case for the ERV1 sequence associated with *Spam1*. We extracted the human genomic DNA sequence containing the full-length exon 1A1 and 1000 bp of upstream sequence (1296 bp in total) from the UCSC genome browser. We also extracted the mouse genomic sequence containing exon 1 and 734 bp of upstream sequence (a total of 858 bp). A Dot Plot showing a comparison of these two sequences is shown in Figure [6A](#F6){ref-type="fig"}.
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**Comparison of the genomic sequence upstream of the human and mouse *SPAM1*genes.**(A) DOTTER comparison of the human (horizontal axis) and mouse (vertical axis) genomic DNA sequences upstream of the *SPAM1 / Spam1*locus. Nucleotide positions in bp are given on the upper horizontal and left vertical axes. The approximate position of the human ERV1 MER34 *pol*region is shown below the lower horizontal axis. The approximate positions of the annotated mouse ERV1 MER34 *pol*region (solid box) and of the extended *pol*sequence (dashed box) are shown on the right hand side. The 5\'-most transcriptional start site of each gene is represented by a bent arrow. An asterisk marks the approximate position of a conserved CRE in the proximal human and mouse promoters. (B) Multi-species alignment of the well-conserved sequence marked with an asterisk in Figure 6A. Nucleotides identical between the human and mouse sequence are joined by vertical lines. Nucleotides identical between the mouse and rat sequence are highlighted in gray. The mouse-rat alignment is incomplete in this region due to a small (56 bp) insertion into the rat sequence. The MER34 ERV1 consensus sequence is shown above the human *SPAM1*promoter sequence; nucleotides identical between the two sequences are shown in reverse type. Solid lines above and below the sequence indicate the position of the conserved CRE. The 5\'-most human and mouse transcriptional start sites are marked with bent arrows.
:::
:::
The sequence annotated as the ERV1 *pol*region in the human genome corresponds to nucleotides 246 -- 1296 in Figure [6A](#F6){ref-type="fig"} (nucleotides -754 to +296 in Figure [2](#F2){ref-type="fig"}). The positions of the ERV1 *pol*region and the exon 1A transcriptional start site are shown below the lower horizontal axis. The mouse genomic sequence from approximately nucleotide 100 -- 800 in Figure [6A](#F6){ref-type="fig"} shows some sequence similarity to nucleotides 300 -1050 of the human sequence. Therefore the region of the mouse *Spam1*locus derived from the ERV1 *pol*region is considerably larger than that annotated by RepeatMasker, extending approximately 700 bp upstream of the transcriptional start site. The positions of the annotated and extended ERV1 *pol*regions are represented by solid and dashed boxes, respectively, on the right hand side of Figure [6A](#F6){ref-type="fig"}. A similar DOTTER result was observed upon comparison of the corresponding rat and human genomic sequences (data not shown).
The level of sequence similarity between the human and mouse *SPAM1*promoter regions is highest at position 900 -- 950 in the human sequence and 650 -- 700 in the mouse (Figure [6A](#F6){ref-type="fig"}, region marked with asterisk). A sequence comparison revealed that this conserved region contains the functional CRE identified in the murine *Spam1*promoter (Figure [6B](#F6){ref-type="fig"}, reference \[[@B19]\]). The relatively high level of primate -- rodent conservation of this element and the surrounding sequence indicates that this region may be functionally important.
We performed 5\'-RACE on mouse testis RNA to identify the transcriptional start site(s) and to search for alternative promoters of *Spam1*. As shown in Figure [5](#F5){ref-type="fig"}, a single first exon with multiple transcriptional start sites was identified for *Spam1*. This exon is orthologous to exon 1A of the human gene (Figure [6A](#F6){ref-type="fig"}). No sequence equivalent to human exon 1B was detected in mouse *Spam1*transcripts. Two splicing variants were identified for the mouse *Spam1*gene. Variant 2 utilized an alternative transcription start site and splice donor site within exon 1 to generate a truncated first exon, and spliced into a short (35 bp) novel downstream exon (Figure [5](#F5){ref-type="fig"}). As with human *SPAM1*, the murine splicing variants affect only the 5\'-UTR, leaving the downstream ORF intact (Figure [5A](#F5){ref-type="fig"}).
Expression of the mouse *Spam1*gene is largely testis-specific
--------------------------------------------------------------
We performed non-quantitative RT-PCR on a panel of normal C57BL/6 mouse tissues to determine the expression pattern of *Spam1*. As shown in Figure [7](#F7){ref-type="fig"}, primers specific to the *Spam1*ORF detected transcripts only in the testis. Transcripts initiating within the ERV1 *pol*region were detected primarily in the testis, and to a lesser degree in the kidney. As with the human gene, the ERV1 *pol*-derived promoter of murine *Spam1*is therefore largely testis-specific.
::: {#F7 .fig}
Figure 7
::: {.caption}
######
**Detection of *Spam1*transcripts by RT-PCR.**Primer pairs specific for *Gapdh*, the *Spam1*ORF, and *Spam1*transcripts initiating within the ERV sequence were used in RT-PCR assays. The assays were carried out on cDNAs derived from a range of normal C57/BL6 mouse tissues: 1, brain; 2, heart; 3, kidney; 4, liver; 5, lung; 6, colon; 7, small intestine; 8, spleen; 9, stomach; 10, thymus; 11, muscle; 12, testis; 13, placenta. Approximate molecular weights are indicated on the left.
:::
:::
Discussion
==========
In this study we have experimentally confirmed a previous *in silico*observation \[[@B23]\] that transcription of the human and murine *SPAM1*genes initiates within an antisense ERV common to both species. *SPAM1*is the only hyaluronidase gene to initiate within an ERV (data not shown); this TE insertion therefore took place after the small segmental duplication of three ancestral hyaluronidase genes, but before the divergence of primates and rodents. Interestingly, human *HYAL4*, but not its mouse orthologue, appears to initiate within an antisense LINE1 element (\[[@B23]\] and our unpublished observations). This element therefore inserted after the primate-rodent split, indicating an ongoing contribution by TEs to human hyaluronidase transcriptional regulation.
A previous study by our group determined that TE insertions were more common in transcripts with a high K~a~/K~s~value \[[@B23]\]. The K~a~/K~s~ratio for the human-Old World monkey SPAM1 orthologous pair is high at 0.57 \[[@B32]\]. This is in line with our hypothesis that TE insertions are more likely to be tolerated by rapidly-evolving genes \[[@B23]\]. High K~a~/K~s~ratios are a common characteristic of primate genes that are involved in male reproduction. This may be due to positive selection, driven by competition between the sperm of individual males of the more promiscuous primate species \[[@B32]\]. In the case of SPAM1, the requirement for species-specific sperm-zona pellucida recognition may also have contributed to the high inter-species divergence of the protein sequence.
We have identified two closely-spaced ERV-derived promoters for human *SPAM1*. Both were active primarily in the testis, albeit with an approximately 10-fold difference in promoter activity. This close physical proximity and similar tissue specificity suggests that the two promoters may be regulated by a shared testis-specific enhancer element, rather than by individual tissue-specific proximal promoter regions. We have also identified alternative splicing variants for the human and mouse genes. Alternatively-spliced transcripts of *HYAL1*and *HYAL3*have been described that cause an in-frame deletion of the putative catalytic site and abolish hyaluronidase function \[[@B33]\]. Evidence from the NCBI database suggests that an alternative splicing event in *SPAM1*exon 6 generates an extended 3\'-transcript, which encodes a C-terminally truncated SPAM1 isoform. However, the presence of this splicing variant has yet to be confirmed. In contrast, the alternatively-spliced transcripts of *SPAM1*and *Spam1*described in this study differ only in the sequence and length of the 5\'-UTR, and are not predicted to affect enzyme function. Changes in the 5\'-UTR sequence may however alter the stability and / or translation efficiency of the transcripts (reviewed in \[[@B34]\]), and hence impact indirectly on SPAM1 expression.
We have shown that all *SPAM1 / Spam1*alternative promoter and splicing variants are expressed primarily in the testis. Lower levels of expression were also observed in the human prostate and murine kidney. This contradicts previous reports that human *SPAM1*is expressed in the placenta \[[@B2]\] and that murine *Spam1*is expressed in tissues of the female reproductive tract \[[@B7]\]. Expression of SPAM1 is confined to a subset of specialized cells in some tissues \[[@B5],[@B7]\], which may explain these contradictory results.
In contrast to all known examples of host gene transcriptional regulation by ERVs, *SPAM1*and *Spam1*initiate not within an LTR, but rather within a fragment of the *pol*coding region. While the *SPAM1 / Spam1*promoters have not yet been fully analyzed, a non-consensus CRE at position -39 has been shown to be important for activity of the murine *Spam1*promoter in an *in vitro*testis system \[[@B19]\]. This site, and a similar sequence in the human promoter, are clearly derived from the ERV1 *pol*region and are well conserved between the two species (Figures [2](#F2){ref-type="fig"}, [5B](#F5){ref-type="fig"}, and [6](#F6){ref-type="fig"}). Various lines of evidence suggest that SPAM1 expression is regulated by sex hormones: the expression of SPAM1 in the male and female reproductive organs; the increased expression of *Spam1*in male kidney compared to female \[[@B8]\]; the seasonal variation in *SPAM1*expression in red fox testis \[[@B35]\]; and the variations in murine female SPAM1 expression at different stages of estrus \[[@B7]\]. Indeed, various groups have identified putative androgen response elements (AREs) in the *SPAM1*and *Spam1*promoters \[[@B5],[@B6],[@B19]\], and estrogen response elements (EREs) in the *Spam1*promoter \[[@B7]\]. Many of these predicted sites also map within the ERV *pol*region. However, none of these sequences represents a consensus binding site, and none has yet been shown to bind its cognate transcription factor or to be required for SPAM1 expression.
Alternatively, hormonal regulation may be mediated through the CRE. Androgen treatment of Sertoli cells was recently shown to rapidly induce phosphorylation of a CRE binding protein and activate transcription of target genes via the MAPK pathway \[[@B36]\]. This mechanism was postulated to represent a common mechanism for activation of testis-specific promoters that do not contain a consensus ARE. Much work remains to be done to elucidate the mechanisms of transcriptional regulation of *SPAM1*and *Spam1*. However, it is clear that at least one functional transcription factor binding site is derived from the ERV1 *pol*region.
ERV LTRs contain the regulatory signals necessary for transcription of the retroviral genes. Insertion of an LTR sequence near a host gene could therefore provide a novel, pre-formed regulatory unit and be rapidly adopted by the gene for use as an alternative promoter. It is less clear how a retroviral protein coding region, which has no known function in transcriptional regulation, could be adopted for use as a promoter by a host gene. We suggest the following scenario.
Prior to the primate-rodent divergence, an ERV inserted upstream of the ancestral *SPAM1*gene, in the antisense orientation. By chance, the antisense *pol*coding region contained sequences that were similar to a CRE, and possibly to other transcription factor binding sites necessary for testis-specific transcription. The region of the human *SPAM1*promoter that contains the CRE is quite divergent from the MER34 consensus sequence (Figure [6B](#F6){ref-type="fig"}). It is therefore unlikely that the CRE was functional, and hence preserved by purifying selection, from the time of ERV insertion. The CRE present in the modern *SPAM1*and *Spam1*promoters is more likely to have evolved by random nucleotide substitution from a similar sequence in the original antisense *pol*gene. The \~50 bp of genomic sequence that contains the CRE is relatively well conserved between human and rodents (Figure [6](#F6){ref-type="fig"}), indicating that purifying selection of this sequence probably occurred at some time after the creation of the functional CRE. The evolutionary origins of other functional transcription factor binding sites in the modern *SPAM1 / Spam1*promoters remain to be determined.
The selective processes driving the evolution of a promoter from a protein coding sequence, and the fate of the original ancestral *SPAM1*promoter, remain unknown. This gene therefore represents an extremely intriguing example of how the host genome can adopt \"parasitic\" ERV sequences for its own purposes.
Conclusion
==========
We have shown that transcription of the human and mouse *SPAM1*genes initiates within an antisense ERV *pol*gene. The first exons and proximal promoters of both genes are derived from this ancient ERV *pol*sequence. Expression of the human and mouse *SPAM1*genes is largely testis-specific, and we have provided evidence that testis-specific transcription factor binding sites are derived from conserved ERV sequence in both species. *SPAM1*can therefore be added to the growing list of mammalian genes that are regulated by TEs. This gene represents the first known example of the evolution of promoter function from an ERV coding sequence, and of gender-specific transcription from an ERV-derived promoter.
Methods
=======
Computational methods
---------------------
The human, mouse and rat *SPAM1*/ *Spam1*loci were examined using the University of California, Santa Cruz genome browser \[[@B37]\]. Homology searches were performed using the Basic Local Alignment Search Tool (BLAST, \[[@B38]\]). The SPIDEY alignment program \[[@B39]\] was used to compare cDNA and genomic DNA sequences for all splicing variants and for 5\'-RACE clones. Human and mouse genomic DNA sequences were compared using the DOTTER program \[[@B40]\].
Reverse transcription and RT-PCR
--------------------------------
C57BL/6 mouse testis total RNA and all human total RNAs were purchased from Clontech. All other mouse RNAs were extracted from C57BL/6 mouse tissues using TRIzol (Invitrogen) according to the manufacturer\'s instructions. 5 μg of each RNA was treated with DNase I and reverse transcribed as described \[[@B41]\]. 35 cycles of RT-PCR were performed using Taq DNA polymerase with 2 ng/μl of each primer in 4 mM MgCl~2~. Primer pairs were as follows. *GAPDH*, HGF1 & HGR1; *SPAM1*ORF, HSF1 & HSR1; *SPAM1*Exon 1A, HSF2 & HSR2; *SPAM1*Exon 1B, HSF3 & HSR2; *Gapdh*, MGF & MGR; *Spam1*ORF, MSF1 & MSR1; *Spam1*ERV, MSF2 & MSR2. All primer positions and sequences are given in Table [1](#T1){ref-type="table"}.
5\'-RACE
--------
5\'-RACE analysis of human or mouse testis total RNA was carried out using the FirstChoice RLM-RACE kit (Ambion) as described \[[@B42]\]. HSR3 and HSR2 were used as the outer and inner primers, respectively, for nested RT-PCR amplification of *SPAM1*5\'-RACE products. MSR3 and MSR2 were used as the equivalent mouse primers.
Real-time RT-PCR
----------------
Real-time quantification of transcript levels was carried out as described \[[@B42]\]. Dissociation curves demonstrated that each primer pair amplified a single product. Standard curves were prepared for each primer pair using serial dilutions of human testis cDNA to enable calculation of the relative abundance of each transcript type. The level of *SPAM1*ORF transcripts for each tissue was normalized to *GAPDH*and expressed relative to the level detected in heart cDNA. The relative amounts of *SPAM1*ERV1A and ERV1B transcripts were assessed only in testis cDNA. The level of each transcript was divided by the amount of ORF transcript detected in testis. This value was then multiplied by the *GAPDH*- and heart-normalized level of ORF transcripts to determine the contribution of each ERV promoter to total *SPAM1*expression. Primer pairs were as follows. *GAPDH*, HGF2 & HGR2; Total *SPAM1*, HSF4 & HSR4; *SPAM1*Exon 1A, HSF5 & HSR5; *SPAM1*Exon 1B, HSF6 & HSR5.
List of abbreviations
=====================
ARE, androgen response element; CRE, cAMP response element; CREM, CRE modulator; ERE, estrogen response element; ERV, endogenous retrovirus; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; LINE, long interspersed nuclear element; LTR, long terminal repeat; MAPK, mitogen activated protein kinase; ORF, open reading frame; RACE, rapid amplification of cDNA ends; SINE, short interspersed nuclear element; SPAM1, sperm adhesion molecule 1; TE, transposable element; UCSC, University of California at Santa Cruz; UTR, untranslated region.
Authors\' contributions
=======================
CAD carried out all experimental work described in the paper, participated in the design of the study, and drafted the manuscript. DLM conceived of the study, participated in its design, and participated in the drafting and critical revision of the manuscript. Both authors read and approved the final manuscript.
Acknowledgements
================
The authors wish to thank the following members of the group for their help: Louie van de Lagemaat for assistance with the DOTTER program and other bioinformatic methods; Mark Romanish, Arefeh Rouhi and Brian Wilhelm for obtaining mouse tissue samples and sharing RNA stocks; Leanne Gutierrez for assisting in the preparation of Figure [1A](#F1){ref-type="fig"}; Greg Baillie for assistance with the DOTTER program, and for helpful comments on the manuscript. This work was supported by a grant from the Canadian Institutes of Health Research.
|
PubMed Central
|
2024-06-05T03:55:55.252731
|
2005-4-1
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079825/",
"journal": "BMC Genomics. 2005 Apr 1; 6:47",
"authors": [
{
"first": "Catherine A",
"last": "Dunn"
},
{
"first": "Dixie L",
"last": "Mager"
}
]
}
|
PMC1079826
|
Background
==========
It was observed that screening of libraries derived by expression cloning for new gene products with a given catalytic activity can be limited by formation of inclusion bodies. In these cases it would be desirable to construct libraries which a higher fraction of soluble gene products. Enrichment of soluble proteins could be achieved either by limiting the screening to coding sequences from phylogenetic groups with mainly soluble proteins or by the use of degenerated primers which are specific for homologous families with mainly soluble proteins.
Alternatively solubility can be improved by the use of fusion proteins \[[@B2]\] like NusA, MBP, Thioredoxin, GrpE, BFR, GST, DsbA \[[@B3]\], the N- terminal domain of IF2 \[[@B4]\] or phage coat protein III \[[@B5]\]. In addition fusion proteins allow affinity chromatography and thus simplify purification \[[@B6]\]. However in some cases the fusion partner has to be removed after protein purification \[[@B7]\] and therefore constitutes and additional step in protein preparation. Other strategies try to improve *in vivo*solubility of a recombinant protein by protein engineering using strategies like molecular evolution \[[@B8],[@B9]\] or rational protein design. Examples for protein design are the insertion of positively charged residues into hydrophobic patches on the surface \[[@B10]\], exchange of phenylalanines by serines \[[@B11]\] or asparagine residues by aspartic acids \[[@B12]\]. It has been shown that single residues can have a major impact on solubility \[[@B13]-[@B16]\]. But also the expression system is important for *in vivo*solubility \[[@B2],[@B17]\]. Other factors are cultivation and disruption conditions \[[@B18]-[@B20]\], rate of protein synthesis \[[@B21]\], fermentation temperature \[[@B20],[@B22]\] and the amount of helper protein \[[@B12],[@B23]-[@B26]\]. Fusion proteins have already been successfully used in high throughput expression- studies to improve solubility \[[@B8],[@B27]-[@B30]\].
However, engineering approaches are limited to specific proteins or growth conditions and are hardly applicable to high throughput expression. Therefore, in projects where libraries are screened for activity, solubility is always an implicit criterion. Previously Harrison et al. introduced a two-parameter model based on an analysis of 81 proteins for which experimental data on solubility exists \[[@B1],[@B3]\]. The model proposes that solubility of proteins in *E. coli*at physiological conditions depends mainly on protein charge and the relative number of turn-forming residues. Interestingly, the model holds for a broad variety of proteins.
We applied this model to a comprehensive analysis of the Lipase Engineering Database (LED) \[[@B31],[@B32]\] which includes more than 1800 α/β-hydrolases. α/β-hydrolases share the same fold but are highly diverse in sequence. They are ubiquitous and include cellular and secreted proteins from a wide range of organisms.
The aim of this study was to find correlations between predicted solubility in *E. coli*and protein size or phylogenetic origin. Homologous families and superfamilies were analysed for the predicted solubility of their members.
Results
=======
Kingdoms of life
----------------
According to Harrison et al. \[[@B1],[@B3]\] the canonical variable CV-CV\' predicts from the protein sequence whether a recombinantly expressed protein is soluble in the cytoplasm of *E. coli*(CV-CV\' \< 0) or will form inclusion bodies (CV-CV\' \> 0). For CV-CV\' = 0 the probability of solubility is 0.5. The probability of solubility or insolubility rises for higher absolutes of CV-CV\'. An analysis of the Lipase Engineering Database indicates that most of the hydrolases are predicted to be insoluble in *E. coli*, with a major peak at CV-CV\' = 0.8 and a minor peak at CV-CV\' = 0.2 (Figure [1](#F1){ref-type="fig"}). A separate analysis of hydrolases from eukaryotic and prokaryotic origin (679 and 686 hydrolases, respectively) demonstrates that these two peaks are formed predominantly by the hydrolases from each of the two kingdoms of life. The distribution of CV-CV\' of bacterial hydrolases is characterized by an average of 0.42 and a first quartile of -0.18. Because the first quartile indicates the minimum solubility of the 25 % most soluble hydrolases, more than 25 % of all bacterial hydrolases are predicted to be soluble. In contrast, for eukaryotic hydrolases the average of CV-CV\' is 0.94; the first quartile is 0.58. Thus, only a small fraction of eukaryotic hydrolases is predicted to be soluble in *E. coli*. α/β-hydrolases from archaea were not investigated because they are only represented with 23 hydrolases.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Distribution of CV-CV\' of bacterial (blue), eukaryotic (green), and all (black) α/β-hydrolases.
:::
:::
Thus, in general, bacterial α/β-hydrolases are predicted to be more soluble in *E. coli*than eukaryotic α/β-hydrolases.
Protein size
------------
The family of α/β-hydrolases falls into three major groups of protein size: small (150--380 amino acids), medium- sized (380--520 amino acids) and large hydrolases (more than 520 amino acids) (Figure [2](#F2){ref-type="fig"}). Large hydrolases are mainly from eukaryotic origin while small hydrolases are mainly from bacterial origin. Correlating sequence length and predicted solubility in *E. coli*demonstrates that large hydrolases are predicted to be less soluble in *E. coli*than smaller ones (Figure [3](#F3){ref-type="fig"}). Hydrolases predicted to be soluble in *E. coli*(CV-CV\' \< 0) have sequence lengths between 200 and 400 amino acids, most hydrolases of more than 400 amino acids are predicted to be insoluble in *E. coli*(CV-CV\' \> 0). A high fraction of bacterial and archaean small hydrolases is predicted to be soluble in *E. coli*, while eukaryotic small hydrolases are predicted to be mainly insoluble. The two outliers with CV-CV\' \< -1 and a sequence length larger than 650 are a lipase from *Staphylococcus xylosus*(AAG35726) and CG6296 from *Drosophila melanogaster*(AAF56648). Both are putative proteins and have not yet been expressed in *E. coli*. Thus, there are two general observations: (1) small α/β-hydrolases are predicted to be more soluble in *E. coli*than large α/β-hydrolases and (2) eukaryotic α/β-hydrolases are larger than prokaryotic α/β-hydrolases.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Distribution of sequence length of bacterial (blue), eukaryotic (green), and all (black) α/β-hydrolases.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Correlation between CV-CV\' and the length of protein sequences in bacterial (blue), eukaryotic (green), and archaean (red) α/β-hydrolases.
:::
:::
Analysis of groups formed by protein size
-----------------------------------------
To distinguish the effects of hydrolase size and phylogenetic origin on solubility, small, medium- sized and large α/β-hydrolases were investigated separately, and hydrolases were grouped by phylogeny of their origin (Tables [1](#T1){ref-type="table"}, [2](#T2){ref-type="table"}, [3](#T3){ref-type="table"}).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Hydrolases from 171 to 379 amino acids sorted by taxa of their host n is the number of hydrolases in the group, charge the charge per residue, turn the relative number of turn- forming residues. Only groups with at least eight hydrolases were included.
:::
**Taxon** **n** **CV-CV\' average** **CV-CV\' first quartile** **Charge average** **Turn average** **Length average**
------------------ ------- --------------------- ---------------------------- -------------------- ------------------ --------------------
d-proteobacteria 8 0.02 -0.87 -0.062 0.230 267
enterobacteria 29 0.02 -0.28 -0.061 0.229 287
a-proteobacteria 66 0.14 -0.26 -0.054 0.223 300
actinobacteria 127 0.23 -0.49 -0.060 0.240 291
proteobacteria 302 0.28 -0.23 -0.052 0.230 293
eubacteria 558 0.29 -0.28 -0.050 0.229 288
archaea 22 0.30 -0.43 -0.052 0.229 272
g-proteobacteria 149 0.31 -0.27 -0.052 0.232 291
b-proteobacteria 46 0.55 -0.04 -0.044 0.236 298
plants 52 0.57 -0.23 -0.048 0.239 301
ascomycetes 40 0.89 0.32 -0.046 0.257 267
fungi 43 0.91 0.37 -0.046 0.258 268
eukaryota 235 0.94 0.36 -0.040 0.250 284
chordata 72 0.98 0.85 -0.036 0.247 269
mammalia 65 1.00 0.88 -0.035 0.245 267
metazoa 140 1.10 0.78 -0.035 0.252 283
arthropoda 48 1.38 0.86 -0.031 0.263 300
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Hydrolases from 380 to 519 amino acids sorted by taxa of their host n is the number of hydrolases in the group, charge the charge per residue, turn the relative number of turn- forming residues. Only groups with at least eight hydrolases were included.
:::
**Taxon** **n** **CV-CV\' average** **CV-CV\' first quartile** **Charge average** **Turn average** **Length average**
------------------ ------- --------------------- ---------------------------- -------------------- ------------------ --------------------
actinobacteria 15 0.30 -0.33 -0.060 0.246 439
ascomycetes 15 0.52 0.09 -0.047 0.235 447
fungi 21 0.67 0.25 -0.046 0.242 435
eubacteria 79 0.84 0.30 -0.044 0.251 437
proteobacteria 38 0.89 0.44 -0.042 0.250 436
g-proteobacteria 11 0.96 0.59 -0.055 0.279 454
arthropoda 40 1.04 0.50 -0.034 0.246 441
eukaryota 165 1.04 0.72 -0.035 0.246 443
a-proteobacteria 19 1.08 0.56 -0.030 0.239 439
metazoa 136 1.08 0.76 -0.033 0.245 444
chordata 76 1.25 0.99 -0.030 0.249 448
mammalia 71 1.25 0.98 -0.031 0.251 446
plants 8 1.42 1.10 -0.034 0.268 452
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Hydrolases larger than 519 amino acids sorted by taxa of their host n is the number of hydrolases in the group, charge the charge per residue, turn the relative number of turn- forming residues. Only groups with at least eight hydrolases were included.
:::
**Taxon** **n** **CV-CV\' average** **CV-CV\' first quartile** **Charge average** **Turn average** **Length average**
------------------ ------- --------------------- ---------------------------- -------------------- ------------------ --------------------
arthropoda 98 0.75 0.51 -0.048 0.251 606
metazoa 262 0.85 0.56 -0.046 0.254 649
eukaryota 279 0.88 0.58 -0.045 0.255 644
mammalia 95 0.88 0.56 -0.046 0.256 720
chordata 108 0.90 0.58 -0.046 0.257 705
enterobacteria 8 0.99 0.74 -0.054 0.279 633
eubacteria 49 1.12 0.76 -0.047 0.276 654
g-proteobacteria 32 1.17 0.79 -0.049 0.285 643
ascomycetes 13 1.21 0.88 -0.046 0.278 552
proteobacteria 35 1.22 0.79 -0.048 0.286 638
fungi 14 1.25 0.88 -0.045 0.279 553
:::
Small hydrolases have an average of CV-CV\' of 0.48 and a first quartile of -0.17, thus more than 25 % of small hydrolases are predicted to be soluble in *E. coli*. All taxa have a positive average of CV-CV\' (Table [1](#T1){ref-type="table"}), indicating that less than 50 % of the hydrolases in each taxon are predicted to be soluble. Hydrolases from eukaryotic taxa are at average predicted to be highly insoluble. All prokaryotic taxa and plants have negative first quartiles of CV-CV\', thus more than 25 % of their hydrolases are predicted to be soluble. All other eukaryotic taxa have positive first quartiles of CV-CV\', thus most of their hydrolases are predicted to be insoluble. The taxa containing hydrolases with the highest predicted solubility average are from bacteria, the taxa containing hydrolases with the lowest predicted solubility average are from eukaryota.
Medium- sized hydrolases have an average of CV-CV\' of 0.98 and a first quartile of 0.54. Hydrolases of all taxa have a positive average of CV-CV\', and thus are mainly predicted to be insoluble (Table [2](#T2){ref-type="table"}). Only actinobacteria have a negative first quartile of CV-CV\', indicating that more than 25 % of its hydrolases are predicted to be soluble, all other taxa have positive first quartiles of CV-CV\'. The taxa containing hydrolases with the highest predicted solubility average are from bacteria and fungi. The taxa containing hydrolases with the lowest predicted solubility average are from metazoa and plants.
Large hydrolases have an average of CV-CV\' of 0.93 and a first quartile of 0.61, they are in general predicted to be highly insoluble. All taxa have positive averages and first quartiles of CV-CV\', and thus most of their hydrolases are predicted to be insoluble (Table [3](#T3){ref-type="table"}). The taxa containing hydrolases with the highest predicted solubility average are from metazoa, the taxa containing hydrolases with the lowest predicted solubility average are from bacteria and fungi.
In general CV-CV\' from bacterial hydrolases is much lower if the hydrolase is small, while metazoan hydrolases have a lower average and first quartile of CV-CV\' if the hydrolase is large. Though large metazoan hydrolases have a higher probability of solubility than small metazoan hydrolases, few large hydrolases are predicted to be soluble in *E. coli*(Figure [3](#F3){ref-type="fig"}). This is consistent with the result from Table [3](#T3){ref-type="table"} that in the analysis of large hydrolases no taxon with a negative first quartile of CV-CV\' could be found.
Thus there are several conclusions. Large α/β-hydrolases from bacteria are predicted to be less soluble than smaller ones. Small bacterial α/β-hydrolases are predicted to be more soluble than both small and large eukaryotic α/β-hydrolases. But large hydrolases from metazoa are predicted to be more soluble than large hydrolases from bacteria and small hydrolases from metazoa. So there seems to be a principal difference between metazoa and bacteria. Fungi, especially ascomycetes, behave differently. Their α/β-hydrolases with the highest predicted solubility are mainly medium- sized.
Analysis of solubility by genera
--------------------------------
Of the 257 genera represented in the database most include only a few hydrolases. Therefore the 29 genera with at least ten hydrolases were analyzed (Table A in the supplementary file [1](#S1){ref-type="supplementary-material"}). Most hydrolases of all eukaryotic taxa with the exception of *Oryza*are predicted to be highly insoluble in *E. coli*.
Hydrolases from bacterial genera show a wide range of averages of predicted solubility. Interestingly the bacterial genera with the highest and the lowest average of CV-CV\' (*Mycobacterium*and *Rhodococcus*) are both from actinobacteria. The third large genus of actinobacteria, *Streptomyces*, is predicted to include more than 25 % soluble hydrolases as it is shown by a negative first quartile.
Thus the averages of different genera of a taxon may differ completely in predicted solubility. Therefore a strategy to identify α/β-hydrolases from actinobacteria that are soluble in *E. coli*should focus on proteins from *Rhodococcus*and *Streptomyces*, but not from *Mycobacterium*.
Analysis of solubility by sequence similarity
---------------------------------------------
Hydrolases with sequence similarity have been assigned to superfamilies which were analysed (Table B in the supplementary file [1](#S1){ref-type="supplementary-material"}). Superfamilies with mostly hydrolases of high predicted solubility contain mainly intracellular α/β-hydrolases from bacteria (supplementary file [2](#S2){ref-type="supplementary-material"}). The averages of CV-CV\' of the superfamilies that contain almost exclusively bacterial hydrolases range from -0.77 to 0.36. Metazoan, fungal and secreted hydrolases can mainly be found in superfamilies with averages of CV-CV\' between 0.38 (hormone sensitive lipases) and 1.45 (cutinases) (supplementary file [2](#S2){ref-type="supplementary-material"}). In general, the average of CV-CV\' depends on sequence length. However, the superfamily with the lowest sequence length in this table (Table B in the supplementary file [1](#S1){ref-type="supplementary-material"}), *Bacillus*lipase, has the lowest predicted solubility.
Proteins from cytosolic hydrolases, a large superfamily that contains mainly epoxide hydrolases and haloalkane dehalogenases, are in general predicted to be more soluble than proteins from most other superfamilies. The average of CV-CV\' is 0.29 and the first quartile is -0.22. This superfamily was chosen for a more detailed analysis of homologous families (Table C in the supplementary file [1](#S1){ref-type="supplementary-material"}). The CV-CV\' averaged over hydrolases of each homologous family of this superfamily ranges from -0.29 to 0.86. This shows that there can be the prediction of families with mostly soluble and of families with mostly insoluble proteins in one superfamily. This observation is analogous to the previous observation with genera and higher taxa. As expected the homologous families with the highest predicted solubility average are dominated by bacterial hydrolases and the families with proteins of low predicted solubility rather contain eukaryotic hydrolases (supplementary file [2](#S2){ref-type="supplementary-material"}). The homologous family with the highest predicted solubility average, which is dominated by eukaryots, is the \'soluble plant epoxide hydrolases\'. Almost 50 % of its members are predicted to be soluble in *E. coli*.
Thus the averages of different homologous families of a superfamily may differ significantly in predicted solubility.
Discussion
==========
The solubility model
--------------------
The dataset of the statistical solubility model were 81 highly diverse proteins, for which solubility data exists, from a wide range of organisms \[[@B1]\]. The published prediction accuracy of the five parameter model was 76 % for soluble proteins and 91 % for insoluble proteins \[[@B1]\], the overall accuracy for solubility prediction of proteins from the dataset was 88% \[[@B1]\]. It has been discovered that only two out of the five parameters are critical for distinguishing between soluble and insoluble proteins \[[@B3]\]. Therefore, CV-CV\', the indicator of predicted solubility, was derived by two terms: the total charge calculated by the relative numbers of arginines, lysines, aspartic acids and glutamic acids, and the relative number of turn forming residues calculated by the relative number of asparagines, glycines, prolines and serines. Those two parameters show a level of significance of 100% \[[@B1]\].
Most of the variation of CV-CV\' among α/β-hydrolases is caused by the charge term. A high negative or positive charge results in the best predicted solubility. However, the pK value of a titratable group is highly dependent on the proteins structure. Short- and long-range interactions can lead to pK shifts of more than two units, changing the total protein charge by more than five unit charges \[[@B33],[@B34]\].
An evaluation of nine frequently used fusion proteins used to improve solubility in *E. coli*\[[@B3]-[@B5]\] showed that seven proteins indeed are predicted to be soluble in *E. coli*(data not shown). The exceptions are MBP which has a CV-CV\' of 0.23, and phage coat protein III which has a CV-CV\' of 0.62 (data not shown).
To test the model in the prediction of solubility of hydrolases, 35 hydrolases from the PDB which were annotated as expressed in *E. coli*were examined (Table D in the supplementary file [1](#S1){ref-type="supplementary-material"}). Because good solubility is prerequisite for successful crystallization, this group of proteins served as a positive control for the predictive value of our method. 24 of 35 hydrolases are predicted to be soluble in *E. coli*(CV-CV\' \< 0). For all 35 hydrolases, the average of CV-CV\' is -0.23, the first quartile is -0.80. Thus, the predicted average solubility of this group of proteins is much higher than any single protein family in the database, including proteins from enterobacteria and the genus *Escherichia*which have an average of CV-CV\' of 0.24 and 0.10 respectively (data not shown), which is a strong support for the reliability of this method.
In addition, the observation that substitution of asparagines by aspartic acids in DsbA-IGFBP-3 fusion proteins improved solubility \[[@B12]\] is consistent with the solubility formula because the fusion protein is already predicted to be negatively charged (data not shown). Thus, increasing the negative charge increases predicted solubility. Similarly, incorporation of solvent exposed positive charged amino acids improved solubility of consensus ankyrin repeat proteins \[[@B10]\]. Here the net charge proposed by the solubility formula was zero (data not shown), so solubility is predicted to be increased by insertion of positively or negatively charged amino acids.
However, the solubility formula exclusively depends on sequence information and neglects the structural context. Therefore, in some cases this simple model resulted in wrong predictions. In the model, substitution of multiple phenylalanine residues by serine \[[@B11]\] led to a predicted increase of the ratio of turn-forming residues in the formula and thus to a lowered predicted solubility in *E. coli*. Instead, an increase in solubility was observed experimentally \[[@B11]\] which can be explained by the structural context: replacement of solvent- accessible phenylalanines increased polarity of the proteins surface and thus its solubility. The fact that single residues can have a huge impact on *in vivo*solubility \[[@B13]-[@B16]\] is not always fully explainable by the solubility formula. Therefore, for the design of proteins with higher solubility the formula should be combined with a careful investigation of the structural context.
Solubility of α/β-hydrolases
----------------------------
Protein size was not included as a parameter in the solubility model, because it showed no significant difference between soluble and insoluble proteins of the dataset \[[@B1]\]. However, the analysis of groups formed by α/β-hydrolase size revealed that both protein size and phylogeny of organisms have a major impact on predicted solubility. Statements like \'Small hydrolases are predicted to be more soluble than large hydrolases\', \'Bacterial hydrolases are predicted to be more soluble than eukaryotic hydrolases\' are generally true, but simplifications.
For small to medium- sized hydrolases, bacterial hydrolases are predicted to be more soluble than eukaryotic hydrolases, archaea are somewhere in between. The relatively high predicted solubility of medium- sized fungal hydrolases is in contrast to the low predicted solubility of small and large fungal hydrolases. Hydrolases from plants are predicted to be much more soluble when they are small than when they are of medium size. However, the significance of this statement is relatively low because there are only eight medium- sized hydrolases from plants.
Though large hydrolases are mainly predicted to be insoluble the CV-CV\' values are much lower for metazoan than for bacterial and fungal hydrolases. So here the situation is opposite to the situation of small and medium- sized hydrolases. Additionally, large metazoan hydrolases are predicted to be more soluble in *E. coli*than small and medium- sized metazoan hydrolases.
These results propose that, in problematic cases where soluble hydrolases are very rare, it makes sense to screen for large hydrolases in coding sequences of metazoa, small hydrolases should be searched in bacterial or archaean and medium- sized hydrolases in actinobacterial or ascomycetic coding sequences. Secreted and eukaryotic hydrolases generally have a low predicted solubility in *E. coli*.
As the data about taxa, the family information could be used to efficiently search for new soluble hydrolases. If a specific catalytic activity is observed in different homologous families or a soluble member of a specific superfamily is searched for a structural genomics project, family specific degenerate primers could preferably be designed for families which are predicted to include mainly soluble hydrolases.
If soluble hydrolases of a given size are rarely found, screening of libraries could be limited to coding sequences from taxa where soluble hydrolases are expected.
Conclusion
==========
General rules for the relationship between predicted solubility in *E. coli*, protein size and phylogenetic origin are:
1\) Bacterial hydrolases are predicted to be more soluble in *E. coli*than eukaryotic hydrolases.
2\) Small hydrolases are predicted to be more soluble in *E. coli*than large hydrolases.
3\) In one taxon huge differences of predicted solubility between genera can exist.
4\) In one superfamily huge differences of predicted solubility between homologous families can exist.
When taking into account the groups formed by protein size there are three additional rules:
5\) Small bacterial hydrolases are predicted to be more soluble than small and large hydrolases from eukaryotes and large bacterial hydrolases.
6\) Large metazoan hydrolases are predicted to be more soluble than large hydrolases from bacteria and small hydrolases from metazoa.
7\) Fungal medium- sized hydrolases are predicted to be more soluble than small or large hydrolases from fungi.
When using the family characteristics family- specific primers could be designed to amplify members from specific homologous families with high averages of predicted solubility in *E. coli*.
Methods
=======
The database
------------
Sequence data for analysis was derived from the Lipase Engineering Database (LED) \[[@B31],[@B32]\] which integrates sequence, structure and annotation information of α/β-hydrolases. The database comprises 1820 hydrolases which are grouped into 149 homologous families and 45 superfamilies.
To remove fragments, only hydrolases larger than 170 amino acids were included in the analysis.
The solubility model
--------------------
A two parameter statistical model by Harrison et al. \[[@B1],[@B3]\] was used to predict *in vivo*solubility of recombinant proteins in *E. coli*. The main parameters for solubility in *E. coli*are the relative number of turn forming residues (asparagine, glycine, proline and serine) and the absolute of charge per residue which is determined by the fraction of positively and negatively charged amino acids (arginine, lysine, aspartic acid, glutamic acid). These values have been combined to a canonical variable CV, where N, G, P, S, R, K, D, E, are the numbers of asparagines, glycines, prolines, serines, arginines, lysines, aspartic acids and glutamic acids, respectively, and n is the total number of residues in the sequence.
To distinguish soluble from insoluble protein a threshold of CV\' = 1.71 was introduced. If the difference CV-CV\' is smaller than zero, the protein is predicted to be soluble in *E. coli*. If it is larger than zero the protein is predicted to be insoluble in *E. coli*. From CV-CV\' a probability of solubility is calculated:
*P*= 0.4934 + 0.276\|*CV - CV*\'\| - 0.0392(*CV - CV*\')^2^
The higher the absolute of CV-CV\', the higher the probability of solubility (CV-CV\' \< 0) or insolubility (CV-CV\' \> 0). CV-CV\' values of -0.4, 0.0 and 1.1 indicate probabilities of solubility of 60 %, 50 % and 25 %, respectively.
The dependency of solubility on the parameters relative number of turn forming residues and absolute of charge per residue can be interpreted as follows: the higher the charge, the higher the repulsion between proteins, thus preventing aggregation. Additionally, many turn-forming residues slow down protein folding, resulting in a high concentration of folding intermediates which can form aggregates.
Statistical methods
-------------------
Averages of CV-CV\', the relative number of turn forming residues, charge per residue and length of protein sequence were computed. Additionally first quartiles of CV-CV\' were determined. For determination of first quartiles the values were ordered by size and divided into four groups with the same number of members. The first quartile is the largest value of CV-CV\' of the group which contains 25 % of the smallest values of CV-CV\'. This means that 25 % of all proteins in the distribution are predicted to have at least the solubility that is represented by the value of the first quartile. When used in combination with averages, quartiles give additional information about the shape of the distribution. They make it possible to compare distributions even if their averages are very similar. While the average value of CV-CV\' gives an upper limit of solubility to 50 % of all proteins of a distribution, the first quartile indicates the solubility of the 25 % best soluble proteins.
Visualisation of distributions
------------------------------
The optimal window size for the visualisation of the distributions d~1~, d~2~and d~3~(Figures [1](#F1){ref-type="fig"} and [2](#F2){ref-type="fig"}) was determined as follows. For each distribution a window size w was determined, where max and min are the largest and the smallest values of the distribution and n is the number of values in the distribution.
As an overall window size for the graphics the largest of the three window sizes was taken.
Abbreviations
=============
*E. coli*, *Escherichia coli*.
Authors\' contributions
=======================
MK wrote the Perl scripts, carried out the analysis and drafted the manuscript. MF provided the database; SB participated in the design of the study. JP supervised the study. All authors read and approved the final manuscript.
Supplementary Material
======================
::: {.caption}
###### Additional File 1
All tables which are too large to be included in the main article. The format is Microsoft Word 2000.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 2
All hydrolases in the analysis were grouped by homologous families and superfamilies. The format is UNIX plain text.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
We wish to thank Jürgen Dippon (Institut für Stochastik und Anwendungen, University of Stuttgart) for fruitful discussions. We are grateful for the suggestion by one of the referees to use PDB files as a positive control.
|
PubMed Central
|
2024-06-05T03:55:55.256184
|
2005-4-2
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079826/",
"journal": "BMC Genomics. 2005 Apr 2; 6:49",
"authors": [
{
"first": "Markus",
"last": "Koschorreck"
},
{
"first": "Markus",
"last": "Fischer"
},
{
"first": "Sandra",
"last": "Barth"
},
{
"first": "Jürgen",
"last": "Pleiss"
}
]
}
|
PMC1079827
|
Background
==========
Transcriptome analysis is a common way of discovering differences in gene expression because regulation of gene activity occurs primarily on transcription level. Numerous low-cost, simple methods are available for gene discovery projects, providing a limited set of transcripts more or less randomly selected from a pool of genes expressed differently between two samples (for example treated sample/control or diseased/healthy tissue). These methods include differential hybridization, subtractive hybridization, EST sequencing and other gene discovery methods, which do not provide quantitative data nor do they aim at a complete coverage of the transcriptome.
DNA microarrays became popular as a tool for genome-wide transcription analysis. Microarray analysis of gene expression is only feasible when extensive sequence information and/or cDNA libraries are available. Therefore, DNA microarrays do not satisfy a growing demand for \"open-end\" transcriptome analysis. Proprietary high-end transcript profiling systems such as Massive Parallel Signature Sequencing \[[@B1]\] and GeneCalling \[[@B2]\] are excellent tools for genome-wide quantitative transcription analysis on any organism, but they are available to only a small fraction of researchers. Serial Analysis of Gene Expression \[[@B3]\] is a free, open-end, quantitative transcription profiling system. The drawback of SAGE is that the size of the sequencing effort required to generate representative data sets limits the practical applicability of the method, particularly when transcripts expressed at low levels need to be covered.
mRNA differential display (DDRT-PCR) \[[@B4]\], cDNA-AFLP \[[@B5]\] and different variants of these two methods \[[@B6]\] are gel-based transcript profiling systems based on electrophoretic fingerprinting of amplified cDNA fragments. Both DDRT-PCR and cDNA-AFLP are open-end methods requiring only standard instrumentation and incurring low costs. They have been used in hundreds of published studies both on model and non-model organisms so far. The reproducibility of cDNA-AFLP patterns was reported to be superior to DDRT-PCR, but its drawback is a high fraction of cDNA molecules escaping detection because of the lack of suitable restriction sites \[[@B7]\]. Because of this limitation and a higher technical demand for cDNA-ALFP analysis as compared to DDRT-PCR, the latter method has largely dominated gel-based transcription profiling \[[@B6]\].
The fundamental goal of transcriptomics is to generate and compare snapshots of mRNA populations. In order to be suitable for comparisons with new data sets, these transcription profiles have to consist of quantitative values stored in a digital form and allowing for normalization procedures compensating for differences in the efficiency of mRNA extraction, cDNA synthesis and detection. Neither DDRT-PCR nor cDNA-AFLP in their established forms fulfill these requirements. So far both methods have been used for gene discovery based on subjective band selection \[[@B8]\], though different steps towards automatic large-scale transcription profiling have been undertaken, including the use of DNA sequencers for cDNA fragment analysis \[[@B9]-[@B12]\], linking expression data to sequences \[[@B13]\] and quantitative evaluation of autoradiographs of cDNA-AFLP gels \[[@B7]\].
Here we describe procedures for the conversion of fluorescent differential display results into quantitative transcription profiles, using a DNA sequencer for cDNA fragment separation and DNA fingerprinting software for band matching, data normalization and densitometry. Transcription profiles are represented by sets of intensity ratios calculated for band pairs (treated sample vs. control) defined by the primer combination used and DNA fragment mobility. We demonstrate the method by analyzing the effect of secondary metabolites of oilseed rape *Brassica napus*on the transcriptome of the rape pathogen *Leptosphaeria maculans*. To our knowledge this is the first report presenting a quantitative evaluation of a mRNA differential display experiment.
Results
=======
Frequency and size distribution of cDNA amplicons
-------------------------------------------------
DDRT-PCR partitions transcripts into non-overlapping sets by synthetising cDNA and amplifying their terminal parts with primer pairs consisting of an anchored poly(dT)-primer and a random primer. We used three anchored primers and 76 random primers (see Arbitrary\_primers.xls), resulting in 228 primer combinations. Amplification products were separated on a DNA sequencer and detector signals were converted into virtual images (Fig. [1](#F1){ref-type="fig"}). A total of about 10,000 distinct fragments of *Leptosphaeria maculans*cDNA were detected. 5,422 bands exceeded the thresholds set for the size (50 bp) and intensity (0.5% elevation as compared to the surrounding). The average number of detected bands produced by one primer pair were 73, the minimum were 0 and the maximum were 263 bands per primer pair. 65% of arbitrary primers produced 40 -- 180 bands each (Fig. [2](#F2){ref-type="fig"}).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Virtual gel image of DDRT-PCR products**. *Leptosphaeria maculans*cultures were treated with secondary metabolites of *Brassica napus*, RNA was extracted and DDRT-PCR reactions performed as described in Materials and Methods section. Amplicons labeled with Cy5 fluorophore were separated on a DNA sequencer and detector signals were converted into a virtual gel image. Thirteen primer combinations were used for the amplification of paired samples consisting of a control (left) and a treated culture (right) each. A size standard was loaded into the first lane.
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Number of bands produced by 76 arbitrary primers**. Arbitrary primers are sorted by the total number of bands
:::
:::
Frequencies of bands produced with different anchored primers are compared in Fig. [3](#F3){ref-type="fig"}. All three anchored primers produced a similar number of bands, indicating that there is no preference for the last non-T nucleotide in the polyadenylation site of mRNA in *Leptosphaeria maculans*.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Number of bands produced with anchored primers (T)~17~A, (T)~17~C, and (T)~17~G**. Box-and-Whisker-plots: dashed line = mean, solid line = median, dots = outliers
:::
:::
The size distribution of amplified fragments (Fig. [4](#F4){ref-type="fig"}) shows that short fragments were amplified preferably, which is typical of methods based on randomly primed PCR. Sequencing very short DDRT-PCR fragments is likely to detect 3\'-noncoding sequences, which are not informative. Fragments exceeding the length of 200 nt, which would preferable be chosen for cloning and sequencing, made up a half of all evaluated fragments.
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Size distribution of amplified cDNA fragments**. Bands corresponding to fragments shorter than 50 bp were discarded. The size of detected bands ranged from 50 to 847 bp.
:::
:::
Coverage and reproducibility
----------------------------
To accurately estimate the coverage of the DDRT-PCR experiment, one needs to know the number of genes expressed under culture conditions used. Because the genome of *L. maculans*has not been sequenced yet, we assume by analogy with other filamentous fungi that it harbours about 10 000 genes. Taking into consideration that many functions are not expressed in mycelial cultures used as the source of mRNA (e.g., activities related to mating, pathogenesis, sporogenesis and spore germination), we assume that the transcriptome analyzed in this work consisted of 4000 to 6000 mRNA molecules. DDRT-PCR generated 10000 cDNA bands, 5422 of which were selected for evaluation. Assuming Poisson distribution, this corresponds to a 59 -- 75% coverage of the transcriptome. The coverage of published DD and cDNA-AFLP experiments varies from less than 20% to 73%.
It has been reported that DDRT-PCR results are afflicted with a relatively high proportion of false positives \[[@B6]\], which originate mainly from ribosomal RNA. The reproducibility of quantitative changes in transcription levels determined by DDRT-PCR has not been reported so far. The fact that induction factors for the vast majority of data points calculated from normalized band intensities were close to 1.0 (Fig. [5](#F5){ref-type="fig"}) indicates that the reproducibility of DDRT-PCR patterns was fairly good in this experiment, though we did not know which part of the differences is attributable to induction/suppression and which to random variance. To determine the reproducibility of the quantitative scoring, we repeated a set of DDRT-PCR reactions on RNA extracted from a second untreated (control) culture. Both cultures were inoculated with the same inoculum and grown in the same incubator to keep the differences in culture conditions at minimum. RNA extraction, cDNA synthesis and DDRT-PCR were performed in parallel, too, and care was taken to limit the variance introduced by handling. Band intensities were normalized and the ratios of intensities for matching bands calculated. The results are summarized in Tab. 2. These data confirmed that reproducibility is an important issue in DDRT-PCR, as is the case with microarrays and other profiling methods. The variance in our experiment was comparable to microarrays \[cp. Table 3 in Ref. \[[@B14]\]\].
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Changes in intensity of DDRT-PCR bands from mRNA of *Leptosphaeria maculans*after treatment with *Brassica napus*extract**. Bands which passed thresholds of relative intensity and elevation from background (see Materials and Methods), sorted by the value of their induction factor (the ratio of normalized intensities of treated sample vs. control), are listed on the abscissa. Vertical lines drawn through the curve correspond to induction factor values of 3.00 and 0.33 (induction or repression by a factor of 3, respectively).
:::
:::
Changes in transcript levels after treatment of fungal cultures with phytoalexins
---------------------------------------------------------------------------------
In order to detect specific effects of *Brassica napus*phytoalexins on *Leptosphaeria maculans*transcriptome and exclude general stress responses, we chose very mild treatment conditions. Firstly, only a small fraction of *Brassica napus*metabolites was used for the treatment. In the purification protocol for these metabolites from plant material we combined two liquid-liquid extractions (from phosphate buffer at neutral pH into ethyl acetate and subsequently from ethyl acetate into hexane). Secondly, the concentration of metabolites in fungal culture medium was adjusted to be at most equal to their concentration in *Brassica napus*stem. Furthermore, a short treatment time was chosen to preclude secondary effects.
DNA yield for each evaluated cDNA fragment was determined as a densitometric value of the corresponding band, normalized to account for differences in loading volumes and used to calculate ratios between treated sample and control, designated as induction factor (see Materials and Methods for details). Fig. [5](#F5){ref-type="fig"} summarizes induction factor values for all 5,422 evaluated bands. The intensity of the majority of bands has changed to a lesser degree than three-times after the treatment. The induction factor was used to divide evaluated bands into five classes. As shown in Fig. [6](#F6){ref-type="fig"}, intensities of 174 bands (0.03 %) increased 3-times to 9-times upon treatment and 16 bands (0.003 %) were induced more than 9-times, the strongest induction observed being 19.5-times. In a gene discovery project these bands would represent candidates for cloning.
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**Classification of cDNA bands in induction classes**. Evaluated cDNA bands were assigned to five classes according to the ratios of intensities sample/control.
:::
:::
In addition to bands appearing in both treated sample and control, a few bands were detected only in the control or only in the treated sample, representing transcripts which were either not present in detectable amounts in the control or fully suppressed by the treatment. These bands cannot be characterized by an induction factor, which amounts to infinity for induced bands without a counterpart in the control and zero for bands fully suppressed in the treated sample. In order to further assess the significance of these bands, we compared their intensities with the intensities of all evaluated bands in the experiment (Table [1](#T1){ref-type="table"}). The strongest band (present only in the treated sample) amounted to 2451% of the mean peak intensity calculated for all bands in the experiment. The strongest suppressed band (present only in the control) amounted to 951% of the mean peak intensity.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Solitary cDNA fragments
:::
Bands detected in treated sample but missing in the control
------------------------------------------------------------- -----------------
Relative intensity\* Number of bands
0 -- 30% 3
30% -- 300% 19
\> 300% 10
Bands detected in the control but missing in treated sample
Relative intensity\* Number of bands
0 -- 30% 3
30% -- 300% 13
\> 300% 6
\*as compared to the average of densitometric values of all bands in the experiment DDRT-PCR bands in *L. maculans*cultures treated with *Brassica napus*secondary metabolites vs. untreated control were matched and their intensities compared (Fig. 5 and 6). For bands with no counterpart, the relative intensity as compared to the average intensity of all bands in the experiment was determined.
:::
Discussion
==========
The main drawback of cDNA differential display as compared to DNA microarrays and industrial gel-based transcription profiling systems is a subjective evaluation of gels and the fact that data are not available in a quantitative form suitable for database storage and numerical manipulations. Though different steps towards quantitative evaluation of DDRT-PCR gels have been undertaken, a practicable system for DDRT-PCR-based transcriptomics based on common equipment has not been described yet.
The system for quantitative evaluation of gel-based transcript profiles developed in this work relies on widely available equipment (DNA sequencer) and software (DNA fingerprinting analysis and spreadsheets). A crucial part of data processing is normalization, which compensates for differences in pipetting and in the efficiency of different steps from RNA extraction up to loading samples on the sequencer. A central issue in normalization is the choice of bands originated from transcripts unchanged by the treatment. Induced and suppressed bands have to be excluded from the calculation of a normalization factor, yet accurate distinction between induced/suppressed and constitutive bands is only possible with normalized data. We solved this problem by assuming that the majority of bands in each lane are not affected by the treatment and calculating a normalization factor from the middle quartiles of bands sorted by their uncorrected induction factor UIF (see Materials and Methods for details.) The efficiency of this approach was proven by the fact that the vast majority of induction factors calculated from normalized data were close to 1.0 (see Fig. [5](#F5){ref-type="fig"}). This result also demonstrated an excellent reproducibility of differential display patterns recorded on a DNA sequencer.
cDNA fragment patterns were originally recorded by autoradiography of radioactively labeled DNA separated in polyacrylamide gels \[[@B4]\] and radioactive labeling is still being used \[[@B7]\]. However, irregular background and typical artifacts of autoradiographs hampers automatic gel evalution. Bauer et al. \[[@B20]\] attempted to replace autoradiography by labeling cDNA fragments with fluorophores and separating them on a DNA sequencer. They used different fluorophores for controls and treated samples; because of the differences in fluorescence yield among fluorophores, quantification of the amount of DNA in cDNA bands was not possible. Therefore, the authors depicted their protocol as merely \"a qualitative method allowing to identify genes which are completely switched on or off\".
In further development of fluorescent DDRT-PCR, labeling random primers \[[@B12]\] was replaced by labeling anchored primers and capillary DNA sequencers were used for cDNA fragment detection \[[@B9],[@B10],[@B23]\]. Although the use of Genotyper Genetic Analysis software (Perking-Elmer ABI) facilitated semi-automatic recognition of unique bands (e.g., bands present in the treated sample but missing in the control), quantitative evaluation of data has not been demonstrated.
Aittokallio et al. \[[@B21],[@B24]\] developed mathematical algorithms for the evaluation of DDRT-PCR patterns recorded on an ALFExpress II sequencer (the same system used in this work) as an alternative to transcript profiling by microarrays. Their algorithms were written from scratch, comprising functions for fitting and smoothing densitometry traces, automatic band finding, band matching (called alignment) and eventually identifying differently expressed bands. Their work provides a solid basis for the development of software for automatic, quantitative processing of DDRT-PCR results, but it is not mature yet. The software is not freely available and it is not clear from the publication whether it is user-friendly enough to be used outside the laboratory of its creators. Finally, certain aspects of the signal processing procedure needs scrutiny. For example, the difference in mobility between corresponding bands (treatment/control) affects the value of Matushita distance (equation 5 in \[[@B21]\]) which is used for the computation of the dissimilarity score (equation 8 in \[[@B21]\]), though these differences reflect merely physical conditions during gel electrophoresis and are unrelated to transcription. In our procedure we completely separated band matching from the comparison of band intensities. Another potential problem with signal processing in \[[@B21]\] is the compensation for differences in loading volumes between lanes. Instead of normalizing lane intensities, the authors attempted to achieve a \"similar background\" by dividing the \"distance value\" for each peak pair by the average distance of all other peak pairs in the lane. This procedure is likely to produce erroneous results when two or more bands in a lane originate from transcripts of differentially expressed genes. Furthermore, it will introduce a bias into distance values of constitutively expressed (unchanged) transcripts in lanes which contain strongly induced or suppressed bands. We avoided this problem by excluding 25% bands with the largest and 25% bands with the smallest induction factor from the computation of the normalization factor (see Materials and Methods for details).
Gellatly et al. \[[@B22]\] recently studied the effect of methylbenzyldithiocarbamate (analog of phytoalexin brassinin) on *Leptosphaeria maculans*. They used DDRT-PCR and observed on the average 65 bands for each primer combination, which is similar to our results. Because the majority of gel-excised bands could not be amplified, the authors later abandonded DDRT-PCR and swiched to cDNA-AFLP. We successfully amplified and cloned bands selected from DDRT-PCR gels for *Leptosphaeria maculans*by visualizing Cy5-labeled DNA on a fluorescence scanner (to be published). There are two differences in the treatment conditions used by Gellatly at al. \[[@B22]\] as compared to our work. The first difference concerns the composition of the inducing agent: a defined brassinin analogue was used by Gellatly \[[@B22]\] while we extracted metabolites from *Brassica napus*stem. Secondly, the treatment time was different: Gellatly at al. \[[@B22]\] induced *L. maculans*cultures for 22 h while we harvested mycelium just 5 h after the induction to prevent secondary effects. It will be interesting to compare genes identified as induced by the treatment in these two experiments.
The group of Breyne et al. \[[@B7]\] recently published an attempt to further develop cDNA-AFLP technique into a quantitative genome-wide transcript profiling system. The major drawback of their protocol as compared to DDRT-PCR is a high fraction of transcripts which escape analysis because of the lack of suitable restriction sites (40% in the work by Breyne at al. \[[@B7]\]). This limitation casts doubts on the suitability of the system for genome-wide expression analysis. Furthermore, the detection and quantification of cDNA fragments was done by autoradiography of radioactively labeled DNA rather than on a DNA sequencer. The noise inherent to autoradiography is likely to increase statistical errors in intensity values assigned to bands. A crucial difference as compared to our approach concerns the normalization of band intensities. The authors calculated their normalization factor from the intensities of bands found to be expressed constantly during a time-course experiment while we automatized this process by calculating intensity ratios for all bands, sorting the bands by ratios and using bands from the 25% -- 75% quartile for the calculation of a normalization factor. Our procedure is suitable for automatization and does not require data from time-course experiments, which are not available in the majority of DDRT-PCR applications.
Conclusion
==========
Our protocol extends the application of cDNA Differential Display (DDRT-PCR) from a gene discovery tool to quantitative transcript profiling. The system is based on a combination of widely accessible hardware (DNA sequencer and fluorescence scanner) and software (DNA fingerprinting analysis software and a spreadsheet program). Transcript profiles generated by quantitative DDRT-PCR according to our protocol consist of sets of numerical values (primer pair/fragment size vs. induction factor) which can be stored in a spreadsheet and easily compared with each other and extended by new experiments. DDRT-PCR thus offers an open end alternative for microarray-based transcriptomics, which is particularly attractive in research on non-model organisms where microarray analysis is not available or economical.
Methods
=======
Cultivation of oilseed rape and phytoalexin elicitation
-------------------------------------------------------
Seeds of the summer oilseed rape cultivar Lichosmos (Deutsche Saatveredelung, Lippstadt, Germany) were germinated in containers and transplanted after 7 days into 13 cm^2^pots. A mixture of steamed compost, sand and peat (3:1:1) was used as substrate. Plants were grown under semi-controlled conditions in the greenhouse (16 h photoperiod, 23°C). Leaves of the plants were removed at growth stage BBCH 61 (beginning of flowering) after about 9--10 weeks of cultivation. Stems were directly treated with a solution of 10 mM CuCl~2~until run-off point using a commercial garden sprayer. Plants were then transferred into a folia tunnel to keep high humidity. The treatment was repeated after 24 h. Finally, plants were harvested 48 h after the last treatment, cut into segments of about 5 cm and stored at -20°C.
Extraction of plant secondary metabolites
-----------------------------------------
Hundred grams of *Brassica napus*stems were homogenised with 200 ml of 0.1 M sodium phosphate buffer pH 7.2 in a vortex. The homogenate was filtered through paper filter and the cleared homogenate was extracted twice with an equal volume of ethyl acetate. Ethyl acetate fractions were combined and extracted with 1 volume of hexane. Hexane was removed on a rotary evaporator and the residue suspended in 30 ml of sterile Czapek-Dox medium (Oxoid Ltd, Hampshire, England) and stored at -80°C.
Fungal cultures
---------------
Cultures of *Leptosphaeria maculans*T12aD34 \[[@B15]\] were started by inoculating 1 × 10^7^pycnidiospores into 35 ml of Czapek-Dox liquid medium (Oxoid Ltd, Hampshire, England) supplemented with 0.2% yeast extract (Difco Laboratories, Detroit, USA) in 300 ml Erlenmeyer flasks and incubated at 20°C in the dark without shaking.
Treatment of fungal cultures and RNA isolation
----------------------------------------------
10 ml of plant crude extract were added to 35 ml of 7-days old *Leptosphaeria maculans*cultures. 10 ml of sterile Czapek-Dox medium was added to control cultures. The incubation was continued for 5 h, after which the mycelium was separated from medium by filtration and frozen in liquid nitrogen. Total RNA was prepared using a guanidine thiocyanate/phenol method \[[@B16]\] modified as follows. Approximately 100 mg of mycelium was ground in a mortar under liquid nitrogen, transferred to a microcentrifuge tube containing 1 ml of GTC-buffer \[[@B16]\], vortexed for 10 s and incubated for 5--10 min at ambient temperature. After centrifugation for 5 min at 16,000 × *g*, the supernatant was transferred to a centrifuge tube containing 0.7 ml of phenol/chloroform/isoamylalcohol mixture (25:24:1), vortexed for 15 s and incubated on ice for 10 min. The emulsion was centrifuged for 10 min at 16,000 × *g*at 4°C. The aqueous phase was transferred to a new tube and mixed with 0.2 vol of 1 M acetic acid and 0.7 vol of 100% EtOH. The suspension was incubated at -20°C for 30 min. Precipitated RNA was centrifuged for 10 min at 16,000 × *g*, 4°C. The pellet was washed with 70% ethanol twice, dried and homogenized with 0.7 ml of GTC buffer without mercaptoethanol. Undissolved residue was removed by centrifugation and RNA was precipitated with 0.14 ml of 3 M sodium acetate, pH 5.4 and 0.7 ml ethanol at -20°C for 30 min. The RNA pellet was collected by centrifugation, washed twice with 70% ethanol, dried and dissolved in 30 μl of DEPC-treated water.
RNA samples were treated with RNase-free DNase (Fermentas, St. Leon-Rot, Germany) to remove the residual DNA contamination as recommended by the manufacturer. The integrity and quantity of RNA was assessed on a 1%denaturing agarose gel \[[@B17]\].
First strand cDNA synthesis
---------------------------
DNase treated total RNA was used for three cDNA synthesis reactions primed with Cy5-labeled anchored oligo-dT (dT~17~M; M = dA, dG or dC). cDNA synthesis mixture was prepared by combining 5.5 μl DEPC-treated water, 4 μl of 5x reverse transcriptase buffer (Fermentas, St. Leon-Rot, Germany), 2 μl of dNTP (10 mM), 5 μl of DNase treated RNA (about 2 μg) and 2.5 μl of 40 μM Cy5-labeled anchored poly(dT)-primers. After incubation at 65°C for 5 min followed by 10 min at 37°C 200 U of reverse transcriptase (Maloney murine leukemia virus (MMLV)) (Fermentas, St. Leon-Rot, Germany) were added and the incubation was continued for 50 min. The reaction was arrested at 72°C for 5 min. The products were diluted by adding 80 μl of distilled water to each reaction.
Polymerase chain reaction and gel electrophoresis
-------------------------------------------------
Second strand synthesis and PCR amplification were peformed in a total volume of 20 μl. The reaction mixture contained 20 μM dNTPs, 1 U of Taq Polymerase (GeneCraft, Muenster, Germany), 4 μM of arbitrary decamer deoxynucleotides (Operon, Cologne, Germany; for primer sequences see supplementary material), 0.5 μM Cy5-labeled anchored primer (dT~17~M) and 1 μl of diluted cDNA solution. PCR reactions with cDNA from the treated culture and control and each of 228 primer combinations (76 random primers vs. 3 anchored poly(dT) primers) were performed according to a published protocol \[[@B12]\] using an annealing temperature of 40°C and 26 cycles.
The whole PCR reaction was reduced to a suitable volume by incubating a PCR plate without a lid in a thermocylcer at 50°C for 1 h, mixed with 5 μl of loading solution (98% formamide, 10 mM EDTA, 0.025% bromophenol blue, 0.025% xylene cyanol FF, pH 8.0) \[[@B18]\] and denatured at 90°C for 3 min. A Cy5-labeled size standard from 50 to 500 nt (ALFexpress TM Sizer, Amersham Biosciences, Piscataway, USA) was used as size standard. Electrophoresis was performed on an ALFExpress II sequencer (Amersham Pharmacia Biotech AB, Uppsala, Sweden) with a 25 cm × 0.5 mm gel cassette and 6% polyacrylamide gel (Reprogel TM Long Read, Amersham Pharmacia Biotech AB, Uppsala, Sweden) at 25 W for 700 min at 55°C.
Data processing
---------------
Detector signals from an ALFExpress II sequencer were converted into 16-bit TIFF images using ALF Sequence Analyzer 2.11 (Amersham Pharmacia Biotech AB, Uppsala, Sweden). Each chromatogram was exported using a low compression level (resampling factor of 2 was used for digital processing). Pseudogel images for digital processing were cropped to a length of 10,000 pixels by removing the lower part of the gel using Adobe Photoshop 6.0 (Adobe Systems Incorporated, San Jose, USA) and imported into the gel analysis software GelCompar 3.5 (Applied Maths, Kortrijk, Belgium). 10,000 pixel is the maximum gel length which can be processed by GelCompar. Gel image parts corresponding to fragments shorter than 50 bp were removed and lanes with artefacts were identified by manual inspection and discarded. A Cy5-labeled size standard (50-bp ladder) was defined as a standard marker lane on the first gel and the following gels were aligned to this marker.
Band finding was performed automatically using a minimum profiling value of 0.5%, allowing only for bands with an elevation of at least 0.5% with respect to the surrounding background. Minimum area and shoulder sensitivity were set to zero. Further editing was performed manually for those band positions in which a band was found only in one of the two corresponding lanes (treatment/control). Bands were added or discarded according to the following criteria: when the densitometric curve revealed an elevation at the position in which a band was found in the corresponding lane, a new band was introduced and verified by checking whether its shape fitted to the densitometric profile. In case no elevation of the densitometry profile was found in a position in which a band was found in the corresponding lane and the corresponding band was weak (elevation barely over 0.5%), the band was discarded. Cases of strong bands in one lane corresponding to no band in the same position in the corresponding lane were very rare. Band matching was performed automatically for each two lanes corresponding to the same primer combination with position tolerance set to 1%. It was rarely necessary to adjust the matching manually, for example when several adjacent bands largely overlapped or when the algorithm failed to find obvious matches because of a slight retardation of lanes at the margin of the gel. Absolute peak area and calculated fragment size for each band were exported to Microsoft Excel (Microsoft, Redmond, USA).
Normalization of densitometry values is crucial both for comparisons among profiles generated in the same experiment and among different experiments. Because the intensities of control and treatment lanes might differ due to differences in the efficiency of RNA extraction and RT-PCR and in the volumes loaded on the sequencer, band intensities (measured as peak areas) have to be adjusted by a normalization factor. This factor can be obtained as the mean of ratios of intensities of corresponding bands (treatment/control) for bands originating from transcripts that are unchanged by the treatment. A technical problem arises from the fact that a rigorous distinction between bands which are affected by the treatment and those which are not affected is only possible after the normalization, yet it is necessary to identify at least some unaffected bands for the calculation of the normalization factor. We solved the problem by assuming that only a minority of bands in each lane is affected by the treatment.
To identify bands which were not affected by the treatment for the estimation of the normalization factor, we calculated \"uncorrected induction factors\" (UIF) for each pair of bands in corresponding lanes as the ratio of areas of peaks corresponding to matched bands. In the next step UIFs for all bands in a lane were sorted by values. Assuming that fewer than 25% of bands in a lane were induced and fewer than 25% suppressed by the treatment, the mean for all UIFs for bands between the first and third quartile was calculated. This value was used as a normalization factor to correct all absolute peak areas. Induction factors (IF) were now calculated as the ratio of normalized peak areas for matched bands in the treated sample and the control. Unaffected bands possessed induction factors close to 1.0, induced bands had IF \> 1.0, suppressed bands had IF \< 1.0.
Statistics
----------
Data were analysed using Microsoft Excel 7 (Microsoft, Redmond, USA) and graphs drafted with SigmaPlot 5.0 (SPSS, Chicago, USA). For results presented as Box-and-Whisker plots, the boxes include 50% of the ranked data, the whiskers show the 10^th^and 90^th^percentile, the outlier points mark the outliers defined as values above and below the 10^th^and 90^th^percentile. Means are expressed as arithmetic mean ± S.D.
Abbreviations
=============
cDNA -- DNA complementary to RNA, generated by reverse transcription of mRNA; cDNA-AFLP -- application of DNA-fingerprinting method Amplified Fragment Length Polymorphism to cDNA; Cy5 -- fluorophore used for DNA labeling (dimeric unsymmetrical cyanine dye); DDRT-PCR -- differential Display of cDNA, transcription profiling method based on the amplification of cDNA by a combination of a random primer and an anchored poly(dT)-primer; DEPC -- diethylbiscarbonate; Dnase -- desoxyribonuclase; dNTP -- desoxyribonucleotide triphosphates; EDTA -- ethylenediamminetetraacetic acid; GTC -- guanidinium thiocyanate; Rnase -- ribonuclease; IF -- corrected induction factor; S.D. -- standard deviation; UIF -- uncorrected induction factor
Authors\' contributions
=======================
BV performed all experiments except the induction of Brassica by copper. UH contributed to the design of algorithms for data processing, processed data and participated in drafting the manuscript. BK performed the induction of Brassica by copper and participated in data analysis and drafting the manuscript. PK conceived the idea, designed and coordinated the study, guided data processing and wrote major parts of the manuscript.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Reproducibility of normalized band intensities
:::
Intensity ratio Proportion
----------------- ------------
0.33 -- 3.0 97.0%
0.40 -- 2.5 88.9%
0.50 -- 2.0 80.7%
Peak areas of corresponding cDNA fragments derived from two independent control experiments generating 423 fragment pairs were normalized and compared to each other. Bands were assigned to intensity ratio classes defined in the left column.
:::
Supplementary Material
======================
::: {.caption}
###### Additional File 1
**Sequences of arbitrary primers**Excel-file containing the sequences of 80 primers used in combination with anchored poly(dT)-primers in DDRT-PCR reactions. 19 KB; <http://www.biomedcentral.com/imedia/6962194924713244/sup1.xls>
:::
::: {.caption}
######
Click here for file
:::
|
PubMed Central
|
2024-06-05T03:55:55.259332
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079827/",
"journal": "BMC Genomics. 2005 Apr 5; 6:51",
"authors": [
{
"first": "Balakrishnan",
"last": "Venkatesh"
},
{
"first": "Ursula",
"last": "Hettwer"
},
{
"first": "Birger",
"last": "Koopmann"
},
{
"first": "Petr",
"last": "Karlovsky"
}
]
}
|
PMC1079828
|
Background
==========
The success of linkage mapping in Mendelian traits led to great optimism that the same approach could be harnessed to identify genes in disorders or traits without clear Mendelian inheritance patterns, so called complex or quantitative traits. In contrast to phenotypes that are controlled by single genes, in complex traits there is no clear-cut mode of transmission. It is likely that quantitative traits are influenced by many genes of small effect size (called quantitative trait loci, QTLs), with factors such as gene-gene interactions (epistasis), pleiotropy, and gene-environment interactions complicating matters further \[[@B1],[@B2]\].
Although many linkage screens for complex disorders have been reported, linkage studies are underpowered to detect genes of small effect size \[[@B3]\]. In contrast, case-control association analysis, even using stringent significance levels, promises to provide the power required to detect QTLs that are too weak to be detected by linkage analysis alone \[[@B2]\]. However, this power comes at an expense, as a systematic genome wide association study requires very large numbers of DNA markers, perhaps as many as 500,000 \[[@B4],[@B5]\]. Furthermore, the ability to detect QTLs of small effect size requires large samples. For example, 80% power (*p*\< .01, two tailed) to detect an effect of 0.5% in an unselected sample requires samples of at least 1000 individuals \[[@B6]\].
Consequently, considerable effort has gone into developing high-throughput genotyping methodologies that allow the genotyping of dense marker sets in large sample sizes quickly, accurately, with minimal optimisation and very low unit cost \[[@B7]\]. Until such technologies become widely available, one way to address the cost, time and labour involved in using large sample sizes is to perform analyses not on individual DNA samples, but on pools made up of DNA from multiple individuals for cases and for controls, a technique that can dramatically reduce the genotyping burden \[[@B8]\]. There is a growing literature addressing methodological issues such as DNA pool construction, genotyping assays, and statistical analysis \[[@B9]-[@B12]\], and the strengths and weaknesses of DNA pooling have recently been reviewed \[[@B13]\]. DNA pooling has been used successfully to identify replicated associations with complex traits \[[@B14]-[@B16]\]. However, using the current SNP genotyping methodologies on DNA pools even for a small number of DNA pools for a dense marker set is still labour-intensive and expensive. One solution to the problem of genotyping many DNA markers is SNP genotyping microarrays which use a one-primer assay to genotype thousands of SNPs, offering the first real hope of a systematic survey of DNA variation in the human genome. However, microarrays can be used only once and are expensive in studies consisting of the large samples needed to detect QTLs of small effect size. One solution to the QTL conundrum is to combine both DNA pooling and SNP microarrays, an approach that we call SNP-MaP (SNP *m*icroarrays *a*nd *p*ooling), which can dramatically reduce the cost of screening large numbers of SNPs on large samples.
We hypothesised that quantitative estimates of allele frequencies -- especially the relative allele frequencies comparing groups like cases and controls -- can be derived from pooled DNA using SNP genotyping microarrays, similar to the way that expression microarrays estimate quantitative frequencies for mRNA transcripts \[[@B17]\].
Affymetrix software (GDAS) uses the hybridisation fluorescence signals from the SNP microarrays to generate \'Relative Allele Signals\' (RAS), a ratio of the measurement of allele A to the summed measurement of alleles A and B. Thus, RAS values vary between 0.0 and 1.0. Two independent RAS values are derived for each SNP from the sense strand (RAS1) and the anti-sense strand (RAS2). As explained in the Methods, Affymetrix software plots RAS1 scores against RAS2 scores and uses empirically derived clustering information to trichotomise these RAS scores as genotypes for DNA of an individual. RAS values near 0.0 are identified as a BB homozygote, 0.5 as an AB heterozygote, and 1.0 as an AA homozygote. We propose that RAS values can be used as quantitative indexes of allele frequencies in DNA pools.
The purpose of our current report is to follow up our previous study that addressed the feasibility of SNP-MaP \[[@B18]\]. We explore the reliability validity and sensitivity of SNP-MaP in greater detail using Affymetrix GeneChip^®^Mapping 10 K Array Xba 131 which genotypes more than 10,000 SNPs.
We constructed a control DNA pool consisting of 100 individuals independently three times (control pool A, B, and C), each assayed on triplicate microarrays. We used these replicate control pools to assess the reliability of estimating allele frequencies from pooled DNA. To assess validity and sensitivity, we compared allele frequency estimates from microarray assays using pooled DNA to individual genotyping. In addition, in order to assess sensitivity experimentally, we reconstructed two \'case\' DNA pools that differed by 15% and 20% in allele frequencies from the controls by spiking an aliquot of a control pool with an individual\'s DNA who was also individually genotyped on the microarray. Each case pool was assayed on duplicate microarrays.
Results
=======
Validation of pool construction
-------------------------------
Before using the microarrays, we genotyped the 9 DNA pools for 3 SNPs using SNaPshot™ ddNTP primer extension reaction kit (Applied Biosystems^©^) in order to confirm that the control DNA pools A, B and C were representative of the 100 individuals used to construct the pools. The results indicated that pool construction was valid in that, for the three SNPs, the pools yielded similar allele frequency estimates. Most importantly, these estimates, and especially their averages across the pools, are highly similar to the allele frequencies based on individual genotyping. For example, the average pooled allele frequencies (corrected for *k*see below and Methods) and individual genotyping allele frequencies, respectively, were 0.914 and 0.918 for rs1003063, 0.438 and 0.464 for rs15643193, and 0.639 and 0.634 for rs956122. SNaPshot™ has previously been shown to estimate pool allele frequencies with a high degree of accuracy if *k*is applied \[[@B12]\].
Summary of 10 K GeneChip results for pooled DNA
-----------------------------------------------
We consistently obtained good signal detection rates across all 14 microarrays used in the study, with an average signal detection of 83.9% (range of 60.54% to 98.59%) using the default analysis parameters. SNPs were excluded if there was inadequate discrimination between specific versus non-specific hybridisation of DNA to the probes.
We obtained RAS1 and RAS2 values for 6077 SNPs across all 9 microarrays for the control pools A, B and C. For the spiked pools, RAS1 and RAS2 values were obtained for 9908 SNPs for the 15% spiked pool and 6668 SNPs for the 20% spiked pools.
Validation: comparing allele frequencies from pooled DNA and population estimates
---------------------------------------------------------------------------------
We have previously shown that an average of RAS1 and RAS2 values (RAS~av~) estimate pooled allele frequencies more accurately then RAS1 or RAS2 alone \[[@B18]\].
RAS~av~values were calculated for each SNP on the 3 microarrays for each of the 3 DNA control pools. These RAS~av~values were then averaged across the 9 microarrays. The resulting allelic frequency estimates from pooled DNA were correlated with SNP allele frequencies determined by individually genotyping for an independent Caucasian sample. The latter data are publicly available from the NetAffx™ Analysis Centre <https://www.affymetrix.com/analysis/netaffx/index.affx>, a web-based tool providing extensive annotation for each SNP on the 10 k microarray derived from Affymetrix internal validation studies.
Figure [1a](#F1){ref-type="fig"} shows a scatterplot between the RAS~av~estimates of allele frequencies for pooled DNA versus the allele frequency estimates from individual genotyping for 11,533 SNPs. The correlation between the pooled DNA and individual genotyping estimates is 0.901, indicating that RAS values can be used to provide a valid quantitative measure of allele frequency in pooled DNA. Nonetheless, the absolute mean difference between microarray estimates of pooled allele frequencies and NetAffx™ population allele frequencies was 0.094 but varies widely (Min 0.00, Max 0.619).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**a**: RAS~av~values derived from between 1--9 replicate assays of the control pool DNA, compared to published population allele frequency estimates for the entire array. *N*= 11,533. **1b**: *k*-corrected RAS~av~values derived from between 1--9 replicate assays of the control pool, compared to published population allele frequency estimates for the entire array (*N*= 11,533).
:::
:::
In Figure [1a](#F1){ref-type="fig"}, the allele frequency estimates from pooled DNA were not corrected for unequal amplification of the two alleles that can occur with pooled DNA. That is, when two alleles are present in equal amounts (as in heterozygous individuals), the allele measurements obtained should be of equal size or intensity. However, for to various reasons \[[@B19]\] equal representation of alleles may not occur.
This bias is corrected by applying a factor, *k*\[[@B20]\], where *k*= A/B, using the equation: A = A/(A+*k*B), where A and B are the measurements of the A and B alleles. *k*corrections for the SNPs on the Affymetrix 10 K microarray were available from a panel of 33 Caucasian individuals assayed on separate microarrays and applied to each SNP (see Methods).
The range of *k*values is very tight, with only a handful of outliers considering the enormous number of SNPs under investigation. 6,813 of 10,084 SNPs have *k*values ranging from 0.5 to 1.5 (we would expect *k*to be 1 if both alleles were being measured equally). The averaged *k*over 10,088 SNPs is 1.14, median = 0.94, max = 93, min = 0.0. It should be noted that in relation to a case-control study design, unless *k*is extreme and allele frequencies rare, *k*correction would not significantly impact on *relative*allele frequency differences between cases and controls, although *absolute*allele frequency estimates would alter. In other words, although *k*correction is needed to compare allelic frequency estimates from pooled DNA and individual genotyping, it is not needed for comparisons between cases and controls.
Figure [1b](#F1){ref-type="fig"} shows the tighter scatter around the line of best fit when the pooled estimates are *k-*corrected. The correlation between the pooled and individual genotyping estimates increased from 0.901 to 0.953. The mean difference between *k*-corrected microarray estimates and NetAffx™ is also attenuated from .094 to .064.
Validation: comparing allele frequencies for pooled DNA and individual genotyping
---------------------------------------------------------------------------------
In addition to using the NetAffx™ population data, we individually genotyped the 100 individuals used to construct the control pool for 104 SNPs. The results are shown in Figure [2](#F2){ref-type="fig"}. As expected, the 104 SNP allele frequencies generated by the micorarrays are highly correlated (0.924) with the NetAffx™ population data, and correlate even better with the individual genotyping data for individuals in the pool (0.942). The mean difference between microarray estimates and individual genotyping was .077. When *k*correction was applied, the correlation between *k-*corrected microarray estimates and individual genotyping increased to 0.986, and the mean difference between *k-*corrected microarray estimates was attenuated from .077 to .036.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
*k*-corrected and uncorrected RAS~av~values derived from nine replicate control pool assays compared to individual genotyping data for 104 SNPs.
:::
:::
Estimation of experimental errors
---------------------------------
The previous analyses are based on RAS~av~values averaged across the 9 microarrays. In order to assess the technical reliability of deriving allele frequency estimates from the DNA pools, we compared RAS~av~for the three technical replicates within each DNA pool and between the three DNA pools. (See Figure [3](#F3){ref-type="fig"}.) As expected given the high correlation between pooled estimates and individual genotyping, correlations of RAS~av~values within pool triplicates are very high (0.945 to 0.968), as are correlations between pools (0.94 to 0.982). The dispersion of the RAS~av~values cross the nine control microarrays is encouragingly narrow: all but 19 SEMs are under 0.1 and 8,695 are 0.025 or less.
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Distribution of standard deviation and standard error in RAS~av~across all nine replicate control pool assays.
:::
:::
For those SNPs where we have RAS values across all 9 pools (*n*= 6077), the median standard deviation is .042. Using this figure to estimate the power of a hypothetical experiment where we have the same number of case pools, for a .05 significance level we would have 80% power to detect an allelic frequency difference of 0.05, and virtual certainty of detecting an allelic frequency difference of 0.10.
By fitting a random intercept model for each SNP to the 9 RAS~av~values, it is possible to estimate the variance components attributable to pool construction (τ) and the measurement error (σ). σ and τ are assumed to be independently normally distributed. The distribution of  is quite similar to the overall standard deviation (see figure [3](#F3){ref-type="fig"}) because tau is, as expected, quite small, ranging from 9 × 10^-7^to 2 × 10^-4^, with a median of 3 × 10^-5^. An estimate of 1 × 10^-4^has been made from comparable pools using different methods \[[@B21]\].
Sensitivity
-----------
The sensitivity of SNP-MaP was assessed empirically in a spiking experiment in which 15% and 20% of one individual\'s DNA was added to control pool D to create two new pools analogous to \'case\' pools. Each of the spiked pools was assessed on duplicate microarrays. The individual\'s DNA was also genotyped on a microarray in order to calculate the allelic frequencies of the spiked pools.
Based on the power calculations presented above, in this spiking experiment we would expect to have roughly 52% power to detect allelic frequency differences of 0.10 between the control and spiked pools. This is less than the power estimate from the control DNA pools because here we have only two replicates of each spiked pool, and nine replicates of the control pool, which is equivalent to having equal groups of *n*= 3.27 (the harmonic mean of 9 and 2). The expected differences in the spiked pool are dependent upon the original allele frequencies in the control pool because we used a fixed total amount of (i.e. the amount of pool DNA is reduced to make room for the spike DNA). For the SNPs where the expected allele frequency difference in the 15% and 20% spiked pools was more than 0.10 and we had complete data (*n*= 658), a significant *t*-test difference was observed for 379 SNPs (58%), supporting our predicted power calculations. As shown in Table 2, the observed and predicted differences between spiked (\'case\') and control correspond well on average over the entire frequency range, although the predicted allele frequency differences are greater than the observed differences. The expected allele frequencies are calculated from frequencies measured in the control pools, and are greatest for low frequency alleles, some of which appear low by chance due to measurement error. The apparent bias shown in table 2 is thus due to regression to the mean.
Discussion
==========
We have demonstrated that the SNP-MaP method can tackle the QTL problem of large samples and large numbers of SNPs. Specifically, allele frequencies can be accurately estimated from pooled DNA allelotyped on microarrays using the sophisticated allele-specific hybridisation method packaged in the commercially available Affymetrix GeneChip^®^system. The validity and sensitivity of the SNP-MaP method was shown by the correlation of 0.986 between estimates of allele frequency from pooled DNA and individual genotyping. The sensitivity \'spiking\' experiment confirmed these estimates, suggesting that case-control pool allele frequency differences on the order of 0.10 could be detected above the noise of measurement error using this method. With this level of sensitivity, the SNP-MaP approach should be able to detect QTLs of modest effect size if large samples are used. The benefit of DNA pooling is that large samples cost no more to genotype than small samples.
These results make it reasonable to proceed to case-control studies of complex traits. For example, we are using the SNP-MaP method in a large-scale study of children of low versus high reading ability. In an actual association study, sampling variation is added to the parameters considered in the present study which focused on technical replicates within and between DNA pools of the same group of individuals and compared allele frequency estimates for pooled DNA from this group of individuals to individual genotyping. However, in an actual association study, pooled DNA estimates are not compared within and between pools of the same individuals but rather between different pools consisting of different individuals. For this reason, in an actual association study, sampling variation would be considered and could be estimated parametrically by including in the design sub-pools of independent samples of cases and controls. This would permit the use of standard parametric statistics comparing the mean allelic frequencies of cases and controls in which *N*is the number of subpools. In addition, using at least two microarrays on each subpool would alleviate technical variation between microarrays and sub-pools. It should be emphasised that DNA pooling is best construed as nominating SNPs that are then confirmed with individual genotyping.
We have used the quantitative RAS values, designed to produce a robust allele call when assessed for individuals, to estimate allele frequencies for pooled DNA. It is likely that a more elegant way can be found to derive an allele frequency estimate from the 40 oligonucleotide probes present for each SNP (20 matches and 20 mis-matches). For example, it has been reported in a study of gene expression differences that increased power can be obtained by testing for differences for each probe rather than for differences averaged over the entire probe set level \[[@B22]\].
The tests used here were not corrected for multiple testing because the intent is to use this method not as a definitive analysis, but rather as a rapid initial screen to reduce the number of candidate SNPs to be submitted to individual genotyping for the large samples needed to detect QTLs of small effect size. With 10,000 SNPs, a large number of false positive results are expected by chance and so we recommend using a multi-stage replication design in order to balance false positives and false negatives in the selection of SNPs for individual genotyping \[[@B13]\].
Our analyses of SNP-MaP were based on the 10,000 SNP microarray, but should also be applicable to the 100 k Affymetrix GeneChip^®^system launched in June 2004 and the 500 k Affymetrix GeneChip^®^system scheduled for the end of 2005. Although the SNPs on the Affymetrix arrays are chosen on the basis of coverage of the whole genome (not on the basis of functionality or location in or near coding regions), 500,000 SNPS approaches a comprehensive genome scan for association based on linkage disequilibrium. The SNP-MaP approach will also be relevant to other types of SNP microarrays such as a microarray that emerges from the exon re-sequencing project at the Sanger Centre which aims to identify all nonsynonymous (nsSNPs), which are likely to be functional because the polymorphism involves an amino acid difference in translation. Microarrays with functional SNPs will facilitate direct association analyses with increased power as compared to indirect association analyses based on linkage disequilibrium \[[@B23]\].
Conclusion
==========
In summary, we have shown that the SNP-MaP method is a reliable and powerful approach to screen thousands of loci using large sample sizes. Employing case-control designs using SNP-MaP will speed-up the SNP nomination process for individual genotyping, and thus help detect SNPs of small effect size more efficiently.
Methods
=======
Samples
-------
The sample consisted of 100 Caucasian individuals (51 females and 49 males) randomly selected from an ongoing study of cognitive ability that is described elsewhere \[[@B24]\].
DNA quantification and pool construction
----------------------------------------
DNA samples were quantified using a spectrophotometer (260 nm), and diluted in Te buffer (0.1 mM EDTA, 10 mM Tris HCL, pH 8.0) to a target concentration of 100 ng/μl, before being quantified using fluorimetry (Picogreen^®^dsDNA quantitation reagent Cambridge Bioscience, U.K) and diluted further to a target concentration of 50 ng/μl (± 5%). Pools were constructed independently in triplicate by combining 50 ng of DNA from each individual, producing a final pool concentration of 35.5 ng/μl.
An aliquot of control pool D was reconstructed to form artificial \'case\' pools by replacing 15 and 20% of the pool sample with the same quantity of DNA of one genotyped individual (the spike). This changes the quantities of alleles present in a way that depends on their frequency in the pool and the genotype of the spike.
Validation of pool construction
-------------------------------
Three SNPs chosen from the microarray set were genotyped independently in duplicate on the three replicate control DNA pools, A, B and C using SNaPshot™, a commercially available dideoxynucleotide primer extension kit (PE applied biosystems). Electrophoresis of the samples was performed on an ABI Prism^®^3100 genetic analyser, analysed with Genotyper^®^3.7, and the allele frequencies derived from peak heights.
The correction factor, *k*, was calculated by individually genotyping seven known heterozygous individuals using the SNaPshot™ method for each of the three SNPs. The pooled allele frequency estimates were then corrected using the equation A = A/(A+*k*B), where A and B are the peak heights of alleles A and B respectively.
Individual genotyping
---------------------
104 randomly chosen SNPs from the microarray were genotyped in the 100 individuals that were used to construct the DNA control pool. The individual genotyping was outsourced to Kbiosciences, who use a mixture of competitive allele specific PCR (KASPar) and TaqMan genotyping assays <http://www.kbioscience.co.uk/>.
Genotyping DNA pools on the microarray
--------------------------------------
The individual spike DNA and 13 pooled DNA samples were genotyped using Affymetrix GeneChip^®^Mapping 10 K Array Xba 131. Standard procedures and default analysis parameters for individual DNA samples were employed, and each assay was independently amplified before hybridization. The scans were performed using GCOS V1.0 and the images (cel.files) were analyzed using GDAS V2.0.
Derivation of allele frequencies from pooled DNA
------------------------------------------------
The Affymetrix microarray uses a probe quartet as the basic unit for detecting different genotypes with a DNA sample. Each probe quartet consists of a perfect match (PM) and a mis-match (MM) probe for alleles A and B on both the sense and anti-sense strands. To make the genotyping more reliable, 7 probe quartets are used, with the polymorphic nucleotide having different shifts from the center of the 25-mer probe sequence. The best five quartets are used to create 40 hybridization intensity values.
A detection filter automatically blocks weak or unreliable signals by comparing the discrimination between perfect match and mis-match cells using the formula (PM-MM)/(PM + MM). Feature extraction processes the intensity values for all SNPs that pass the pre-determined discrimination threshold.
In feature extraction, PM probes for the sense strand of allele A (PM~A~) are corrected for the noise of non-specific hybridisation by averaging the mis-match values of both alleles (A and B) on the sense strand \[(MM~A~+ MM~B~)/2\] and subtracting this from the PM~A~value. This procedure for correcting the sense strand of allele A is repeated for the anti-sense strand of allele A, the sense strand of allele B and the anti-sense strand of allele B, resulting in four relative intensity values.
Relative allele signals (RAS) are then calculated for the sense (RAS1) and anti-sense strands (RAS2) using the formula \[PM~A~/ (PM~A~+ PM~B~)\]. This is done for each of the five quartets for each SNP and the median RAS1 and median RAS2 value is used to identify genotypes for individual DNA \[[@B25]\]. Plotting RAS1 scores against RAS2 scores, a clustering algorithm is used to determine individual genotypes. That is, if the two RAS values cluster near 0.0, they are identified as a BB homozygote; 0.5 for an AB heterozygote, and 1.0 for an AA homozygote. Affymetrix software (GDAS) incorporates empirically derived boundaries for each genotype\'s cluster, which is used to indicate genotypes; RAS scores falling outside these cluster boundaries are not assigned genotypes. For pooled DNA, we used only the detection filter, feature extraction processes, and RAS scores, disregarding the automated genotype calls. We used the average of the sense (RAS1) and anti-sense RAS2; RAS~av~as an estimate of allele frequency in the DNA pools.
As mentioned, a correction factor, *k*, is used to improve the accuracy of allele frequency estimates from pooled DNA. *k*was empirically derived from a panel of 33 individuals assayed on microarrays. *k*was calculated for each strand by dividing the mean of allele A (RAS1 or RAS2) by the mean of allele B (1 minus allele A). The mean of RAS1 and RAS2 was calculated by averaging a set of RAS1 and RAS2 scores from a panel of between 1 and 33 heterozygous individuals. In total, we obtained an estimate of *k*for 10,084 SNPs. Derivation and implementation of *k*in correcting pooled DNA estimates of allele frequency using microarrays are described in detail elsewhere \[[@B26]\].
Authors\' contributions
=======================
EM and LB constructed the DNA pools, performed the microarray assays (with assistance from LL and CF), statistical analysis (with LS), and drafted the manuscript (with LS). AC and VH performed the assays used to estimate k. RP, IC, LS and CF conceived and designed the study. All authors read and approved the final manuscript.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Predicted and observed allele frequency differences. The predicted allele frequency differences are the differences we would expect to see based on the allele frequency estimates from the control pools and the genotype of the spiker. The observed allele frequency differences are the actual frequencies observed in the spiked pool as compared to the control pools. Number of SNPs range between 459 and 2621.
:::
Predicted Observed N of observations
----------- ---------- -------------------
0.001 0.001 1116
0.010 0.007 2621
0.020 0.017 2478
0.030 0.024 2309
0.040 0.033 2043
0.050 0.042 1789
0.060 0.051 1450
0.070 0.062 1113
0.080 0.069 835
0.090 0.076 595
0.099 0.081 459
:::
Acknowledgements
================
We are grateful to Geoff Scopes and Affymetrix U.K. for advice. This work was supported in part by U.K. Medical Research Council grant G9424799 and by Wellcome Trust grant GR07549.
|
PubMed Central
|
2024-06-05T03:55:55.262484
|
2005-4-5
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079828/",
"journal": "BMC Genomics. 2005 Apr 5; 6:52",
"authors": [
{
"first": "Emma",
"last": "Meaburn"
},
{
"first": "Lee M",
"last": "Butcher"
},
{
"first": "Lin",
"last": "Liu"
},
{
"first": "Cathy",
"last": "Fernandes"
},
{
"first": "Valerie",
"last": "Hansen"
},
{
"first": "Ammar",
"last": "Al-Chalabi"
},
{
"first": "Robert",
"last": "Plomin"
},
{
"first": "Ian",
"last": "Craig"
},
{
"first": "Leonard C",
"last": "Schalkwyk"
}
]
}
|
PMC1079829
|
Background
==========
Delirium is a common presentation of illness in frail older adults and is associated with poor outcomes \[[@B1]-[@B7]\]. Risk factors for the development of delirium have been investigated \[[@B4],[@B8]\], but factors that predict poor recovery from delirium remain incompletely understood \[[@B9]\]. A recent study of elderly patients admitted to long-term care facilities found that longer duration of delirium was associated with worse functional outcomes \[[@B10]\]. There is also some evidence that hypoactive delirium is associated with poor outcomes \[[@B11]-[@B13]\], but results have been conflicting \[[@B14],[@B15]\]. Additionally, delirium is often under-recognized \[[@B16],[@B17]\], and this non-recognition is not without consequence. Compared with patients whose delirium is detected, patients in whom delirium goes unrecognized by treating doctors and nurses have higher 6-month mortality \[[@B18]\]. Risk factors for under-recognition by nurses include hypoactive delirium, age \>=80, vision impairment, and dementia \[[@B19]\]. On the other hand, a recent controlled trial found no benefit of an intervention to increase detection \[[@B20]\], suggesting that better detection without better than our current management might well be inadequate. Still, a single study should neither oblige us to consider that the case is closed, nor deter us from seeking better management methods \[[@B21]\]. Identification of factors predictive of outcome remains important, for prognostication, to advise about future care, and for targeting of better intervention and management strategies.
This study set out to identify factors associated with a composite outcome of incomplete functional recovery or death following an episode of delirium coinciding with an acute-care hospital admission. We evaluated both short term (by hospital discharge) and longer-term (at 6 months post-discharge) outcomes.
Methods
=======
The study was conducted in tertiary care medical, surgical, and geriatric ward settings in Halifax, Nova Scotia. The cohort was assembled through usual care and follow-up was done as a combination of usual care and research study protocol. Data were collected prospectively on consecutive patients who were diagnosed with delirium according to standard DSM-IIIR criteria by an attending geriatrician as part of usual care. Included patients were admitted to a geriatric service (N = 37) or seen in consultation on an internal medicine (N = 25) or surgical (N = 15) ward. Age, sex, Mini-Mental State Examination (MMSE) \[[@B22]\] at the time of assessment, presence and severity of dementia, pre-morbid function as indicated by the Barthel Index \[[@B23]\], frailty as measured by the Geriatric Status Score (GSS) \[[@B24]\], probable cause of delirium (categorized as medications, infection, cardiac, metabolic, or other/combination), whether delirium had been recognized as such by the treating service according to notes in the medical record, and presence or absence of agitation while delirious were recorded at the time of assessment. All data were obtained from the best available sources, including chart notes, health care professional or family report, and clinical examination. Pre-morbid function and cognitive status were determined based on a combination of the medical records and proxy report. Vital status and functional status were obtained at hospital discharge and 6 months following discharge, in clinic or telephone interviews by a study nurse.
Measures
--------
The Geriatric Status Score was developed at this institution \[[@B24]\], and used as the basis of a frailty score for community populations \[[@B25],[@B26]\]. Briefly, this ordered categorical scale is used to describe function (including cognition, activities of daily living, mobility, and incontinence) and frailty in older adults and has 7 categories: 1 -- healthy, no functional deficits, no cognitive impairment, 2 -- mild functional impairment, 3 -- moderate functional impairment, 4 -- moderate-severe functional impairment, 5 -- severe functional impairment, 6 -- total dependence, and 7 -- terminally ill and expected to die within 30 days. The version of the Barthel Index \[[@B27]\] used here is that modified by Granger, which is scored on a scale of 0 to 100 \[[@B23]\] with 100 indicting independence in mobility and Activities of Daily Living. Underlying dementia was diagnosed using DSM-IIIR criteria \[[@B28]\] and categorized as absent, cognitive impairment no dementia (CIND) \[[@B29]\], mild, moderate, or severe. Most probable cause of delirium was determined using available source including patient chart and patient and family interview. Two criteria were used: laboratory and/or radiologic evidence of the putative cause and temporal association. Probable cause was analyzed in two ways: as a categorical variable grouped according to category of cause (medication/alcohol, infection, cardiac, metabolic, other) and as a binary variable medications/alcohol *vs.*all other causes. This was done to test the hypothesis that delirium caused by medications may confer a better prognosis than other causes \[[@B30]\]. The determination of whether delirium had been recognized was made by one of the investigators (KR) or by a study nurse, using a protocol developed for another study \[[@B31]\]. Briefly, from the written record, we recorded whether any physician responsible for day-to-day care (attending or housestaff) had recorded within 24 hours of the delirium being present the terms \"delirium\", \"acute confusion\" or a reasonable synonym. This determination was done at the time of initial assessment by study personnel. Duration of delirium was recorded using best available information, as described in an earlier study \[[@B32]\].
Follow-up information on function was obtained for participants who were alive and contactable, either as part of usual care (n = 29) or in proxy-verified telephone follow-up (n = 21) for this study. The primary outcome was a composite of death or functional decline in both short- (at hospital discharge) and long- (6-months post-discharge) term. Functional decline was defined as a decrease by ≥10 points compared with pre-morbid BI score. This cut-off was chosen to represent clinically detectable functional decline (*e.g*. associated with loss of full independence in one functional domain, or lesser decreases in two separate functional areas). Based on our experience \[[@B31],[@B32]\], this would be clinically detectable (translating into an effect size of \~0.5, given a pooled standard deviation of \~20). Blinding was not possible in usual care follow-up patients.
Statistical analysis
--------------------
Data were analyzed using Stata 8 \[[@B33]\] and Statistix 8 \[[@B34]\] analytical software packages. Following descriptive analysis of associations using chi-square testing, logistic regression models were constructed. Functional decline (defined as a decrease of \>9 points on the Barthel index) and a composite outcome of death or functional decline were studied at hospital discharge and 6 month follow-up. All models adjusted for age, sex, and frailty according to the GSS.
Ethics
------
The project was approved by the Research and Ethics committee of Camp Hill Medical Centre, Halifax, Nova Scotia, Canada. Additional written, informed consent or telephone consent was obtained for the follow-up interview.
Results
=======
Of 77 patients with delirium during their acute-care admission, 6 died in hospital. Vital status at 6 months is known for 65 of the 71 survivors, of whom another 15 had died (6 months mortality = 30%). See Figure [1](#F1){ref-type="fig"}. At baseline, most patients were frail, with 19 (27%) having a pre-morbid BI score of 100, although a minority had dementia (Table [1](#T1){ref-type="table"}). No patient had a pre-morbid BI of zero (lowest score = 42), so all patients had the potential to show a decline in function.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Study cohort flowchart.
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Characteristics of the study population of patients with delirium (N = 77).
:::
**Patient characteristics** **Value** **N**
--------------------------------------------- ------------------------------ -------------- ----
Age (years) mean (SD) 78.5 (7.2) 77
range 64--93 years
Sex -- male N (%) 34 (44%) 77
Duration of admission mean days (SD) 38.6 (47.8) 77
median (interquartile range) 24 (15--41)
range 1--292
Duration of Delirium mean days (SD) 6.3 (6.1) 77
median (interquartile range) 5 (2--7)
range 1--35
Medications or alcohol as cause of delirium N (%) 30 (39%) 76
Cause of delirium N (%) 76
Medications, alcohol 30 (39%)
Infection 8 (11%)
Cardiac 12 (16%)
Metabolic 2 (3%)
Other 24 (32%)
Agitation present N (%) 12 (18%) 65
Poor recognition of delirium N (%) 25 (32%) 77
GSS -- Frailty N (%) 77
1 -- healthy, independent 4 (5%)
2 -- mild impairment 10 (13%)
3 -- moderate impair. 26 (34%)
4 -- moderate-severe 25 (33%)
5 -- severe impairment 10 (13%)
6 -- totally dependent 2 (3%)
7 -- terminally ill 0
Dementia 77
None 44 (57%)
Mild dementia 14 (18%)
Moderate dementia 15 (19%)
Severe dementia 4 (5%)
MMSE score/30 (SD) (on initial assessment) range 0--27 12.5 (7.4) 75
Barthel Index score (SD) range
Pre-morbid 42--100 86.6 (17.4) 77
At hospital discharge 0--100 78.9 (24.9) 71
At 6 month follow-up 6--100 78.2 (22.3) 50
Mean change in Barthel Index (SD)
Pre-morbid to discharge -8.9 (19.7) 71
Pre-morbid to 6 month -12.7 (16.9) 50
Functional decline N (%)
At hospital discharge 26 (37%) 70
At 6 months 27 (54%) 50
GSS = Geriatric severity score \[21\]. MMSE = Mini-Mental State Examination. SD = standard deviation.
:::
Functional status was known for 71 patients at hospital discharge and 50 survivors at 6 months. 6 patients were lost to follow-up after hospital discharge. By 6 months, most patients (48/71 = 68%) had a poor outcome (death or functional decline). Presence and severity of dementia, length of hospital stay, and duration of delirium did not show a statistically significant association with short- or long-term outcomes. Cardiac cause of delirium was associated with 8 times the odds of poor outcome at hospital discharge (95% CI 1.3--47.6) and a trend (p = 0.06) towards poor outcome at 6 months. Causes other than medications were also associated with death or functional decline at 6 months.
Poor recognition of delirium by the primary service was associated with death or poor functional recovery at hospital discharge (p \< 0.0001). This is a crude estimate only, as a model that adjusting for age, sex, and frailty was not possible because lack of recognition perfectly predicted poor outcome (*i.e.*no patient who was poorly recognized had a good outcome at hospital discharge). At 6 months, this crude association persisted (p \< 0.0001), and was robust to adjustment for age, sex, and frailty (Table [2](#T2){ref-type="table"}). [1](#T1){ref-type="table"} patient (4%) whose delirium had been poorly recognized had a good outcome at 6 months. Of those whose delirium was well recognized early in their clinical course, 7 patients (13%) had a poor outcome at hospital discharge, and 25 (53%) had a poor outcome at 6 months.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Odds of death or functional decline, adjusting for age, sex, and frailty.
:::
-------------------------------------------------------------------------------------------------------------------------------
**At hospital discharge**\ **At 6 month follow-up**\
**N = 77** **N = 71**
--------------------------------------- ---------------------------- --------------------------- ---------- -------------------
Age 0.7 1.0 (0.9, 1.1) 0.2 1.0 (1.0, 1.1)
sex 0.4 1.5 (0.5, 4.4) 0.9 1.1 (0.3, 3.4)
Frailty (GSS) 0.008 2.0 (1.2, 3.5) 0.03 1.8 (1.1, 3.2)
Length of admission 0.8 1.0 (1.0, 1.0) 0.5 1.0 (1.0, 1.0)
Duration of delirium 0.2 1.1 (1.0, 1.2) 0.08 1.2 (1.0, 1.4)
BI admission 0.8 1.0 (1.0, 1.0) 0.9 1.0 (0.9, 1.0)
Meds causing delirium 0.1 0.4 (0.1, 1.2) 0.02 0.3 (0.1, 0.8)
Cause: Meds Baseline Baseline
Infection 0.9 0.9 (0.2, 5.8) 0.4 2.2 (0.3, 15.1)
Cardiac 0.02 8.0 (1.3, 47.6) 0.06 9.2 (0.9, 91.3)
Metabolic 1.0 0.9 (0.0, 19.6) 0.08 3.3 (0.9, 12.5)
Other/combo 0.2 2.4 (0.7, 8.5) \*\* \*\*
Poor recognition \*\* \*\* 0.008 18.2 (2.2, 153.2)
Agitation 0.02 0.1 (0.0, 0.6) 0.2 0.4 (0.1, 1.8)
MMSE when assessed 0.1 0.9 (0.9, 1.0) 0.2 0.9 (0.8, 1.0)
Dementia severity 0.9 1.0 (0.6, 1.7) 0.9 1.0 (0.5, 2.0)
Dementia: none Baseline Baseline
Mild 0.2 0.4 (0.1, 1.8) 0.6 0.7 (0.2, 3.0)
Moderate 0.8 0.8 (0.2, 3.1) 0.7 0.8 (0.2, 3.5)
Severe 0.7 1.6 (0.1, 20.5) \*\* \*\*
Change in BI: pre-morbid to discharge 0.07 0.9 (0.9, 1.0)
BI at hospital discharge 0.1 1.0 (0.9, 1.0)
-------------------------------------------------------------------------------------------------------------------------------
\*\* unable to construct model as success is predicted perfectly
CI = confidence interval, OR = odds ratio, GSS = Geriatric severity score \[21\], MMSE = Mini-Mental State Examination, BI = Barthel Index.
:::
Adjusting for the effects of age and sex, increasing frailty was associated with poor functional recovery at the time of hospital discharge and at follow-up. Taking age, sex, and frailty into account, agitated delirium appeared to be protective against poor outcome at hospital discharge, but no statistically significant association was seen at 6-month follow-up.
The possibility of interaction was investigated where considered plausible. No evidence of interaction was found between frailty & age, dementia & age, or MMSE score & Barthel Index. Interaction between agitation and poor recognition could not be tested because poor recognition predicted failure perfectly.
Discussion
==========
Both frailty and poor recognition of delirium by the primary managing service were associated with greatly increased odds of poor outcome, defined as death or functional decline, in both short and long term. Absence of agitation was associated with poor outcome at hospital discharge. We also identified a trend towards a decline in functional status over the course of the hospital admission being a predictor of poor outcome at 6 months. In contrast to the results of another study that found longer duration of delirium symptoms to be associated with poor functional outcomes among patients admitted to long-term care facilities \[[@B10]\], neither duration of delirium nor length of hospital stay was associated with functional recovery in our hospital-based study.
Our findings must be interpreted with caution. Our small sample size (N = 77) likely resulted in our not detecting statistically significant associations between some of the factors investigated and the outcomes, such as the relationship with dementia, which here was swamped by the association with frailty. At a significance level of 0.05, our study had a power of 0.8 to detect a BI difference of 10 points (corresponding to an effect size of 0.7) between patients whose delirium had been well *vs.*poorly recognized, whereas it was under-powered to detect such a difference in outcomes between patients with no dementia *vs.*those with dementia (power 0.6 to detect a difference of 10 points by hospital discharge and power 0.7 at six months). Additionally, as delirium is a clinical diagnosis, there is always some diagnostic uncertainty. This might have impacted both the diagnosis (and thus inclusion of patients in our study population) as well as the assessment of duration of delirium. On the other hand, assessments were done by an experienced geriatrician using predetermined criteria. The outcome measure was a composite of death and poor functional recovery, as defined by a drop of \>=10 points on the BI. This endpoint thus relies on the inter- and intra-rater reliability of the BI, which has been shown to be good \[[@B35],[@B36]\]. Additionally, some of the follow-up data were collected by telephone interview. BI scoring based on telephone interview has been found to be reliable \[[@B35],[@B37]\].
Although delirium is common in frail elderly patients, of whom 10--60% have delirium, an estimated 22--66% of cases go unrecognized \[[@B16]\]. Our findings suggest that poor recognition of delirium in hospitalized patients is a risk factor for mortality and poor functional outcomes. This is consistent with findings from a study of elderly patients discharged home from emergency departments, in which undetected delirium was associated with increased 6-month mortality \[[@B18]\]. Interestingly, since all of our patients were seen at some point by a geriatrics service at which time delirium was recognized and management suggestions made, it appears that detection early in the course of delirium is crucial. However, since the precise duration of delirium prior to its recognition is unknown, it is uncertain how long the delay in recognition must be before it becomes associated with adverse outcomes. It is also possible that differences in implementation of management recommendations between clinical services may have confounded the apparent association between recognition and outcomes. The relationship between poor recognition and adverse outcomes was recognized using a composite outcome as developed for the primary analysis. Considered separately, the relative risk (RR) of death given poor recognition of delirium was 10.4 (95% CI 1.3--84.5) at hospital discharge and 2.6 (95% CI 1.3--5.3) at 6 months. The RR of functional decline was 8.5 (4.0--18.0) at hospital discharge and 2.2 (1.4--3.3) at 6 months. While poor recognition was associated with mortality and functional decline, there was no statistically significant association with institutionalization (OR 2.3, 95% CI: 0.7--7.4). However, only 20 patients were discharged to long-term care facilities, so sample size may have been insufficient to demonstrate association. Lack of recognition was also not associated with duration of delirium or length of hospital stay. Insufficient power may have been an issue in these analyses as well. There was no relationship between cause of delirium and lack of recognition (p = 0.2).
Poor recognition of delirium is relative here -- ultimately, each patient was recognized to have had delirium, and some attempt at management, however late in the course, was made. Identification of patients in the course of usual care may have resulted in some profiles of delirious patients being disproportionately excluded (*e.g.*hypoactive delirium or mild cases). Without a systematic assay of delirium in patients who were not referred, we cannot be sure of the effect of this selection bias. Arguably, patients in whom delirium was not recognized, who were never referred for geriatrics consultation, and who were thus excluded from the study, might either have been well enough to go home, or were seen to have done so badly that no attempt at improving outcomes was made.
Six patients were lost to follow-up at 6 months, of whom 5 had been well recognized as being delirious by their treating services. In order to assess the impact of missing follow-up data, we conducted a best/worst case scenario sensitivity analysis, the results of which demonstrated our findings to be robust: if all 6 patients whose outcomes were unknown at 6 months were assumed to have had a poor outcome, the odds ratio of unfavourable composite outcome given poor recognition was 16.1 (95% CI: 2.0--133.4), while if all 6 were assumed to have had a good outcome the OR was 11.6 (95% CI: 2.3--57.7).
In our cohort, agitation was apparently protective for both short- and long-term outcomes. This might have been due to agitation leading to better recognition. Considering agitation as the exposure and recognition as the outcome, those who were agitated were less likely to have poor recognition of their delirium. Adjusting for age, sex, and frailty according to the GSS, the odds ratio for poor recognition in patients with agitation (compared to those without) was 0.09 (95% CI 0.0--0.8; p = 0.03). Among those whose delirium was recognized, there was no statistically significant association between presence of agitation and outcome at hospital discharge (p = 0.4) or after 6 months (p = 0.9).
Better outcomes with agitation is consistent with some previous studies, which found hypoactive delirium to be associated with longer hospital stay, more severe illness, increased pressure ulcers and hospital-acquired infections \[[@B11]\], and longer duration of delirium \[[@B12]\]. However, others have reported equivalent \[[@B14]\] or better \[[@B15]\] outcomes in patients with hypoactive delirium. Although such differences may point towards etiological and pathophysiological differences between clinical subtypes of delirium \[[@B13],[@B38],[@B39]\], it is also possible that the agitation of hyperactive patients leads to better recognition by health care staff and thus to more timely and better management of the delirium and its underlying causes \[[@B11]\]. Our finding that absence of agitation was associated with increased odds of poor recognition by the primary treating service lends support to this hypothesis, given that poor recognition was also associated with much higher odds of poor outcome at both hospital discharge and 6-month follow-up.
Additionally, we found that among patients whose delirium had been appropriately recognized, agitation was not associated with better or worse outcome. This lends support to the idea that agitation is beneficial only in so far as it draws attention to the patient and his or her delirium rather than signaling an intrinsically more benign pathophysiological state.
Adjusting for frailty, age, and sex, cardiac cause of delirium was associated with worse outcomes at hospital discharge, and medications as cause were found to portend a better outcome at 6 months. This is consistent with previous findings \[[@B30]\] and may point towards differences in underlying pathophysiology.
Interestingly, we found that the magnitude of change in BI scores from pre-morbid to hospital discharge, more so than the absolute value of the discharge BI, may be a relevant predictor of death and poor functional recovery 6 months following delirium. This is only a trend (p = 0.07), although a larger sample might have found a statistically significant difference. Of note, it is also consistent with an earlier study, \[[@B40]\] which found that change in the BI was associated with adverse outcomes by the time of hospital discharge. Marked change in function over the course of hospital admission may reflect a more severe delirium associated with greater momentum of a downward spiral from which functional recovery is difficult. Adjusting these models for length of hospital stay did not affect the results.
Conclusion
==========
We have identified some factors that are associated with unfavourable outcomes at hospital discharge and at 6-month follow-up in patients who have had an episode of delirium. While pre-existing frailty and cause of delirium are not readily modifiable risk factors, poor recognition by treating physicians early in the course of delirium may be, suggesting that a potential remedy for the poor outcomes associated with delirium may be within the grasp of attentive practitioners. While further research is needed, our findings contribute to the literature on the important subject of patients\' recovery following delirium. It is our hope that awareness of factors that portend a poor outcome will be useful in discussions relating to management, prognosis and future care needs for older adults with delirium.
Authors\' contributions
=======================
KR collected the clinical data. All three authors participated in the design of the analyses. MKA did the analyses and wrote the first draft of the paper. SHF & KR reviewed the analyses and revised the manuscript. All authors read and approved the final manuscript.
Competing interests
===================
The author(s) declare that they have no competing interests.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2318/5/5/prepub>
Acknowledgements
================
Melissa Andrew was supported by a Ross Stewart Smith Fellowship and a Dalhousie University Internal Medicine Research Foundation Research Fellowship. Kenneth Rockwood receives support from the Canadian Institutes for Health Research through an Investigator award, and from the Dalhousie Medical Research Foundation as Kathryn Allen Weldon Professor of Alzheimer Research.
|
PubMed Central
|
2024-06-05T03:55:55.265321
|
2005-3-17
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079829/",
"journal": "BMC Geriatr. 2005 Mar 17; 5:5",
"authors": [
{
"first": "Melissa K",
"last": "Andrew"
},
{
"first": "Susan H",
"last": "Freter"
},
{
"first": "Kenneth",
"last": "Rockwood"
}
]
}
|
PMC1079830
|
Background
==========
A randomized, controlled field trial of a multifaceted intervention to improve preventive care tailored to the needs of participating family practices was conducted in Southern Ontario and delivered by nurses trained in the facilitation of prevention. This report is a cost-consequences analysis of the intervention. Specifically, it provides information about the cost of the outreach facilitator intervention and money saved the health care system as a result of reducing inappropriate and increasing appropriate preventive screening tests.
Improving preventive performance is both important and necessary. There is substantial room to improve rates of appropriate preventive practice \[[@B1]\]. The Canadian Task Force on the Periodic Health Examination \[[@B2],[@B3]\] has established guidelines for the delivery of preventive care that are supported by clinical evidence as effective in decreasing the impact of disease. However, evidence-based guidelines are not self-implementing \[[@B4]-[@B6]\]. Changing physicians\' long-held patterns of behaviour and the environments in which they work is essential yet complex and difficult. Unless the barriers to change can be overcome and actions taken to put preventive care guidelines into practice, evidence-based guideline development efforts will be wasted and the quality of preventive care will not improve \[[@B7]\].
Several reviews have focussed on the effectiveness of different interventions for implementing guidelines and improving care \[[@B5],[@B6],[@B8]-[@B12]\]. Multi-faceted outreach facilitation interventions employing trained individuals who meet with providers in their practice setting to provide information and assist the practice in implementing evidence-based guidelines have been shown to be more effective than single interventions such as guideline dissemination efforts or physician prompts \[[@B10]-[@B14]\]. Tailoring interventions to the requirements of the practice has also been proposed as important in supporting practice changes and in attaining more successful and sustained outcomes in preventive care performance as compared to interventions that are fixed and lack this flexibility \[[@B15]-[@B19]\]. Given the diversity of practice environments, it is unlikely that \"one size fits all\" approaches to improving preventive care will be able to address the needs of all providers and their patients \[[@B20],[@B21]\].
Successful interventions designed to improve compliance with evidence-based guidelines for preventive care could have an important influence on the health of Canadians. In addition, interventions designed to reduce the ordering of inappropriate tests in delivering preventive care have the potential to offset some of the intervention costs.
Economic evaluations of outreach facilitation
---------------------------------------------
By its nature, outreach facilitation is multifaceted. In the United Kingdom and the United States specially trained nurse facilitators organized preventive care in busy practitioners\' offices through outreach visits and using approaches such as academic detailing, chart audit and feedback for the prevention and early detection of cardiovascular disease and cancer \[[@B22],[@B23]\]. Very few evaluations of outreach facilitation have studied the costs of delivering these interventions. The Cochrane Effective Practice and Organization of Care Group has concluded that outreach visits are effective, however, cost-effectiveness needs to be determined \[[@B24]\]. Soumerai and Avorn have suggested that the savings from outreach facilitation may outweigh the costs if the intervention is targeted at inappropriate and costly practice behaviour \[[@B25],[@B26]\].
Cockburn and colleagues conducted a randomized controlled trial in which the effectiveness of three approaches to marketing a quit smoking intervention kit to physicians was evaluated \[[@B27]\]. They conclude that educational outreach facilitators do not appear to be cost effective strategies for distributing smoking interventions. The actual use of the kit by the physicians for their smoking patients did not differ significantly across groups. However, there was a trend toward higher use in the facilitation group for one of the components of the kit as compared to those who received the kit by courier or standard mail.
Conversely, McCowan et. al. conducted a randomized controlled trial to examine the effect of a facilitator intervention on the management of children with asthma by family physicians \[[@B28]\]. They found that the facilitator intervention reduced asthma care costs in the intervention group as compared to the control resulting in an overall net saving of 12,000 (U.K. 1991) pounds or one pound less per child per annum. The facilitator accomplished this by inserting guidelines for the management of asthma into intervention practices\' case records. The authors estimate that the net savings to the health system would recoup the facilitator\'s salary at 1991 rates.
The literature on the costs of outreach facilitation is limited. It can be argued that facilitation is a costly intervention \[[@B27]\]. However, a costly intervention that achieves success may be preferred to a cheaper one that demonstrates very little or has no lasting effect. More research on the costs of successful facilitation and other effective alternative interventions to outreach facilitation is necessary.
Intervention description
------------------------
The nurse facilitation in our study was a tailored multifaceted approach to getting evidence into action. Three nurse facilitators focused on the educational, attitudinal and organizational barriers to change in the practice setting and tailored a multi-component intervention to the specific needs of the practice \[[@B22],[@B23],[@B29],[@B17]-[@B34]\]. The facilitators completed a 30-week intensive training program before being assigned to intervention practices. The training covered an orientation session, medical office computer systems, medical practice management, prevention in primary care, evidence-based medicine, and facilitation and audit skills development. Approximately 28 hours per week were spent in training and 7 hours per week in preparation and planning. Six of the 30 weeks of training were spent applying skills in a primary care office setting.
The intervention period was 18 months ending December 1998. During this period each intervention practice was visited an average of 33 times (range 21 to 50) at an average visit length of one hour 45 minutes. The facilitators delivered primarily three intervention strategies to improve preventive care: chart audit and feedback, educational consensus building, and reminder systems. They discussed the strategies with the physicians and practice staff, working with them to adapt the strategies to the practice needs and wishes. All 22 of the intervention practices participated in an initial audit and received feedback on preventive care practice patterns. Twenty practices requested subsequent audits and analyses of data to follow their rates of performance. All of the practices were involved in meetings with the facilitator to identify opportunities for improvement, assess needs, receive and discuss critically appraised evidence from the literature for the preventive maneuvers, and select strategies for improving preventive care performance. All of the intervention practices implemented some form of a reminder system as a strategy to improve performance. Eighteen practice sites implemented a preventive care flow sheet; two sites used a chart stamp; and two sites implemented a computerized reminder system.
The facilitators provided management support to practices and followed a quality improvement framework similar to that proposed by Leininger and colleagues \[[@B30]\]. For each practice the facilitators were to: (1) present preventive performance rates prior to the intervention, (2) facilitate the development of a practice policy for preventive care, (3) assist in the setting of goals and desirable levels of performance, (4) assist in the development of a written plan for implementing preventive care, (5) assist in the development and adaptation of tools and strategies to implement the prevention plan, (6) facilitate meetings to assess progress and modify the plan if necessary, and (7) conduct performance feedback to measure the effect of changes made.
The facilitators had no interaction with control practices. The latter were told that they were involved in a study on prevention but were not told which preventive manoeuvres were being measured. More details on the multi-component nature of the intervention are published elsewhere \[[@B35]\].
Setting
-------
The Prevention Facilitator intervention involved Health Service Organizations (HSOs) in Ontario. HSOs are community primary care practices that have a payment system based primarily on capitation and not fee-for-service. At the time of the study, there were 72 physician-sponsored HSOs located at 100 different sites in Ontario. The study involved 106 physicians from 45 HSOs. All physicians gave informed consent to participate in the trial.
Intervention outcomes
---------------------
The goal of the intervention was to increase those preventive manoeuvres supported by evidence as appropriate and decrease those preventive manoeuvres supported by evidence as inappropriate. Eight grade A and B, and 5 grade D preventive manoeuvres were chosen by a panel of practicing family physicians from the Canadian Guide to Clinical Preventive Health Care\[[@B3]\] to represent a broad spectrum for both male and female adult patients (see Table [1](#T1){ref-type="table"}). The grade A and B manoeuvres are supported by evidence as appropriate and the grade D manoeuvres are supported by evidence as inappropriate. As determined by chart audit, an absolute change overtime of 11.51% in preventive care performance in favour of intervention practices was achieved. More detailed outcome results of the randomized controlled trial are reported elsewhere \[[@B36]\].
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Preventive manoeuvres studied
:::
**Level of Evidence** **Preventive Manoeuvre**
------------------------------------------ -------------------------------------------------------------------
**A^Ω^& B^¥^-categories**: (Appropriate) 1\. Folic acid for primary prevention of neural tube defects
2\. Smoking cessation and nicotine replacement
3\. Treatment for Hypertension
4\. Mammography and exam in women over 50
5\. STD screening for high risk groups
6\. Papanicolaou smears for sexually active women
7\. Influenza vaccination to patients 65 and older
8\. Blood pressure measurement for patients 21 to 64 years of age
**D^ψ^- category:**(Inappropriate) 1\. Proteinuria screening for general population
2\. Blood glucose for the general population
3\. Prostate-specific antigen testing for men over 50
4\. Chest radiography
5\. Mammography in women under 50
^Ω^There is excellent evidence from repeated randomized controlled trials to support the manoeuvre.
^¥^There is good evidence from cohort and case-control studies to support the manoeuvre.
^ψ^Not recommended on the basis of fair evidence not to perform the manoeuvre.
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Demographic profile comparison of intervention and control practices
:::
**Measure** **Intervention Group (N = 22)** **Control Group (N = 23)** **Significance (P value)**
----------------------------------------------- --------------------------------- ---------------------------- ----------------------------
Percentage of Group Practices 77.3% 60.9% .34
Percentage Teaching Affiliated 54.5% 52.2% 1.00
Percentage in communities greater than 50,000 86.4% 65.2% .17
Mean number of physicians in group practices 2.91 2.70 .71
Mean number of registered nurses in practices 1.16 1.64 .48
Mean year of graduation from medical school 1975 1975 .92
Mean proportion of female physicians 12.6 20.4 .37
Mean roster size 4317 3874 .55
Mean number of patients seen per day 34.4 33.0 .59
Percentage of female patients served 53.4 53.8 .89
Mean age of patients served 46.4 46.8 .87
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Comparison of intervention and control practices on delivery of preventive manoeuvres to eligible patients post intervention (N = 4501)
:::
**Preventive Manoeuvres** **Proportion of Eligible Patients**
----------------------------- ------------------------------------- -------------------- --------------
Intervention (n = 2201) Control (n = 2300) Significance
**A & B Manoeuvres** \% (N) \% (N)
Folic Acid Pre-conception 15.4% (325) 4.9% (369) .0001
Cessation Counselling 41.7% (571) 40.6% (549) N.S.
Mammography 50 to 69 68.3% (325) 57.5% (358) .005
Hypertension Treatment 79.7% (169) 82.7% (185) N.S.
STD Screening 23.3% (382) 19.1% (366) N.S.
BP Measurement 74.6% (1666) 72.5% (1781) N.S.
FLU Vaccination 66.0% (692) 53.8% (652) .0001
Cervical Cytology 66.2% (826) 60.2% (958) .01
**D Manoeuvres**
Blood glucose screening 32.8% (1844) 38.7% (1980) .0001
PSA Testing 30.6% (379) 30.0% (393) N.S.
Mammography 40 to 49 ^1^ 11.6% (267) 9.1% (309) N.S.
Chest X-Ray 3.7% (571) 4.9% (549) N.S.
Urine proteinuria screening 16.5% (1772) 29.8% (1887) .0001
^1^This was a grade D manoeuvre at the time of the study; the Canadian Task Force on Preventive Health Care has recently changed it to a Grade C manoeuvre.
:::
We wanted to estimate the cost savings associated with an effective outreach facilitation intervention designed to reduce inappropriate and improve the delivery of appropriate preventive care.
Methods
=======
The analysis was part of a randomized controlled trial to determine if outreach facilitation was effective in improving prevention in primary care. The design of the trial and details of the method have been described in full elsewhere \[[@B36]\]. Practices were randomized to either intervention (22 practices) or control status (23 practices).
Patient eligibility
-------------------
Patients eligible for the intervention were greater than 19 years of age, asymptomatic and had received the intervention for screening and prevention purposes.
Cost analysis methodology
-------------------------
We performed a cost-consequences analysis to evaluate the outreach facilitation intervention \[[@B37]\]. We determined the incremental costs of the preventive manoeuvres performed and the overall cost of the intervention. We estimated the cost-savings as a result of having improved appropriate preventive care and reduced inappropriate preventive care between intervention and control conditions. All of the appropriate and inappropriate preventive manoeuvres from the trial were included in the analysis, not only those that showed significant improvement. Multiple data sources were used to determine the costs of preventive procedures, treatment, and efficacy of preventive manoeuvres (see Tables [4](#T4){ref-type="table"} and [5](#T5){ref-type="table"}). We determined the costs and savings associated with outreach facilitation by calculating the mean difference between intervention and control groups in the number of eligible patients screened or treated for appropriate preventive manoeuvres and the mean difference in the number of eligible patients not screened for inappropriate preventive care. The cost-consequences analysis is conducted from the perspective of the Ontario Government. All costs are presented on the basis of one year rather than over the 18-month period of the intervention to correspond with the government one-year planning and budgeting cycle. Costs were converted into Canadian dollars using the nominal rate method and adjusted for inflation (1999 current dollars) where necessary. The net cost savings per patient and per facilitator are presented.
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Input variables and cost estimates for appropriate manoeuvres (1999 dollars)
:::
**Variable** **Value and Range^a^** **Source**
----------------------------------------------------- ------------------------ ------------
**A & B Manœuvres**
Folic Acid
Difference in no. of eligible patients counseled 1495 (854, 2136) (36)
Probability of reducing neural tube defects .00058 (55)
Life-time treatment costs of Spina Bifida \$201,822^b^ (51)
Smoking Cessation Counselling/NRT
Difference in no. of eligible patients counseled 253 (-1072, 1577) (36)
Efficacy of NRT .06 (3)
Incidence of lung cancer in smokers .0024 (48)
Treatment costs for lung cancer \$7,074 (56)
Mammography 50 to 69 years of age
Difference in no. of eligible women screened 1513 (505, 2521) (36)
Incidence of breast cancer .003165 (42)
Cost of a mammogram \$76.54 (57)
Treatment costs saved for each breast cancer \$2,522 (42)
Hypertension Treatment
Difference in no. of eligible patients treated -145 (-739, 449) (36)
Efficacy of treatment of stroke .42 (3)
Incidence of stroke .0018 (47)
Efficacy of treatment of heart disease .14 (3)
Incidence of heart disease .00196 (47)
Treatment cost per case for stroke \$3,815 (56)
Treatment cost per case for heart disease \$3,303 (56)
STD Screening
Difference in no. of eligible patients screened 645 (-253, 1542) (36)
Cost of gonorrhoeae and chlamydia culture \$55 (57)
Incidence of gonorrhoeae or Chlamydia infection .08 (49)
Pelvic inflammatory disease (PID) prevented .14 (50)
Treatment cost per case for PID \$1,782 (49)
Flu Vaccination
Difference in no. of eligible patients vaccinated 3364(1928,4799) (36)
Cost of flu vaccination \$3.75 (43)
Pneumonia hospitalizations averted 4.1 per 1000 (43)
Chronic respiratory hospitalizations averted 10.4 per 1000 (43)
Congestive heart failure hospitalizations averted 2 per 1000 (43)
Emergency room visits avoided 21.6 per 1000 (43)
Emergency room visit cost \$76.00 (44)
Cost of pneumonia hospitalization \$4,462 (56)
Cost of chronic respiratory hospitalization \$4,445 (56)
Cost of heart failure hospitalization \$5,417 (56)
Cervical Cytology
Difference in no. of eligible patients screened 2196 (559, 3833) (36)
Cost of PAP test \$57.17 (57)
Incidence of cervical cancer .00013 (46)
Treatment costs saved for each women screened \$9,813 (41)
^a^95% Confidence Intervals
^b^10% discount
:::
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
Input variables and cost estimates for inappropriate manoeuvres (1999 dollars)
:::
**Variable** **Value and Range^a^** **Source**
------------------------------------------------ ------------------------ ------------
**D Manoeuvres**
Blood Glucose Screening
Difference in eligible patients not screened 4709 (2329,7089) (36)
Specificity of blood glucose test 89% (3)
Cost of fasting glucose test \$10.34 (57)
Cost of glucose tolerance test \$23.26 (57)
Cost of HgA1c \$19.12 (57)
PSA Testing
Difference in eligible patients not screened 95 (-932, 1122) (36)
Cost of PSA Test \$25 (57)
Specificity of PSA test 40% (3)
Cost of Biopsy\* \$232.83 (58)
Mammography 40 to 49
Difference in eligible patients not screened 295 (-296, 887) (36)
Cost of mammogram \$76.54 (57)
Incidence of breast cancer .001616 (42)
Specificity of mammogram 96.5% (3)
Cost of Biopsy \$2,164 (42)
Chest X-Ray
Difference in eligible patients not screened 276 (-271, 822) (36)
Cost of chest x-ray \$233 (40)
Urine Proteinuria Screening
Difference in eligible patients not screened 9947 (7934,11961) (36)
Specificity of urine dipstick test 95% (3)
Cost of Urine culture and urinalysis \$27.18 (57)
^a^95% Confidence Intervals
\*Includes intracavitary ultrasonography, ultrasound guidance of biopsy and needle biopsy of the prostate
:::
Intervention costing methodology
--------------------------------
The actual cost of the intervention over 18 months was gathered from administrative expenditure records for labour, supplies, telephone and travel. The investigation team determined that the cost of training the nurse facilitators was important to include but that the investment in training was worth more than the 18-month period of the intervention. Therefore, the cost of the training of the three nurse facilitators was depreciated over a 5-year life at a discount rate of 5% using the double-declining balance method for the 18-month period of the intervention. To correspond with the government planning and budgeting perspective, research costs were not included but supervision was. The justification for excluding research costs and including supervision was to model the program as if it were implemented as an actual government sponsored intervention. As a government sponsored program there would be no research salary and operations costs, but there would still be the need to supervise the facilitators. In addition, the cost of physician time for administering preventive manoeuvres was not included. The rationale for excluding physicians\' time for administering preventive manoeuvres is based upon the fact that the physicians were not reimbursed on a fee-for-service basis but rather were reimbursed on a capitation basis and the intervention did not result in an increased number of visits keeping costs the same in both the intervention and control groups. Within a capitation based system, physicians are reimbursed according to the size of their patient roster and not by fee-for-service. Physicians within a capitation based system do not receive additional fees for providing preventive services, nor do they receive any additional remuneration from seeing a patient more frequently. Therefore, this is not an additional cost to the Ontario provincial government.
Preventive manoeuvre costs
--------------------------
Using the most recent fee schedule for laboratory services, published by the Ontario Ministry of Health in April of 1999, we calculated the direct costs of performing seven appropriate and five inappropriate measures. Where cost estimates were unavailable from these two sources, published estimates for the cost of bilateral mammography\[[@B38]\], administering a flu shot\[[@B39]\], a chest x-ray\[[@B40]\] and cervical cancer screening\[[@B41]\]were used. The cost of folic acid, nicotine replacement therapy, hypertensive medication and antibiotics were not included since from the perspective of the Ontario government we determined that these costs would be substantially paid for directly by patients. The measurement of blood pressure was included as a case finding manoeuvre for hypertension as described by the Canadian Task Force on Preventive Health Care \[[@B3]\] but was not costed as 80% of eligible patients in treatment and control groups received a blood pressure measurement and as a consequence did not impact the cost analysis. Calculations were performed using Microsoft Excel Version 7.0.
Direct costs associated with inappropriate tests avoided are considered savings to the government whereas the costs associated with increasing the number of appropriate tests are considered costs to the government (see Tables [4](#T4){ref-type="table"} and [5](#T5){ref-type="table"}). The direct cost for each inappropriate test, according to the Ontario Ministry of Health, includes not only the test cost for performing the manoeuvre but also a \"Patient Documentation and Specimen Collection Fee\" which can be applied to a test or series of tests performed for one patient. The costs for appropriate tests were gathered from the estimates of other relevant cost studies (see Table [4](#T4){ref-type="table"}). For the purpose of calculating costs, the total population of 184,670 patients as of November 1998 for all 45 HSO practices was used to estimate the total number of intervention patients eligible (90,283) to receive a test according to the patients eligible for the selected tests from the study sample of 4,501 patients (2,201 intervention patients, 2300 control patients).
The cost of the initial visit to the physician was not included given that this cost would have been included whether a test was performed or not and these costs would be similar between the intervention and control groups and thus would not significantly impact the cost analysis. In addition, direct costs to the patient such as travel time to a laboratory to have a blood test have not been included. However, down-stream costs (follow-up visits and further tests) for the subsequent follow-up of false positive results for all screening manoeuvres where appropriate have been included. The down-stream costs included the cost of biopsies to determine breast cancer and prostate cancer as well as the cost of follow-up glucose tolerance tests to screen for diabetes. Any down-stream costs associated with other preventive manoeuvres were not considered for the analysis.
The number of expected false positive screening tests was determined from the specificity of a blood glucose test (89%), a urine dipstick test for protienuria (95%), and a PSA test (40%) \[[@B3]\]. A panel of five family physicians with approximately 100 years of combined practice experience determined through consensus the proportion of follow-up tests and visits needed. For estimating the total costs of performing follow-up on initially positive results for blood glucose, it was agreed that 40% would receive only a glucose tolerance test and 60% would receive both a fasting blood glucose and an HgA1c. For estimating the total costs of performing follow-up on positive urine protein tests, it was assumed that 100% of patients would receive both a urine culture and a microscopic urinalysis and a 24-hour urine. The documentation and specimen collection fee of \$7.75 was only included once for each patient visit. For estimating the total costs of performing follow-up on a positive PSA result, it was assumed that 90% of patients would be referred to an urologist \[[@B58]\]. The cost of family physician follow-up visits was not included given that the HSO physicians were not reimbursed on a fee-for-service basis. The input variables and cost estimates for appropriate and inappropriate manoeuvres are provided in Tables [4](#T4){ref-type="table"} and [5](#T5){ref-type="table"}.
Preventive manoeuvre savings
----------------------------
The input variables and cost estimates for savings associated with preventive manoeuvres are provided in Tables [4](#T4){ref-type="table"} and [5](#T5){ref-type="table"}. The difference in the number of patients receiving preventive manoeuvres between the intervention and control practices gave us the increase in savings to the government. Mammography can reduce breast cancer mortality by 20% to 30% \[[@B3]\]. For savings associated with Mammography in women 50 to 69, Salzmann et al. \[[@B42]\] was used to determine the difference in lifetime treatment costs of screened and unscreened women with breast cancer. The estimated difference is \$1,682 \$US 1995 or converted and inflated to 1999 dollars \$2,522 \$Can less for the screened group. It is assumed that the incidence of breast cancer in women 50 to 69 is 3.165 per thousand \[[@B42]\].
For savings associated with administering flu vaccine in the elderly, Nichol et. al.\[[@B43]\] provided the data on the number of hospitalizations of the elderly averted due to flu vaccination for pneumonia (4.1 per 1,000), chronic respiratory conditions (10.4 per 1,000), congestive heart failure (2 per 1,000) and the number of emergency room visits avoided (21.6 per 1,000). The estimated cost of an emergency room visit of \$76.00 \$Can 1999 as determined by Jacobs and Hall was used \[[@B44]\]. The Ontario Case Costing project provided the average length of stay and the total cost of hospitalization for pneumonia (\$4,462 \$Can 1999), chronic respiratory conditions (\$4,445 \$Can 1999) and congestive heart failure (\$5,417 \$Can 1999) for patients 65 years of age and older.
Performing Pap tests every three years reduces invasive cervical cancer by 91.2% \[[@B45]\]. Fahs et al. \[[@B46]\] provided the cervical cancer incidence rate (10 per 100,000 women) and Helms et al.\[[@B41]\] provided the difference in cervical cancer treatment at stage of diagnosis. The difference between the net costs of treatment for patients in the initial stages of cervical cancer and more invasive cancer is \$6,368 (\$US 1988) or \$9,813 (\$Can 1999).
For determining savings associated with smoking cessation counselling and treatment of hypertension, the Canadian Task Force on Preventive Health Care \[[@B3]\] provided the percent efficacy of cessation counselling and the treatment of hypertension for stroke and heart disease. The Canadian Institute for Health Information \[[@B47]\] provided the incidence of stroke and heart disease and Statistics Canada provided the incidence of lung cancer in smokers \[[@B48]\]. The Ontario Case Costing project provided the total cost for lung cancer hospitalization (\$7,074 \$Can 1999), stroke hospitalization (\$3,815 \$Can 1999), and hospitalization for heart disease (\$3,303 \$Can 1999) for determining overall treatment cost savings.
For savings associated with STD screening, Gift et. al.\[[@B49]\] provided the figure for the incidence of gonorrhoeae or chlamydia infection for eligible women as well as the cost to treat a case of pelvic inflammatory disease (PID) and the Centers for Disease Control and Prevention provided the figure for the prevention of pelvic inflammatory disease from treatment \[[@B50]\].
Finally, for determining the costs averted as a result of increasing folic acid intake, Waitzman et. al.\[[@B51]\] provided the data on the net direct medical costs associated with Spina Bifida (\$123,485 \$US 1996 or \$201,822 \$Can 1999) and the Canadian Task Force on Preventive Health Care and the CDC provided the estimates for the reduction in neural tube defects due to folate, an estimated 50% of all neural tube defects or 500 per million births \[[@B3]\].
After calculating the current (1999) cost of the intervention and the cost of the increase in delivery of recommended preventive manoeuvres we calculated the potential direct savings to the government for increased appropriate and decreased inappropriate preventive care by multiplying the costs with the estimated difference in patients screened or treated using the estimates and input variables provided in Tables [4](#T4){ref-type="table"} and [5](#T5){ref-type="table"}. We also calculated a low and high estimate of costs using the 95% confidence intervals of the difference in patients screened between intervention and control groups. These calculations are summarised mathematically in Appendix -- \[see [Additional file 1](#S1){ref-type="supplementary-material"}\].
Sensitivity analysis
--------------------
To account for uncertainty in the cost-consequences analysis, a 1,000 iteration Monto Carlo simulation was conducted to determine the impact of the range of input cost parameters on the net savings to government using Microsoft Excel. The cost input parameters included the normally distributed range of values between the 95% confidence intervals for percent difference between intervention and control practices on the preventive manoeuvre performance outcomes. These parameters are considered the most important in influencing overall costs and cost savings in the analysis. The ranges of randomly chosen estimates of each parameter were varied simultaneously to determine the impact on costs and cost savings. Finally, the ranges of input parameters were assessed individually on the ranges of the net savings outcome generated by the simulation to generate a Tornado diagram. Percent variance was used to determine the degree to which the outcome net savings was sensitive to the values of each input parameter and to determine which input parameter had the greatest impact on net savings.
Results
=======
Intervention (n = 22) and control (n = 23) group practices did not differ significantly on any of the demographic characteristics presented in Table [2](#T2){ref-type="table"}.
Intervention effect
-------------------
Table [3](#T3){ref-type="table"} presents the proportion of eligible patients that received recommended (A & B Manoeuvres) and inappropriate (D Manoeuvres) preventive care for both intervention and control groups of patients after the outreach facilitator intervention.
For inappropriate testing there was a statistically significant but small reduction in the proportion of eligible patients that received a random blood glucose test to screen for diabetes (32.8% vs. 38.7%) and a larger reduction for urine protein testing to screen for kidney disease (16.5% vs. 29.8%). There were no significant differences for PSA testing to screen for prostate cancer, mammography for women 40 to 49, or chest radiography to screen for lung cancer in smokers. For appropriate preventive manoeuvres there was a statistically significant difference for folic-acid counselling, mammography for women 50 to 59, flu vaccination, and cervical cytology.
Intervention costs
------------------
Table [6](#T6){ref-type="table"} provides data on the costs of the outreach facilitator intervention in 1999 dollars. The 18-month intervention cost a total of \$357,583.00 for three nurse facilitators including all travel, telephone, supplies and supervision or \$238,388 per year. Telephone costs include the cost of a cellular phone as well as the long distance costs of a home telephone. Supply costs include the costs of home-office supplies as well as the cost of materials for intervention purposes in the practices. The cost of the intervention on a yearly basis equates to \$10,835 per intervention practice (n = 22) or \$4,584 per intervention physician (n = 52) or \$328 per visit based on an average of 33 visits to the practice per year.
::: {#T6 .table-wrap}
Table 6
::: {.caption}
######
Intervention costs over 12 months
:::
**Cost Item** **1999 Dollars**
------------------------- ------------------
Staff training \$20,450
Salaries & Benefits \$178,200
Supplies \$7,060
Telephone \$8,629
Car Mileage & Insurance \$12,049
Supervision \$12,000
INTERVENTION COST \$238,388
:::
Table [7](#T7){ref-type="table"} provides the costs of providing recommended preventive care. The intervention resulted in 1,513 more mammograms having been done which is estimated to be an additional cost to the government of \$77,192 per year with a range of \$25,743 to \$128,641 depending on the number of mammograms. 645 additional cultures for gonorrhoeae and chlymadia were carried out in the intervention arm resulting in an estimated \$23,631 in costs. In addition, 3,364 more flue shots were provided in the intervention arm of the trial costing \$8,409 per year and 2,196 more women received a PAP test at a cost of \$83,679 per year (see Table [7](#T7){ref-type="table"}).
::: {#T7 .table-wrap}
Table 7
::: {.caption}
######
Costs associated with the increase of appropriate preventive manoeuvres (1999 Dollars)
:::
**Manoeuvre** **Baseline** **Low** **High**
---------------------- -------------- ------------- --------------
STD Screening \$23,630.87 -\$9,268.28 \$56,530.01
Mammography 50 to 59 \$77,192.11 \$25,742.87 \$128,641.34
Influenza shot \$8,409.23 \$4,820.83 \$11,997.62
Pap Test \$83,679.41 \$21,286.01 \$146,072.81
TOTAL \$192,911.62 \$42,581.43 \$343,241.78
:::
The total cost to the government for the intervention and additional manoeuvres performed was \$431,300 (95% CI: \$280,969 -- \$581,630) on a yearly basis, that equates to \$19,604 per intervention practice, or \$143,766 per facilitator or \$4.67 per HSO patient rostered (see Table [10](#T10){ref-type="table"}).
Cost savings
------------
Table [8](#T8){ref-type="table"} gives the estimated cost savings in 1999 dollars as a result of having significantly reduced inappropriate care in the intervention group for unrecommended screening manoeuvres. For the laboratory and diagnostic services cost model extrapolated to the entire eligible patient roster for intervention practices there is an estimated \$148,568 (95% CI: -\$175,898 -- \$473,035) in total savings including follow-up tests. This represents 34% of the costs of the outreach facilitator intervention and reduces the program to a cost of \$282,732.
::: {#T8 .table-wrap}
Table 8
::: {.caption}
######
Savings associated with the reduction of inappropriate preventive care (1999 Dollars)
:::
**Manoeuvre** **Baseline** **Low** **High**
---------------------- -------------- ---------------- --------------
Chest X-Ray \$42,827.36 (\$42,065.12) \$127,719.83
Mammography 40 to 49 \$16,107.12 (\$16,131.58) \$48,345.82
PSA Testing \$16,125.11 (\$158,156.15) \$190,406.37
Blood glucose \$59,990.54 \$29,672.59 \$90,308.49
Urine Protein \$13,518.31 \$10,782.33 \$16,254.29
TOTAL \$148,568.44 (\$175,897.93) \$473,034.80
:::
Table [9](#T9){ref-type="table"} provides the cost savings to the government as a result of averted hospitalizations and treatment costs. There is an estimated \$455,464 (95% CI: \$237,935 -\$672,992) per year in averted hospitalization and treatment costs as a result of increasing the provision of recommended care to eligible patients. Most of the cost averted (\$254,633 per year) is due to the estimated reduction in hospitalizations for pneumonia \[[@B14]\], acute chronic respiratory conditions \[[@B35]\], and cases of congestive heart failure \[[@B7]\]. The treatment costs saved of an estimate of one case of Spina Bifida equated to \$174,999. Costs averted due to hospitalizations and treatment equate to \$20,703 per practice, \$151,821 per facilitator, and \$4.93 per patient.
::: {#T9 .table-wrap}
Table 9
::: {.caption}
######
Savings from the provision of appropriate preventive care (1999 Dollars)
:::
**Condition** **Baseline** **Low** **High**
--------------------- -------------- -------------- --------------
Breast Cancer \$12,075.20 \$4,026.97 \$20,123.44
Influenza \$254,633.54 \$145,976.01 \$363,291.07
Neural Tube Defects \$174,999.39 \$99,997.21 \$250,001.56
Cervical cancer \$2,154.49 \$548.05 \$3,760.92
Lung Cancer \$122.27 (\$518.51) \$763.06
Heart Disease (\$808.13) (\$4,116.12) \$2,499.86
STD Treatment \$12,862.75 (\$5,044.91) \$30,770.42
Stroke (\$575.90) (\$2,933.28) \$1,781.48
TOTAL \$455,463.61 \$237,935.42 \$672,991.81
:::
::: {#T10 .table-wrap}
Table 10
::: {.caption}
######
Estimated cost savings
:::
--------------------------------- -------------- ---------------- ----------------
**Baseline** **Low** **High**
[Costs]{.underline}
Intervention Cost \$238,388 \$238,388 \$238,388
Manoeuvre Costs \$192,911.62 \$42,581.43 \$343,241.78
Total Costs \$431,299.62 \$280,969.43 \$581,629.78
[Savings]{.underline}
Inappropriate manoeuvre savings \$148,568.44 (\$175,897.93) \$473,034.80
Treatment cost savings \$455,463.61 \$237,935.42 \$672,991.81
Total Savings \$604,032.05 \$62,037.49 \$1,146,026.61
NET Savings \$172,732.44 (\$218,931.95) \$564,396.83
--------------------------------- -------------- ---------------- ----------------
:::
Table [10](#T10){ref-type="table"} shows that the cost of the program over one year combined with the increased costs associated with delivering more recommended preventive care is \$431,300. The costs averted from reducing inappropriate tests and the reduction in hospitalization and treatment are \$604,032. This is a net cost savings to the government of \$172,732 (95% CI: -\$218,932 -- \$564,397) per year as a result of the intervention with 22 practices and equates to \$1.87 per patient saved or \$3,321 per physician, an estimated return on the intervention and investment in additional manoeuvres of 40%. Based on a Consumer Price Index of \$1.11 the net cost savings to the government in 2003 dollars are \$191,733 per year or \$8,715 per practice, \$3,687 per physician, or \$2.08 per patient.
Sensitivity analysis
--------------------
The sensitivity analysis using Monte Carlo simulation revealed that the mean net savings in 1999 dollars to the government would be \$175,510 and that 90% of the expected net savings to the government would fall between \$-43,000 and \$394,000 (see Figure [1](#F1){ref-type="fig"}). Based upon 1,000 iterations of the cost model and the assumptions regarding input costs, it is unlikely that the government would not receive some return on its investment in outreach facilitation. Finally, Figure [2](#F2){ref-type="fig"} shows the relationship between each of the input variables to net savings. As would be expected net savings is very sensitive to PSA testing with the more men tested the less net savings accounting for 47.4% of the variance. Flu vaccination also impacts net savings in terms of pneumonia prevented and accounting for 18.4% of the variance in net savings. Due to the poor specificity of the PSA test and the non significant effect of the intervention to reduce PSA tests in eligible men, the cost associated with the PSA test varies considerably and its reduction is critical for attaining cost savings whereas the increase of PSA tests is negatively associated with savings. In contrast, PAP testing and mammography for 50 to 59 year old women in the cost model are not as negatively associated with net savings despite the increase in costs as more of these manoeuvres are performed.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Distribution of outcome for net savings
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Sensitivity analysis for net savings
:::
:::
Discussion
==========
The primary objective of this study was to determine if decreasing inappropriate preventive tests and increasing appropriate testing performance could offset the cost of an outreach facilitation intervention to improve preventive practice. The significant reduction in inappropriate testing and increase in appropriate testing resulted in net savings of \$191,733 per year in 2003 dollars to the government or a return on investment of 40%. The limitations of the cost analysis are typical to this type of economic evaluation \[[@B37]\]. They include:
• the perspective of the government necessitated not having included all possible costs in the model. For example, the cost for patient time, travel or patient discomfort and anxiety associated with the manoeuvre were not included;
• the total indirect costs associated with hospitalizations averted were not included;
• the estimate of the frequency of downstream events was based on a panel of experts;
• the estimate for screening for cervical cancer was based on a yearly screening rather than once every three years;
• downstream costs were estimated and included in the model for follow-up visits as a result of a false positive test only. Other possible downstream costs such as visits to other allied health professionals or consults to specialists were not included;
• the benefit of inappropriate screening tests for some patients and the associated cost savings have been ignored; and
• rates of delivery for preventive screening tests were from a randomized controlled trial in a HSO setting. Therefore, caution must be used when generalizing the potential cost savings to other settings.
Our cost estimates in this analysis are conservative since patient costs and other downstream costs were not included. Nonetheless, the analysis shows that all of the costs of the outreach facilitator intervention can be recouped as a result of having reduced inappropriate testing and increased appropriate testing for the manoeuvres under study. Similarly, McCowan et. al. were able to show that a facilitation intervention was able to improve primary care asthma management and that the cost savings to the health care system could completely offset the annual salary of one facilitator serving a large number of family physicians \[[@B28]\]. McCowan\'s study also included the costs associated with hospital admissions and secondary costs and involved the improvement in treatment of an acute illness and not prevention in primary care. Including the downstream costs associated with inappropriate tests averted for the outreach facilitator intervention has allowed for an additional 35% in estimated cost savings and the inclusion of costs averted associated with appropriate testing has completely offset the cost of outreach facilitation.
The successful outreach facilitator intervention described in this study was a very intensive intervention with each practice being visited an average of 33 times over 18 months. This compares to other successful trials such as Dietrich et. al. \[[@B23]\] where outreach facilitators visited only 3 times over a three-month period at an average of 120 minutes per visit and Hulscher et. al. \[[@B32]\] where facilitators visited practices an average of 25 times with an average duration of only 73 minutes. Both of these studies used outreach facilitation to improve preventive practice for a number of different maneuvres. Unfortunately, the cost of outreach facilitation was not included in these less intensive studies.
The cost of an outreach facilitator in our study per year was over \$4,497 per physician but resulted in an overall net savings to the government of \$3,289 per physician. In comparison, Cockburn et. al. tested an educational outreach facilitator intervention to improve physician smoking cessation counselling performance which cost \$A142 in 1992 per practitioner \[[@B27]\]. However, the facilitator only visited each physician twice at an average of 12 minutes per visit or \$A72 per visit and achieved very little in the way of improved outcomes. As a consequence, unsuccessful outreach facilitation was shown to be not cost-effective compared to other alternatives. In our intervention the cost per visit was \$590 and nothing after having adjusted for cost savings associated with a successful intervention. Our intervention was targeted at changing the entire practice and not just physician behaviour for a number of preventive measures, and as a result more time was spent on-site and more visits were required. More research is necessary to determine the most appropriate intensity of intervention for a given level of outcome.
The Cochrane Effective Practice and Organization of Care Group has compiled evidence that supports outreach visits combined with additional interventions as effective in improving professional practice and health outcomes \[[@B13]\]. Our study has demonstrated the effectiveness of outreach facilitation in improving overall preventive care performance. This is the first cost-consequences analysis of an outreach facilitation intervention that we are aware of and we have shown that the savings attributable to the reduction in inappropriate testing and increases in appropriate testing can offset all of the intervention cost and in fact result in a net savings to the government of approximately two dollars for every rostered patient. Further, the sensitivity analysis has revealed that the likelihood of net savings to the government is substantial despite variation in the inputs. However, it has also revealed that the appropriate primary preventive manoeuvres such as vaccination of the elderly contribute more to net savings to the government than some secondary preventive manoeuvres such as Pap testing and mammography. The net savings associated with primary prevention help to cover the net costs to the government associated with secondary prevention. As well, in the case of PSA testing even greater cost savings may be possible by significantly reducing this and other inappropriate tests and the consequent downstream costs.
Filak et. al. \[[@B52]\] calculated the lifetime charges of office visits, procedures, laboratory tests, and patient purchases required to comply with the US Preventive Services Task Force screening recommendations. They determined that the lifetime charges in 1999 US Dollars for all required preventive services ranged from \$5,432.60 to \$7,529.60 for men and from \$15,307.10 to \$18,525.10 for women. If physicians could deliver all the necessary preventive care to their patients, the costs over a lifetime are very reasonable in comparison to all of the costs associated with treating a stage I (\$14,000) or stage IV (\$64,000) cancer of the breast in a woman over 50 \[[@B53]\] or a case of pneumonia (\$35,700) caused by influenza in a person 65 years old \[[@B54]\]. The outreach facilitator intervention was effective in improving the uptake of preventive manoeuvres shown to be very cost-effective with the added benefit of reducing preventive manoeuvres that are less cost-effective. If the estimated 9,850 family physicians in Ontario received the benefit of outreach facilitation, the estimated savings to the government equate to \$36.3 million (2003 dollars). However, this study involved HSO physicians who may not be representative of all family physicians and research has demonstrated that the intervention does not work in chaotic practices \[[@B59]\].
Conclusion
==========
This paper has provided information on the cost of outreach facilitation and the potential for cost savings to the health system. The results can be considered an underestimate of the true potential cost savings given that not all the costs associated with inappropriate and appropriate preventive care were considered. Further, this was an efficacy trial and as a consequence there is potential for reducing the cost of the intervention through efficiency improvements such as increasing the number of practices per facilitator as well as savings in administration and training through economies of scale. In addition, additional savings to the health system may be possible through on-going outreach facilitation for chronic illness care.
There are no magic bullets to changing primary care practice patterns \[[@B9]\]. However, a facilitation approach that incorporates a number of intervention strategies tailored to the environment and needs of the practice holds promise. The literature has shown that outreach facilitation is one of the most effective means of improving the delivery of primary care preventive health services \[[@B13]\]. Further economic evaluations of outreach facilitation and other intervention alternatives are needed to assist in important public policy and administrative decision-making on getting preventive care guidelines into practice.
Competing interests
===================
The research project was funded by a grant from the Ontario Ministry of Health. All of the authors were involved in the research project to evaluate the effectiveness of outreach facilitation in improving the delivery of preventive health care guidelines. WH and JL are practicing family physicians, professors with the University of Ottawa and were investigators on the project. NB is a PhD candidate at the University of Waterloo and was employed as the project co-ordinator for the evaluation.
Authors\' contributions
=======================
WH is the Principal Investigator for the Closing the Loop on Prevention project. WH and JL conceived the economic evaluation study, participated in the study design, preparation of first drafts and critical revisions of the manuscript and contributed to all other aspects of the study. NB contributed to the economic study design, acquired the economic data, performed the economic analysis and interpretation, prepared, and revised the manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6963/5/20/prepub>
Supplementary Material
======================
::: {.caption}
###### Additional File 1
Mathematical Summary
:::
::: {.caption}
######
Click here for file
:::
|
PubMed Central
|
2024-06-05T03:55:55.268685
|
2005-3-9
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079830/",
"journal": "BMC Health Serv Res. 2005 Mar 9; 5:20",
"authors": [
{
"first": "William",
"last": "Hogg"
},
{
"first": "Neill",
"last": "Baskerville"
},
{
"first": "Jacques",
"last": "Lemelin"
}
]
}
|
PMC1079831
|
Background
==========
In many countries there is a substantial potential for savings if less expensive drugs, particularly thiazides, are prescribed rather than the more expensive drugs for hypertension \[[@B1]\]. The potential savings in the UK are £132 million (\$200 million) per year (£2.22 (\$3.36) per inhabitant; figures from the year 2000). The US and Norway could potentially save even more per inhabitant (£3.21 (\$4.86) and £3.55 (\$5.38) respectively, year 2000). An important reason for these differences in potential savings is that thiazides are used more in the UK than in the US and Norway. In this article we describe and attempt to explain international variation in prescribing patterns of antihypertensive drugs.
Methods
=======
We had access to sales figures for anihypertensive drugs for six countries (Canada, France, Germany, Norway, the UK and the US) for the year 2000. We also had survey-based information describing the diagnoses for which the drugs were being prescribed. This was relevant since antihypertensive drugs are also used for other indications, such as heart failure (e.g. ACE-inhibitors) and post-myocardial infarction (e.g. beta-blocking agents). The information was provided by IMS-Health.
The sales-figures were originally expressed as physical units (kg), which we transformed to defined daily dosages/1000 inhabitants/day \[[@B2]\]. The defined daily dose (DDD) is the assumed average dose used for a drug \[[@B3]\].
For each drug-class we estimated the total consumption by summarizing the consumption for each drug within a class. The total consumption for each class was then multiplied with the proportion of prescribing that was done specifically for hypertension. We estimated the consumption of the various drug classes for each country, and compared them.
The following drug-classes (and ATC-numbers) were included: alpha blocking agents (C02C A), thiazides (C03A, C03B og C03E), beta blocking agents (C07), calcium channel blockers (C08), ACE-inhibitors (C09A), ACE-inhibitors combined with a diuretic (C09B), angiotensin II antagonists (C09C), and angiotensin II antagonists combined with a diuretic (C09D).
We also obtained official sales statistics for antihypertensive drugs in the Nordic countries (Denmark, Finland, Iceland, Norway and Sweden) for 1999, and compared the patterns of consumption \[[@B4]\]. For these countries we did not adjust for the proportion of prescribing being made specifically for hypertension, as we did not have access to such information.
We circulated those results (figures [1](#F1){ref-type="fig"} and [2](#F2){ref-type="fig"}) to a convenience sample of one academic in each of the included countries and asked about possible reasons for inter-country variation in prescribing patterns. The results were also sent to the drug regulatory agency in each country. The recipients were asked to answer the following five questions:
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Consumption of drugs for the treatment of hypertension for six countries (percentage distribution within each country) \*.**\*Consumption as defined daily dosages/1000 inhabitants/day. Based on figures on sales (year 2000) and indications for prescribing from IMS-Health
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Consumption of antihypertensive drugs within the Nordic countries (percentage distribution within each country)\***\*Consumption as defined daily dosages/1000 inhabitants/day. Based on official sales statistics from 1999 \[4\]. No adjustment made for the relative proportion of prescribing being done for other indications than the treatment of hypertension
:::
:::
1\. Are there specific policies, rules or regulations in place in your country that may influence the choice of drug in the treatment of hypertension (e.g. pricing policies, treatment-protocols etc.)?
2\. What do you think are the main factors that influence the choice of drug in the treatment of hypertension in your country?
3\. In particular, why do physicians prescribe newer, expensive drugs, such as calcium channel blockers or angiontensin II receptor blockers?
4\. Similarly, why do physicians prescribe older, less expensive drugs, such as thiazides or beta-blockers?
5\. We have attached a graph illustrating the profile of sales of antihypertensive drugs for some countries, including yours. How would you explain your country\'s profile in comparison with the other countries?
Lastly, we contacted the medical directors of the British and Norwegian affiliates of four major drug companies with antihypertensive drugs, and conducted a semi-structured telephone interview exploring possible explanations for the differences between these two countries, which had the largest differences in prescribing patterns (See [additional file 1](#S1){ref-type="supplementary-material"}: Interview guide).
Results
=======
Calcium channel blockers and ACE-inhibitors are generally the most widely used drugs for the treatment of hypertension (figure [1](#F1){ref-type="fig"} and [Additional file 2](#S2){ref-type="supplementary-material"}: Sales of antihypertensive drugs, IMS). Apart from that, there is large variation among the countries with regards to the use of drugs for the treatment of hypertension.
In the UK thiazides account for 25% of consumption, while the corresponding figure for Norway is 6%. In Norway alpha-blocking agents account for 8% of consumption, which is more than twice the percentage found in any of the other countries. The use of combination drugs in the UK is strikingly low. The prescribing patterns also vary largely within the Nordic countries (figure [2](#F2){ref-type="fig"} and [Additional file 3](#S3){ref-type="supplementary-material"}: Sales of antihypertensive, Nordic countries).
The overall consumption of antihypertensive drugs is considerably higher in the US than in Canada, France, Norway or the UK (table [1](#T1){ref-type="table"}).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Consumption of antihypertensive drugs in six countries, based on sales figures provided by IMS-Health (year 2000)
:::
**Total consumption of antihypertensives (DDDs/1000 inhabitants/day)** **Consumption of antihypertensives for the treatment of hypertension (DDDs/1000 inhabitants/day)**
--------- ------------------------------------------------------------------------ ----------------------------------------------------------------------------------------------------
Canada 163.6 119.6
France 171.7 133.4
Germany 205.9 145.3
Norway 171.6 115.7
UK 170.6 105.8
US 225.5 165.3
:::
Possible explanations for inter-country variation
-------------------------------------------------
We received answers from researchers in all the ten countries and drug regulatory agencies in Finland, Norway and Sweden and from a Department of Health official in the UK.
There was nearly full consensus among our respondents that marketing by pharmaceutical companies is the main explanatory variable for prescribing choices, and particularly the driving force behind the prescribing of newer drugs, such as calcium channel blockers and angiotensin II receptor blockers.
There was some variation among the respondents regarding the policies, rules and regulations related to prescribing of antihypertensive drugs in the various countries, but few believed that this had a major impact on prescribing. The exceptions were the Finnish respondents, who believed that their national clinical guidelines program could have had an impact, and the UK Department of Health official who considered central guidance (\"especially NICE guidance\") a major factor.
The answers given to our question on why doctors prescribe older drugs was mixed. Some researchers pointed to price considerations (Denmark, Finland, Canada, US), and some made statements like \"Some physicians prescribe according to evidence-based principles, but they are probably a minority\" (Canada, France). There were several possible explanations for the differences in prescribing patterns between their own and the other countries:
\....\"it is MOST unusual for UK GPs to use combination preparations. It is regarded as poor practice.\" (British researcher)
\"The very low usage of thiazides probably reflects the views of the prominent specialists and researchers in this field in Norway\" \...\... \"Many of the experts are deeply involved in industry sponsored research and tend to favour new products and their own research.\" (Norwegian drug regulator)
\"Denmark\'s relatively large share of thiazides may to some extent be explained by factors like domestic production\"\....\"but may also reflect their better system for continuing drug education among their GPs (e.g. academic detailing). In Norway this education is almost entirely left to the pharmaceutical companies.\" (Norwegian researcher)
\"Beta-blockers have a long tradition in Sweden.\" (Swedish researcher)
\"I once asked an older cardiologist about the reason for this difference and the answer was that there was no major tradition for beta-blockers in Denmark\" (Danish researcher)
\"The most striking difference for Canada is the lack of combination use of thiazides with ACEI and ARBs. This is probably due to the fact that these combination drugs are not funded in many of the provinces. This leads to a seemingly greater use of thiazides and ACEIs alone.\" (Canadian researcher)
Seven of the eight drug company medical directors (including two alternates) agreed to be interviewed: four from Norway and three from the UK.
Differences in physicians\' attitudes were thought by all the directors to partly explain the differences in prescribing between the UK and Norway; e.g. that British doctors generally are conservative and slow implementers of new interventions while Norwegian physicians are \"early adopters\":
\"My perception is that we Norwegians are, in general, \"early adopters\" -- we are attracted to the new and \"hot\", be it kitchen equipment or other things.\" (N2)
\"UK is in general a conservative country when it comes to prescribing new drugs. New therapies are slowly taken up\" (UK1)
\"I think this \[low use of combination drugs\] reflects the general conservatism that is typical of the British.\" (N2)
\"Could be that the British doctors are more conservative -- it wouldn\'t be surprising knowing how conservative the British are, in general!\" (N4)
\"It is, perhaps, true that British doctors comply more with guidelines than their Norwegian colleagues. The British are perhaps more respectful towards authorities while the Norwegians are more individualistic in their attitudes.\" (N4)
\"It may be true that Norwegian tend to disregard clinical guidelines and rather adhere to their own convictions. \...\...It is typically Norwegian that each individual has his own personal opinion.\" (N2)
\"There is usually a slow uptake of newer drugs in the UK. New data typically doesn\'t impact immediately -- unless they\'re very strong.\" (UK2)
\"One new trial result is not enough for a UK-doctor -- again a consequence of the conservatism we have.\" (UK3)
The Norwegian directors also believed that conducting trials in general practice has an impact on prescribing and that a substantially higher number of physicians involved in such trials in Norway could explain differences in prescribing patterns:
\"Many trials are being run in Norway.\"\..... \"I believe that the hand-on experience doctors get from participating in the trials increases their awareness and influences practice.\" (N1)
\"I do believe that running a study like this is, by itself, of value from the marketing perspective.\" (N2)
\"Of course we conduct clinical trials with the aim of developing newer and better drugs, but a marketing effect is unavoidable, and is an added benefit for the company.\" (N2)
\"Yes, involvement by doctors in the running of clinical trials probably has an impact. In Norway a much higher proportion of GPs participate in trials compared to, for instance, the UK.\" (N2)
\"I don\'t think that is true \[that the differences can be explained by the higher number of trials that are conducted in Norway\]. I can only speak for our company, but we have run several trials in UK general practice -- often in collaboration with sites in Norway and the other Scandinavian countries. So this does not fit with my understanding.\" (UK1)
\"Norwegian physicians are in general rapid implementers of new therapies. A major explanation for this is probably the high number of large clinical trials that are run in Norway. Relative to our population the number of patients participating in clinical trials is about the double of what can be expected compared to other countries.\" (N3)
\"Norway is a key area for clinical drug trials, e.g. studies of organoprotective effects of antihypertensive drugs. This has been driving prescribing to the right in the graph \[i.e. towards the more expensive alternatives\]. Recently we have seen a slight increase in the use of ARBs, for the treatment of migraine. Again, an example of high quality trial results being rapidly taken up in clinical practice.\" (N3)
\"I don\'t think participating in trials will lead to a change in prescribing habit for a UK-doctor.\" \...\... \"We do \[run clinical trials where GPs participate\] \... but maybe we do it less than in Norway.\" (UK3)
British doctors were also believed to be more cost-aware, mainly due to local budgets which include drug expenditures, thus driving prescribing towards less costly alternatives:
\"In the UK they have given budget-responsibility to the local GPs, which could limit spending on drugs, for instance.\" (N1)
\"It\'s probably true that UK physicians are more cost-aware. One reason may be their tradition of local budgets. Actually, I think Norwegian doctors have become less cost-aware now than they were before. At least that\'s the impression I have from speaking to our representatives -- they tell me that the doctors put very little emphasis on price.\" (N2)
\"Cost considerations are also part of the picture. It is my impression that UK doctors are more cost sensitive. This has to do with the limited drug budgets for each practice, but I think UK doctors also are happy to prescribe thiazides and beta-blockers.\" (UK1)
\"Maybe the British physicians are more cost-aware. I think Norwegian doctors have regarded the drug-reimbursement system as a bottomless pit. They may be concerned about the costs for their patients, but not the societal costs.\" (N4)
\"I have mentioned drug formularies\... Also, primary care trusts have budgets they are responsible for, which includes spending on drugs.\" (UK2)
\"I don\'t think that UK GPs are particularly acquisition-cost aware, but the drugs that are included in their formularies have gone through a process of appraisal where cost-effectiveness is an important element.\" (UK2)
\"It could be that in Norway you focus more on the patient in front of you. The UK-system is very cost-conscious.\" (UK3)
Local influence from pharmaceutical advisors was also mentioned as an important force that \"pushes the use of thiazides -- for pure cost reasons\" (UK3). This \"local guidance\" was also seen as a reason why \"UK-doctors are hard work for industry to influence\" (UK3).
Views on the role of evidence among Norwegian and British physicians were contradictory. On the one hand the directors seemed to agree that there is a \"bias for evidence-based medicine\" (UK1) in the UK, but on the other hand the Norwegian directors claimed that physicians in Norway are very much focused on trial results and particularly hard endpoints, e.g. effects on mortality rates:
\"Our impression is that the doctors in Norway put great emphasis on documentation -- they look at the evidence. They want trials with hard end-points.\" (N1)
\"I think there is a particular focus on large studies with survival endpoints among Norwegian doctors.\" (N2)
\"I think Norwegian physicians are particularly interested in research and the results of clinical trials.\" (N2)
\"I do have the impression that Norwegian physicians are focused on hard endpoints, but if there is any difference compared to British doctors I don\'t know.\" (N4)
\"I think guidelines are important in explaining the UK prescribing pattern and there is a strong bias for evidence-based medicine. Thiazides and beta-blockers have been recommended as first line, and doctors have followed this recommendation.\" (UK1)
\"UK doctors may be more difficult to persuade, perhaps, with their strong focus on evidence. They demand trials that have shown the effectiveness of drugs on hard end points -- their clinical usefulness.\" (UK1)
\"Also in the UK some physicians want hard end-points -- especially the physicians in secondary care. Many GPs are more focused on controlling the blood pressure.\" (UK2)
\"Maybe Norway is less evidence-based and more patient-focused? \..... There is a fundamental belief in endpoints and outcomes -- there is a hang-up on this in the UK.\" (UK3)
The British directors agreed that the low use of combination drugs in the UK was due to therapeutic traditions largely seeded at medical school. One of them also believed that \"the general push for generic prescribing plays a role, since the combinations usually are branded drugs\" (UK3).
The high use of alpha-blockers in Norway was considered to be an artefact by some since current regulation only allows for the reimbursement of such drugs if they are used for hypertension, and not for benign prostate enlargement. However, one of the directors refuted this, citing a study conducted among general practitioners where only 5--10% of reimbursed prescriptions for alpha-blockers were made for individuals without hypertension.
Two Norwegian directors confirmed that the relationship between specialists and industry might be a factor:
\"Conflicts of interest of opinion leaders may be an issue\" \...\... \"Norway has a rather high number of opinion leaders in the cardiovascular area that have been active in collaboration with industry.\" (N2)
\"Their engagement with industry is important for the services they provide to their patients, e.g. the chance of accessing new therapies, more intense follow-up of patients etc. This is less so in the UK. For this reason industry may find it more difficult to convince physicians in the UK with their arguments.\" (N3)
\"To get opinion leader endorsement is important. They rarely go against established guidelines, so they generally support the use of thiazides. Also the opinion leaders put great importance on cost-effectiveness, while tolerability for instance is less emphasised. Regarding their links to industry I think they are cautious -- they have their integrity to maintain.\" (UK3)
\"One difference might by that Norwegian opinion leaders seem to be somewhat less keen on promoting a specific drug -- they worry about their credibility.\" (N2)
\"In fact, my guess is that marketing in general -- also the use of opinion leaders -- is more aggressive in the UK, if there is any difference at all.\" (N1)
One Norwegian director mentioned the role of patients: \"In the UK I think the doctor-patient relationship is more traditional, with the physician deciding for the patient, while here the patients are more involved in decision-making; and when given the choice they prefer the \'latest model\"\' (N1).
Financial incentives were mentioned by one UK director: \"In the UK there are local prescribing incentive schemes that reward generic prescribing, reward high use of drugs in accordance with guidelines, etc. The reward can be money, which goes to the clinic. The GPs are independent contractors -- running their own business.\" (UK3)
One director suggested that a higher threshold for initiating treatment among Norwegian physicians means that a higher proportion of those who are treated need more than one drug, and that this influences the overall prescribing pattern.
Discussion
==========
We have based our analysis on a cross-sectional data from one year. This may represent an incomplete picture of the prescribing in the ten countries since we are not able to detect prescribing trends, which may vary across countries. Another potential weakness of our analysis is that we have not taken epidemiological differences into account. For instance, a lower use of ACE-inhibitors in the US may be appropriate given the high number of Afro-Americans and the fact that these drugs are less effective in persons of African origin. Interestingly, however, the use of ACE-inhibitors in the US is particularly high.
There is considerable variation in prescribing patterns of antihypertensive drugs among the countries included in this study. The UK and Norway are at opposite ends of the spectrum.
The high use of alpha-blockers in countries such as Norway and the US is of concern since evidence in support of these drugs is particularly weak. The first major comparative trial of alpha blocking agents and other antihypertensives was stopped early due to the high rate of cardiovascular disease events among those allocated to the alpha-blocker \[[@B5]\]. Despite this, the sales of alpha-blockers in Norway have only gone down slightly (from 8.8 DDDs/1000 inhabitants/day in 2000 to 7.8 DDDs/1000 inhabitants/day in 2003) \[[@B3]\]. In the US the publication of the trial results appears to have had a greater impact on physicians\' prescribing habits \[[@B6]\].
Not only do Norwegian physicians tend to choose drugs from the most expensive drug-classes, they also tend to select the more expensive drugs within a class (table [2](#T2){ref-type="table"}). For example, metoprolol is the most sold beta-blocking agent in Norway and it is one of the most expensive drugs within its class \[[@B7]\]. Moreover, metoprolol is available from several manufacturers, and the most expensive version (Selo-Zok^®^) is the most sold \[[@B7]\]. This explains why beta-blocking agents have a higher average price than ACE-inhibitors in Norway (table [2](#T2){ref-type="table"}), despite the fact that most beta-blocking agents are much less expensive than ACE-inhibitors \[[@B7]\].
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Drug-costs for one year\'s treatment, in Norway (ranked according to price)\*
:::
**Drug cost\***
----------------------------------------- -----------------
Thiazide \$33 (£22)
ACE-inhibitor \$150 (£99)
Beta-blocking agent \$174 (£115)
Calcium channel blocker \$211 (£139)
Alpha blocking agent \$269 (£178)
ACE-inhibitor with thiazide \$270 (£178)
Angiotensin II antagonist \$288 (£190)
Angiotensin II antagonist with thiazide \$328 (£216)
\*Based on prices to consumer. The costs were calculated by dividing total sales of each drug class by consumption (year 2000). It is assumed that the average dose used equals the defined daily dose (DDD)
:::
If Norwegian physicians were to prescribe antihypertensive drugs similarly to Danish physicians, the annual expenditure would be reduced by £26 million (\$US 40 million), which is about 30% of current spending on these drugs in Norway. Interestingly, it was pointed out 20 years ago that the Norwegian drug expenditures for antihypertensives would be reduced by 36% if Norway were to adopt a Danish profile \[[@B8]\].
It is widely believed that marketing efforts from the pharmaceutical industry have a major impact on physicians\' prescribing \[[@B9]\]. Our informants agreed with this, but the directors we interviewed did not indicate that marketing strategies differ much from country to country. According to them companies base their marketing on global strategies that are adapted locally. However, it is possible that Norwegian physicians interact more frequently with drug company representatives, for instance, and that this may explain some of the variation in prescribing behaviour. We have collected data on the scope of outreach visits by industry representatives to general practitioners in Norway, but we have not found similar data for any of the other countries in this study. Thus, no comparison is possible.
Non-industry pharmaceutical advisors play an active role in trying to influence prescribing behaviour among UK primary care physicians. They use strategies such as locally developed guidelines, feedback on prescribing, and outreach visits. A similar system does not exist in Norway and this may be one important explanatory factor for the observed differences between the UK and Norway.
Our survey of researchers and drug regulators yielded few plausible explanations for the variation in prescribing patterns. It is unlikely that practice guidelines alone have had an important impact \[[@B10]\]. The higher use of thiazides in Denmark and the higher use of beta-blockers in Sweden may be related to domestic production of these agents in the respective countries, and the lack of reimbursement may help to explain the lower use of combination drugs in Canada. These explanations are not likely to explain other differences.
The medical directors from drug companies in Norway and the UK suggested that fund holding for general practitioners was an important reason for the higher use of thiazides in the UK. However, the findings from numerous evaluations addressing the effect of fund holding on prescribing do not necessarily support this belief \[[@B11]\].
The assertion that Norwegian physicians are less conscious about the societal costs of their prescribing compared to their British colleagues is difficult to validate. Jacoby and colleagues observed in their interviews with 56 general practitioners in the UK that there was considerable variation concerning the level of cost-consciousness \[[@B12]\].
There seemed to be a consensus among the Norwegian medical directors that physicians in their country are particularly focused on research findings, especially trials that measure effects in terms of hard endpoints. However, this does not fit with our impressions. For example, in recent debates over first line antihypertensive drugs metabolic effects (soft endpoints), not hard end points, have been at the centre of the debate. Medical directors from the UK also did not share the impression that British physicians show less interest in research findings or put less emphasis on hard endpoints than their Norwegian colleagues.
Statins are another group of drugs for which the national prescribing patterns vary greatly \[[@B13]\]. In their recent paper, Walley and colleagues speculated that a major reason why Norway has the highest use of statins in the region is \"the involvement of Norwegian doctors in seminal trials\" \[[@B13]\]. However, this does not explain the variation in consumption of antihypertensive drugs, since there are essentially no trials that provide evidence against the use of thiazides or beta-blockers. A better explanation may be that participation in trials as such, and not necessarily seminal trials, influences prescribing. Pharmaceutical companies may include marketing considerations when planning clinical trials, in particular those conducted after drug approval has been granted. Studies conducted after marketing-approval has been granted are sometimes labelled \"seeding trials\", as they may be designed to \"seed\" the use of a drug among physicians \[[@B14]-[@B16]\]. All the Norwegian directors in our study seemed to agree that trial-participation influences prescribing. Despite searches in several databases, we did not find research data to support or refute this belief. This is surprising considering concerns that have been raised over the impact of seed trials \[[@B15],[@B16]\]. Norwegian doctors are typically paid a bulk sum in the order of NOK 5000 -- 15000 (£425 -- £1275; \$800 -- \$2400) for every patient they enrol in clinical trials. The hourly fee for doctors participating as investigators in clinical trials in Britain is set at £193.50 (\$360), according to an agreement made by the British Medical Association and the Association of the British Pharmaceutical Industry. This fee includes all relevant overheads (Richard Tiner, personal communication).
For trial-participation to be an effective marketing strategy it is necessary to recruit as many physicians as possible. Also, a double blind trial is not the best design if the idea is for the physicians to obtain \"provider-experience\" with a specific drug \[[@B14]\]. In fact, many of the recent trials of antihypertensive drugs have recruited a very large number of physicians, and a non-blinded design is common. National registers of ongoing clinical trials, which could provide data regarding the extent of seed trials and phase III randomised trials, are generally either non-existent or not accessible.
We are not aware of data that document that physicians in Norway have a higher threshold for initiating treatment than in the UK. The fact that the consumption of drugs for hypertension is somewhat higher in Norway suggests the opposite (table [1](#T1){ref-type="table"}).
There was a common belief among the medical directors that UK physicians are more conservative than their Norwegian colleagues, and it was also suggested that this mirrors differences in attitudes that exist in general between the British and Norwegian societies. We have not identified research findings that support or challenge this hypothesis. However, Jacoby et al did observe that the British general practitioners they interviewed \"consistently described themselves as cautious and conservative\" \[[@B12]\].
Two additional hypotheses to explain inter-country variation were suggested during the review process of this paper: (1) the use of drug-samples and (2) educational interventions. There is evidence suggesting that both may be effective in influencing prescribing of drugs \[[@B17],[@B18]\]. However, we do not have information about the relative use of such interventions in the countries we have studied.
Conclusion
==========
The fourfold contrasts in use of thiazide and alphablockers in Norway versus the UK suggests that relative risks of potentially inappropriate prescriptions to Norwegian hypertensive patients are in the vicinity of 4. Two compelling explanations for the observed variation that warrant further investigation are the promotion of less-expensive drugs by pharmaceutical advisors in UK and the promotion of more expensive drugs through seeding trials in Norway.
Competing interests
===================
Both authors are employed by the Norwegian government, which has substantial interest in containing the cost of health care.
Authors\' contributions
=======================
AF conceived the study, carried out the survey and interviews, prepared the first manuscript and contributed to all other aspects of the study. ADO contributed to the design of the study, to the interpretation of data, and made critical revisions to the manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6963/5/21/prepub>
Supplementary Material
======================
::: {.caption}
###### Additional File 1
List of questions used in semi-structured telephone interviews with medical directors, or their alternates, of pharmaceutical companies.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 2
Sales figures for each drug-class, and proportions of drugs prescribed for hypertension, presented country wise (Canada. France, Germany, UK, USA, Norway). Based on IMS-data for the year 2000.
:::
::: {.caption}
######
Click here for file
:::
::: {.caption}
###### Additional File 3
Sales figures for each drug class, presented country-wise (Denmark, Finland, Iceland, Sweden, Norway). Based on official sales statistics for the year 1999.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
We thank all of the individuals who participated in our survey and interviews, including: Bjørn Beerman, Neil Brickel, Martin Eccles, Curt D. Furberg, Francois Gueyffier, Sigurdur Helgason, Paul Hjemdahl, Tessa Ing, Tahir Khan, Jan Peter Kösters, Henrik Lund, Steinar Madsen, Marjukka Mäkelä, Pirkko Paakkari, Atle Skattebøl, Jørund Straand, Heidrun Sturm Karen Marie Ulshagen, James M. Wright, and John Young. We also thank Signe Flottorp and Morten Aaserud for important contributions, Elisabeth Eriksen and Marit Rønning for their assistance with the use of the Norwegian Prescription Database, and Peter Stephens from IMS-Health. The Norwegian Health Services Research Centre, where both authors are employed, funded the study. Finally, we thank the peer-reviewers, Joel Lexchin and Malcolm Maclure, for valuable comments and criticism.
|
PubMed Central
|
2024-06-05T03:55:55.274760
|
2005-3-11
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079831/",
"journal": "BMC Health Serv Res. 2005 Mar 11; 5:21",
"authors": [
{
"first": "Atle",
"last": "Fretheim"
},
{
"first": "Andrew D",
"last": "Oxman"
}
]
}
|
PMC1079832
|
Background
==========
Patient access to care within a certain time is an important performance indicator of health systems\[[@B1],[@B2]\]. In publicly funded health care, wait lists are commonly used to manage access to elective procedures raising concerns about delaying necessary treatment\[[@B3],[@B4]\]. In patients with coronary artery disease (CAD) requiring coronary artery bypass graft (CABG) surgery, delaying the operation may lead to deterioration in the patient\'s condition, worsening of clinical outcomes and increased risk of death\[[@B5],[@B6]\]. Queuing CAD patients according to urgency of treatment is generally perceived as a method of facilitating access to care within clinically appropriate time\[[@B6]\].
In the Canadian province of British Columbia (BC), all medically necessary services are publicly funded\[[@B7]\]. Concerned with growing wait lists for cardiac surgery in the mid-1990s the provincial government started providing supplementary funding to increase the number of CABG operations by 15% annually starting in 1998\[[@B8]\]. Although aimed at improving access, it is not known whether these measures changed the time patients spent on CABG wait lists.
Previous studies showed inconsistent results regarding the impact of supplementary funding on time to surgery in existing hospitals in publicly funded health systems\[[@B9]\]. It has been suggested that the effect may vary according to the scope and the term of funding commitment, such as single hospital versus all the hospitals in a region, and one time versus on-going increases\[[@B10]\]. Although access to surgery from wait lists depends on the assigned priority, there is little information on the impact of supplementary funding across different priority groups.
In this paper we compare the number of weeks between being registered for CABG and undergoing the operation for equal proportions of patients registered in different years before and after the provincial government started providing supplementary funding. In the study period, patients were prioritized according to the established guidelines to expedite access to surgery if they were considered at greater risk of deterioration or death. The specific research questions were: 1) did the period of registration have an effect on the time patients spent on wait lists for CABG? 2) was the period effect similar across priority groups?
We use all relevant records from the provincial population-based registry of CAD patients identified as needing bypass surgery. Primary comparisons are done across synthetic cohorts of patients defined by two-year periods of registration on the wait lists: 1991--92, 1993--94, 1995--96, 1997--98, or 1999--2000.
Methods
=======
Data sources
------------
The provincial Cardiac Surgery Registry, a part of BC Cardiac Registries, was created in 1990 to collect data for reporting, planning and research purposes of participating surgeons and hospitals, and the provincial Ministry of Health\[[@B11]\]. The Registry prospectively captures the occurrence and timing of registration, surgery, or removal from the wait lists without surgery, for all patients accepted for cardiac surgery procedures in the four hospitals delivering all adult open-heart surgery services to four million residents of BC.
Between 1991 and 2000, from 15 to 20 cardiac surgeons were performing bypass surgery in BC, with less than 30% turnover. Although cardiac surgeons manage their wait lists independently, they all routinely provide information to the Registry entry modules: surgery registration, operative report, wait-list reconciliation, and discharge summary. When accepting patients on their wait lists, the surgeons document the indication for the procedure as well as the priority for treatment using common criteria (see below).
Once the operation is completed, the operative report containing the procedure and clinical data is entered in the Registry. Patients are removed from the wait lists after undergoing the operation or for other reasons: if they died, declined the operation, accepted surgery from another surgeon, moved away, or switched to medical management. The crude agreement between the Registry and hospital charts for ten demographic and clinical data elements has been estimated at 86%\[[@B11]\].
We deterministically linked the Registry records to administrative databases storing records of all hospital episodes in BC\[[@B12]\]. These records include the dates of admission, procedure and discharge, as well as diagnoses at discharge\[[@B13]\]. These data were used to corroborate the service dates and to identify coexisting medical conditions\[[@B14]\].
Patients
--------
If angioplasty is not indicated when the cardiologist evaluates the arterial lesions on the coronary angiogram, then a cardiac surgeon is consulted to assess the patients\' suitability for CABG. Patients are transferred to an in-patient ward directly from the catheterization laboratory if expedited assessment is necessary. If deemed suitable, these patients wait for CABG in hospital without registration on a wait list. Alternatively, a consultation with the surgeon can be scheduled at a later date. Surgeons register on their wait lists patients who need CABG and for whom the operation can be safely delayed. As in-patients were not added to wait lists, they were not included in analyses of wait-list times.
There were 9,366 records of registration for isolated CABG added to the Registry between January 1991 and December 2000. We excluded 135 records of patients who were: emergency cases (30), removed on the registration date (101), and had missing operating room reports (4). All remaining 9,231 records had either the surgery date or the date and reason of removal from the list without surgery. We restricted the analyses to the first 52 weeks after registration so that 475 (5%) patients remaining on the lists at 12 months were censored. Of those, 167 eventually underwent surgery; seven died; 78 received medical treatment; 104 declined surgery; 17 were transferred to another surgeon or hospital; and 102 were removed for other reasons.
Priority groups
---------------
When assigning priority, all cardiac surgeons in BC apply common guidelines developed in 1990 \[see [Additional file 1](#S1){ref-type="supplementary-material"}\]. Using the location and degree of affected coronary anatomy and symptoms, the guidelines help to: identify patients for whom CABG can increase survival or improve quality of life\[[@B15]\]; classify patients according to urgency of treatment; and assign a maximum recommended waiting time (MRWT). Patients are assigned priority 1 if they require CABG urgently (eg, left main coronary artery stenosis greater than 70%, MRWT three days); priority 2 if there is moderate urgency (eg persistent unstable angina, MRWT six weeks); or priority 3 if there is less urgency (eg intractable chronic angina, MRWT 12 weeks). These guidelines did not undergo any major revisions through the entire period under study.
Comorbidity
-----------
Using the administrative data, coexisting medical conditions were identified using all primary and secondary discharge diagnoses recorded in all hospital discharge abstracts within one year prior to registration\[[@B13]\]. This time frame was chosen in order to capture the presence of chronic diseases that could have affected the waiting time\[[@B14]\].
For each patient, we identified the presence of major and minor comorbid medical conditions present at registration.
Statistical methods
-------------------
Waiting times were analyzed as prospective observations beginning at the time of registration. Each subject had a wait-list time calculated in calendar weeks from registration to surgery or removal for other reasons. The cumulative probability of undergoing surgery as a function of wait-list time was estimated using the Kaplan-Meier method\[[@B16]\]. Patients removed from the list for reasons other than surgery were treated as censored observations.
Primary comparisons were done across synthetic cohorts of patients defined by two-year periods of registration on the wait lists. Differences in the distributions of wait-list times across cohorts were examined using the log rank-test\[[@B17]\]. The average weekly surgery rate was calculated by dividing the number of operations by the total number of patient-weeks on the list. The effect size for each registration period was estimated by hazard ratios for surgery derived from a Cox proportional hazards model\[[@B18]\]. Hazard ratios (HR) associated with registration periods evaluated the conditional weekly probability of undergoing CABG relative to the 1991--92 period. The priority groups and the presence of comorbidity at registration were included as independent variables in the Cox model to estimate adjusted effects. Age and sex were entered into the regression models as strata variables to avoid the proportionality assumption on these factors while using the proportional hazards model.
The Clinical Research Ethics Board of the University of British Columbia approved the study protocol.
Results
=======
In BC in the 1990s, 9,231 patients were registered on wait lists for CABG and spent a total of 137,126 person-weeks waiting. Over the same period, 9,433 patients underwent isolated CABG without registration on wait lists. The most prevalent groups at registration were men (82%), those without major comorbidities (52%), those registered in priority group 2 (70%), patients aged 60--69 (38%) and 70--79 (30%) years, and those registered in 1995--96 (22%), Table [1](#T1){ref-type="table"}. The proportion of patients registered in priority group 1 was lowest in 1999--2000 and highest in the 1995--96 cohort, Table [2](#T2){ref-type="table"}. The opposite pattern was observed in priority group 3. Of 8,756 patients who left the lists within 52 weeks: 7,991 underwent surgery; 90 died while waiting; 176 received medical treatments; 188 declined surgery; and 311 were removed due to other reasons.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Characteristics of 9,231 subjects registered for isolated coronary artery bypass surgery in British Columbia, 1991--2000.
:::
**Characteristic** **N** **(%)**
---------------------------------------------------- ------- ---------
Age group (y)
\<50 732 (7.9)
50--59 2005 (21.7)
60--69 3530 (38.2)
70--79 2770 (30.0)
≥ 80 194 (2.1)
Sex
Women 1634 (17.7)
Men 7597 (82.3)
Urgency at registration
Priority 1 659 (7.1)
Priority 2 6496 (70.4)
Priority 3 1963 (21.3)
Unknown 113 (1.2)
Major comorbidity at registration
None 4769 (51.7)
Minor comorbidity 2450 (26.5)
CHF, diabetes, COPD, rheumatoid arthritis, cancer 2012 (21.8)
Registration period
1991--1992 1724 (18.7)
1993--1994 1889 (20.5)
1995--1996 2010 (21.8)
1997--1998 1888 (20.5)
1999--2000 1720 (18.6)
Abbreviations: CHF -- congestive heart failure COPD -- chronic obstructive pulmonary disease
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Distribution of subjects registered for isolated coronary artery bypass surgery in British Columbia, 1991--2000, by priority group and registration period
:::
**Registration** **Priority 1** **Priority 2** **Priority 3**
------------------ ---------------- ---------------- ---------------- --------- ------- ---------
**period** **N** **(%)** **N** **(%)** **N** **(%)**
1991--1992 116 (6.7) 1221 (70.8) 334 (19.4)
1993--1994 110 (5.8) 1381 (73.1) 388 (20.5)
1995--1996 249 (12.4) 1363 (67.8) 374 (18.6)
1997--1998 117 (6.2) 1327 (70.3) 428 (22.7)
1999--2000 67 (3.9) 1204 (70.0) 439 (25.5)
Note: Excludes 113 subjects with unknown priority
:::
In all registration cohorts combined, the average weekly number of operations was 5.8 (95% confidence interval 5.7--6.0) per 100 patients listed, the median time on the list was 11 weeks (25th percentile 5 weeks; 75th percentile 22 weeks), and the probability of undergoing surgery after 26 weeks on the list, twice the MRWT for priority group 3, was 20%.
As expected, there were significant differences among priority groups, with larger proportions undergoing CABG at every week among more urgent patients (log rank test = 1611.9, P \< 0.0001). The average weekly number of operations per 100 patients on the list differed from 20.6 (19.0--22.2) in group 1 to 6.7 (6.5--6.8) in group 2 to 3.3 (3.1--3.4) in group 3. However, considerable variation in wait-list times was observed within each priority group. For instance, although half of group 1 underwent surgery within two weeks and 90% underwent surgery by 12 weeks, the remaining 10% waited another 1 to 32 weeks (total 13 to 44 weeks).
This can be seen in Figure [1](#F1){ref-type="fig"}, which shows access probabilities for CABG in each priority group. The abscissa shows the number of weeks on the waiting list and the ordinate shows the probability of undergoing operation by that week. Higher probabilities correspond to shorter wait list times. While all patients were removed at 52 weeks, for graphical simplicity we show the first 36 weeks. Access probabilities in priority 3 (blue) were systematically lower indicating longer wait list times than among priority 2 (red line) or priority 1 (green line).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Estimated probabilities of undergoing isolated CABG within a certain time after registration on wait lists, by priority group.
:::
:::
Access to surgery by registration period
----------------------------------------
The differences in the proportion of patients undergoing CABG were significant across registration periods (log rank test = 97.3, P \< 0.0001), with longer wait-list times for those registered between 1995 and 1998, Figure [2](#F2){ref-type="fig"}.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Estimated probabilities of undergoing isolated CABG within a certain time after registration on wait lists, by registration period.
:::
:::
Table [3](#T3){ref-type="table"} shows the number of weeks required for a specified proportion of patients to undergo the operation across registration periods. Wait-list times in 1995--96 were such that 10%, 25%, 50%, and 75% patients underwent surgery within 1, 6, 15, and 26 weeks, respectively, whereas half of the 1991--92 cohort underwent surgery within 9 weeks, and 75% did so within 19 weeks. Comparing the 1995--96, 1997--98 and 1999--2000 cohorts we observed a compression in access to surgery, i.e., reduction in the length of wait-list interval required for a specified proportion to undergo the operation. As measured by the difference between 90^th^and 50^th^percentiles of the wait time distributions, 40% of the 1995--96 cohort underwent surgery within 33 weeks following the median time, while it took 29 weeks for the 1999--2000 cohort.
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Percentiles of wait-list time (weeks) for subjects registered for isolated coronary artery bypass surgery in British Columbia 1991--2000 by registration period
:::
**Registration period** **Percentile**
------------------------- ---------------- --- ---- ---- ----
1991--1992 1 3 9 19 44
1993--1994 2 4 9 18 45
1995--1996 1 6 15 26 48
1997--1998 2 6 14 25 43
1999--2000 3 6 10 19 39
All periods 2 5 11 22 44
Note: probability of undergoing outpatient surgery within 26 weeks of registration is 0.804
:::
While the median wait-list time was 11 weeks (the MRWT of priority group 3) in all cohorts combined, 15% of the 1991--92 and 1993--94 cohorts, 22% of the 1995--96 cohort, 19% of the 1997--98 cohort, and 14% of the 1999--2000 cohort experienced an excessive wait, defined as longer than 26 weeks (data not shown). The average weekly number of operations per 100 patients listed varied from 6.5 (6.2--6.9) in the 1991--92 cohort to 5.1 (4.8--5.3) in the 1995--96 cohort to 6.2 (5.9--6.6) in the 1999--2000 cohort, Table [4](#T4){ref-type="table"} (fourth column). Corresponding hazard ratios and 95% confidence intervals (CI) are shown in Table [4](#T4){ref-type="table"}, columns 6 and 7. Relative to the 1991--92 cohort, the conditional weekly probabilities of undergoing surgery were 30% lower among 1995--96 patients, HR = 0.70 (0.65--0.76), and 23% lower in 1997--98 patients, HR = 0.77 (0.71--0.83), after adjusting for priority, comorbidity, age and sex. There were no differences between periods 1991--92 and 1999--2000, HR = 0.94 (0.88--1.02).
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Average weekly rate of undergoing the operation from wait list for isolated coronary artery bypass surgery in British Columbia 1991--2000 and adjusted rate ratios by registration period
:::
**Registration period** **Number of operations** **Total wait time, weeks** **Crude Rate, per 100** **SE** **Hazard ratio** **95% CI**\*
------------------------- -------------------------- ---------------------------- ------------------------- -------- ------------------ --------------
1991--1992 1504 23047 6.5 0.2 1.00 referent
1993--1994 1646 25480 6.5 0.2 1.00 0.93, 1.08
1995--1996 1727 34186 5.1 0.1 0.70 0.65, 0.76
1997--1998 1613 30384 5.3 0.1 0.77 0.71, 0.83
1999--2000 1501 24029 6.2 0.2 0.94 0.88, 1.02
All periods 7991 137126 5.8 0.1 \- \-
Abbreviations: SE = standard error; CI = confidence interval
\*adjusted for priority group and comorbidity; stratified by age and sex
\*\*0 patients were on the wait list on December 31 2001
:::
Access to surgery by registration period within priority groups
---------------------------------------------------------------
In each priority group, the proportions of patients undergoing CABG at each week on the wait-list was lower among those registered in 1995--96 compared to 1991--92 as measured by log-rank tests (priority 1: chi-square = 5.6, P = 0.0183, 1 df; priority 2: chi-square = 58.2, P \< 0.0001, 1 df; priority 3: chi-square = 20.5, P \< 0.0001, 1 df). By 1999--2000, the pattern of change was different between priority groups.
In priority group 1, the average weekly number of operations per 100 patients listed declined from 42.4 (34.6--50.2) in the 1991--92 cohort to 20.3 (17.7--22.9) in the 1995--96 cohort to 12.2 (9.2--15.3) in the 1999--2000 cohort (data not shown). Corresponding HRs and 95% CIs are shown in Table [5](#T5){ref-type="table"}, columns 2 and 3. The conditional weekly probabilities of undergoing surgery were 34% lower for the 1995--96 cohort, HR = 0.66 (0.50--0.87), and 53% lower for the 1999--2000 cohort, HR = 0.47 (0.33--0.68), relative to 1991--92. There was a difference in the distribution of wait-list times between 1995--96 and 1999--2000 cohorts (chi-square = 9.9, P = 0.0017, 1 df).
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
Access to surgery by registration period and priority group for subjects registered for isolated coronary artery bypass surgery in British Columbia 1991--2000, as measured by adjusted hazard ratios\*
:::
**Registration period** **Priority 1** **Priority 2** **Priority 3**
------------------------- ---------------- ---------------- ---------------- -------------- ------ --------------
1991--1992 1.00 referent 1.00 referent 1.00 referent
1993--1994 0.79 (0.57, 1.09) 1.10 (1.01, 1.20) 0.78 (0.65, 0.92)
1995--1996 0.66 (0.50, 0.87) 0.71 (0.65, 0.78) 0.69 (0.58, 0.82)
1997--1998 0.49 (0.36, 0.67) 0.82 (0.75, 0.89) 0.75 (0.63, 0.88)
1999--2000 0.47 (0.33, 0.68) 0.99 (0.90, 1.08) 1.07 (0.90, 1.26)
Abbreviations: HR = hazard ratio; CI = confidence interval
\*adjusted for priority group and comorbidity; stratified by age and sex
:::
In priority group 2, the average weekly number of operations per 100 patients listed varied from 7.3 (6.8--7.7) in the 1991--92 cohort to 5.3 (5.0--5.6) in the 1995--96 cohort to 7.3 (6.9--7.7) in the 1999--2000 cohort. The adjusted HR in 1999--2000 was 0.99 (0.90--1.08) relative to 1991--92 (Table [5](#T5){ref-type="table"}, columns 4 and 5). There was no difference in the distribution of wait-list times between 1991--92 and 1999--2000 cohorts (chi-square = 0.5, P = 0.5, 1 df).
In priority group 3, the average weekly number of operations per 100 patients listed changed from 3.9 (3.4--4.3) in 1991--92 to 2.8 (2.4--3.1) in 1995--96 to 4.0 (3.6--4.5) in 1999--2000. The adjusted HR associated with the 1999--2000 registration period was 1.07 (0.90--1.26) relative to 1991--92 (Table [5](#T5){ref-type="table"}, columns 6 and 7). There was no difference between the between 1991--92 and 1999--2000 cohorts (chi-square= 0.6, P = 0.4, 1 df).
Discussion
==========
In this paper we studied the amount of time that patients with CAD spent on CABG wait lists before and after the provincial government started providing supplementary funding to increase the annual number of CABG operations. We sought to determine whether the period of registration had an effect on the wait-list time and whether the period effect was similar across priority groups. Using the population-based registry, we compared the number of weeks from registration to surgery for equal proportions of patients across different registration periods. In these comparisons, we accounted for the priority mix at registration. We used prospective follow-up of all patients registered to avoid biases inherent in wait-list statistics based on patients undergoing the procedure only\[[@B19]\].
We found that the registration period had an effect on the amount of time that patients spent awaiting CABG in BC in the 1990s. Wait-list times in the 1995--96 cohort were such that 50% and 75% patients underwent surgery within 15 and 26 weeks, respectively, whereas one-half of the 1991--92 cohort underwent surgery within nine weeks and three quarters did so within 19 weeks. This trend was reversed later, such that the 1999--2000 patients waited no longer than did their 1991--92 counterparts. Relative to the 1991--92 cohort, the conditional weekly probabilities of undergoing surgery were 30% lower in 1995--96 patients, and 23% lower in 1997--98 patients, while there were no differences between periods 1991--92 and 1999--2000.
We also found that the effect of registration period was different across priority groups. In priority group 1, the wait-list time increased by the middle of the decade and increased even further by the end of the decade. This may reflect changes in queuing patients with more severe CAD including lessened concern about safety of delaying patients with left main stenosis\[[@B20]\] as well as increasing use of angioplasty to treat patients who would have formerly been treated surgically\[[@B21]\]. In priority groups 2 and 3, the wait-list time also increased by the middle of the decade. In contrast to priority group 1, however, the wait-list time in groups 2 and 3 decreased later, such that there were no differences between the 1999--2000 and 1991--92 cohorts.
Studies examining access to elective care in Canada and elsewhere often report median or mean times \[[@B22]-[@B25]\]. We found that reporting the probability of undergoing CABG as a function of wait-list time helps overcome some limitations of using single-value statistics in understanding differences between periods \[[@B26]-[@B29]\]. For instance, we were able to conclude that not only did changes in waits reduce the median delay from 15 to 10 weeks in the 1995--96 and 1999--2000 cohorts, respectively, but also provided 20% compression in access for 40% patients staying on the lists longer than the median time. Studying the distributions of wait-list times, we also were able to compare the conditional weekly probability of undergoing CABG across registration periods while adjusting for priority, comorbidity, age and sex.
The lack of information on hospitals or surgeons could be a limitation of this study as we were not able to adjust for the volume of CABG between the four tertiary care hospitals where the operation was performed or for the wait lists between cardiac surgeons.
Conclusion
==========
Our results provide evidence for a significant reduction in wait-list time after supplementary funding was provided on an annual basis to tertiary care hospitals within a single publicly funded health system. While system-level factors such as changes in the organization or delivery of services may have affected the wait-list time, one plausible reason for the observed reduction was that the hospitals had capacity to increase the number of operations. Compared to 1995--96, there was a 12% increase (from 3,696 to 4,174) in the total number of CABG operations in 1999--2000, Table [6](#T6){ref-type="table"}. Also, between 1995--96 and 1999--2000, there was a 13% decrease (from 54% to 41%) in the proportion of patients accessing the operation through wait lists, indicating that supplementary funding was used to provide more operations without delay.
::: {#T6 .table-wrap}
Table 6
::: {.caption}
######
Distributions of patients who were registered on wait lists or operated without delay in British Columbia 1991--2000, by registration period
:::
**Registration Period** **Patients identified as needing CABG**
------------------------- ----------------------------------------- -------- ------ --------
1991--1992 1724 (49.3) 1770 (50.7)
1993--1994 1889 (55.3) 1526 (44.7)
1995--1996 2010 (54.4) 1686 (45.6)
1997--1998 1888 (48.6) 1997 (51.4)
1999--2000 1720 (41.2) 2454 (58.8)
:::
The relatively short time frame following the funding increase is a limitation of our study. As discussed elsewhere, supplementary funding may not result in shortening wait lists if hospitals function near full capacity \[[@B30]\], or, if it is expected that funding will be withdrawn after wait lists are reduced\[[@B10]\]. Reducing wait lists may require investing in new health services facilities. In Denmark, rates of open-heart surgery increased by 70% and the median waiting times declined by half since 1994 when additional capacity for cardiac surgical care was established by increasing the number of operating theatres, equipment and personnel \[30\]. On-going study of wait-list times for CABG in BC will help determine the permanence of the impact of supplementary funding.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
ARL conceived and designed the study, acquired the data, interpreted the results, and drafted the manuscript. BGS conceived and designed the study, analysed the data, interpreted the results, and drafted the manuscript. RH participated in the design of the study, helped acquire the data, and interpreted the results. MK helped acquire the data, and interpreted the results. JMF participated in the design of the study and interpreted the results. MTS participated in the design of the study and interpreted the results. All authors read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6963/5/22/prepub>
Supplementary Material
======================
::: {.caption}
###### Additional File 1
The Microsoft^®^Word 2002 file \"BC consensus guidelines for CABG priority.doc\" shows the guidelines used by British Columbian cardiac surgeons for assigning priority to patients registered for coronary artery bypass grafting.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
The authors gratefully acknowledge the contributions of the following individuals: Rita Sobolyeva, Lisa Kuramoto, Laurie Kilburn, Christopher Buller, Min Gao, and Gordon Pate.
The following cardiac surgeons are contributors to the BCCR Surgical Research Committee: Drs. James Abel, Richard Brownlee, Larry Burr, Anson Cheung, James Dutton, Guy Fradet, Virginia Gudas, Robert Hayden, Eric Jamieson, Michael Janusz, Shahzad Karim, Tim Latham, Jacques LeBlanc, Sam Lichtenstein, Hilton Ling, John Ofiesh, Michael Perchinsky, Peter Skarsgard and Frank Tyers
This study received financial support from the: St Paul\'s Hospital Foundation (ARL), Vancouver Coastal Health Research Institute (BGS, JMF), Michael Smith Foundation for Health Research (ARL), Canada Foundation for Innovation (ARL, BGS), and Canada Research Chairs program (BGS). None of the sponsors had any role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or the decision to submit the paper for publication.
|
PubMed Central
|
2024-06-05T03:55:55.278324
|
2005-3-14
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079832/",
"journal": "BMC Health Serv Res. 2005 Mar 14; 5:22",
"authors": [
{
"first": "Adrian R",
"last": "Levy"
},
{
"first": "Boris G",
"last": "Sobolev"
},
{
"first": "Robert",
"last": "Hayden"
},
{
"first": "Michael",
"last": "Kiely"
},
{
"first": "J Mark",
"last": "FitzGerald"
},
{
"first": "Martin T",
"last": "Schechter"
}
]
}
|
PMC1079833
|
Background
==========
Medicare, the national health insurance program for the elderly in the U.S., consists of various plans or \"parts \[[@B1]\]\". Part A covers hospitalization costs and has no premium, but limits and deductibles apply. Part B covers outpatient care and requires a small monthly fee as well as a deductible. Limited prescription drug coverage will be added in 2006, but medications, co-payments, and deductible costs presently create a large and growing private market for Medicare supplemental insurance \[[@B2]\]. Retirees without access to supplemental insurance as a work-related benefit may choose to purchase Medicare supplemental insurance out-of-pocket, a phenomenon that may increase in the near future \[[@B3]\]. However, about 15% of elderly persons forego private supplemental insurance policies, a number that may increase as the cost of policies increase.
There is evidence that elderly persons who lack private supplemental insurance may be at increased risk of morbidity and mortality \[[@B4],[@B5]\]. Approximately 35% of the near poor elderly lacking supplemental insurance either skip doses or do not fill their prescriptions at all because they cannot afford the medications they are prescribed. Medicare supplemental insurance may improve access to care by reducing co-payments and medication costs, thereby reducing barriers to timely care for acute but treatable illnesses, chronic disease care, as well as preventive services. Among younger cohorts, co-payments have been shown to reduce utilization and possibly increase mortality \[[@B6]-[@B9]\]. Though improved access to medical care appears to confer health benefits, there is considerable uncertainty as to whether these benefits are worth their cost.
For the government, sensible decision-making requires an understanding of the health benefits associated with purchasing additional services. Much of the debate about the Medicare Modernization Act of 2003 has focused on its overall cost in the face of large deficiencies in coverage \[[@B2],[@B10]\]. With more employers likely to withdraw from carrying retiree costs past the age of Medicare eligibility, and the costs of insurance spiraling, greater numbers of older Americans are likely to be underinsured. Despite a rapidly growing budget, policymakers will likely be asked to consider improving the package that Medicare offers.
Given the uncertainty surrounding the efficiency of Medicare supplemental insurance however, policymakers have little rational basis upon which to design future Medicare policy changes. Information on the cost-effectiveness of the additional medical goods and services provided by a supplemental policy to Medicare would provide a foundation for evaluating Medicare reform initiatives. It also has implications for estimating the cost-effectiveness of universal coverage proposals. To these ends, we conducted a cost-effectiveness analysis of the average private supplemental insurance policy relative to Medicare alone. Such an analysis, together with sensitivity analyses of key parameters, may provide a benchmark with which to assess the efficiency of other services and technologies that vie for addition to the Medicare market basket.
Previous studies have examined the costs associated with providing health insurance to younger cohorts \[[@B11]\]. Other studies have examined the effectiveness of health insurance in general \[[@B6]-[@B9]\]. In this study, we use a natural history Markov model to estimate the cost-effectiveness of supplemental medical insurance among the elderly, updating and amalgamating these earlier studies. Our cohort consists of persons over the age of 65 who presently do or do not have insurance.
Methods
=======
Overview and definitions
------------------------
To estimate the expenditures and health effects of private supplemental Medicare insurance to persons over age 65 relative to those who lack such coverage, we used a 3-step process. First, we performed regression analyses on nationally representative data of Medicare recipients with private supplemental insurance. These analyses modeled the effects of socio-demographic, behavioral and clinical variables on expenditures, health-related quality of life (HRQL) values, and mortality.
Those without supplemental insurance coverage differ in socio-demographic characteristics from those that do have coverage. We therefore apply the socio-demographic characteristics of the uninsured to regression models of insured persons. The use of the predictive functions of linear regression analysis allows for a better estimate of the impact of providing supplemental insurance to the uninsured.
Finally, using Markov decision analysis modeling that incorporates the predicted costs and health benefits, we estimated the cost-effectiveness of providing private supplemental insurance to the average 65-year-old with Medicare only coverage.
We use the quality-adjusted life year (QALY) as an outcome measure. The QALY combines health-related quality of life with life expectancy; one QALY represents a year of life lived in perfect health \[[@B12],[@B13]\]. In our analysis, the incremental cost per QALY gained is the additional money spent on supplemental health insurance divided by the additional gains in quality-adjusted life expectancy. This ratio is called the incremental cost-effectiveness ratio (ICER). Our study was conducted from the societal perspective, employed a community derived measure of HRQL to calculate effectiveness in terms of QALYs, and used a 3% discount rate, following recommendations of the Panel on Cost-Effectiveness in Health and Medicine \[[@B12]\]. We did not capture over-the counter drug costs, costs associated with institutionalized persons, or direct non-medical costs.
Health-related quality of life scores were derived from the Health and Activities Limitation Index (HALex) \[[@B14]\]. The HALex is comprised of two health domains, self-rated health and role limitations. The HALex exhibits reasonable validity, but because it captures only 2 health domains, its sensitivity to the full spectrum of morbidity is limited \[[@B15],[@B16]\]. Details of these methods have been published elsewhere \[[@B14]\].
The most recent publicly available nationally representative information on the relationships between health insurance and HRQL and mortality is the 1993 National Health Interview Survey (NHIS) linked to the 1995 National Death Index \[[@B17],[@B18]\]. We therefore used these data to the to obtain prospective HRQL and mortality differentials by private supplemental insurance coverage status. We obtained costs from the 1996 Medical Expenditure Panel Survey (MEPS) \[[@B11],[@B19]\].
Regression analyses
-------------------
We developed regression models using SUDAAN (Research Triangle Park, NC) and STATA (College Station, TX), adjusting for the complex sample designs used in both NHIS and MEPS. In these models, HRQL scores, survival, or total medical expenditures were entered as dependent variables. Age, race, ethnicity, gender, education, family income (NHIS) or percent of federal poverty level (MEPS), family size, marital status, behavioral risk factors (smoking status and seatbelt use in NHIS), number of conditions reported (NHIS) or self-rated health (MEPS), and area of residence (region of country, and urban or rural location) were entered as independent variables. In the expenditure models (MEPS), only persons whose insurance status did not change throughout the year were included.
We estimated the effects of the socio-demographic and health variables upon HRQL scores using linear regression. Expenditures were analyzed using generalized linear regression models with a gamma distribution and a log link function (to account for the skewed distribution of expenditures). The resulting parameter estimates from the analyses using those with private supplemental insurance were used to calculate predicted HRQL scores and expenditures for those with Medicare only using their socio-demographic, behavioral and health characteristics. The predicted expenditures were adjusted to include administrative costs for health insurance plans. All costs were deflated to constant 1994 dollars using the medical portion of the consumer price index.
Proportional hazard models were used to estimate the risk of death due to lacking private supplemental health insurance.
Decision analysis models and sensitivity analyses
-------------------------------------------------
We examined these costs and benefits over the lifetime of the average 65-year-old using a Markov decision analysis model using DATA Professional 4.0 (TreeAge Software, Williamstown, MA). Our models evaluated two possible events: giving private supplemental insurance to those lacking it, or receiving only Medicare parts A and B. For each health state, subjects were exposed to an annual, age-specific risk of death, with survivors gaining one HRQL-adjusted year and medical costs. The values for these year-to-year changes in health by insurance status were obtained from our regression analyses. Thus, the average cost of medical care for a 65 year-old given private supplemental insurance is tabulated and the model is advanced one year. Age-specific mortality rates are used to determine the proportion of subjects dying, and survivors are assigned a discounted QALY and costs for the next year. This process is repeated until over 99% of the subjects are dead. The same process is applied to the Medicare only cohort, and incremental values are calculated.
The variables used in our analyses were subjected to a Monte Carlo simulation and 1-way sensitivity analyses. In a 1-way analysis, all variables are held constant but one. In a Monte Carlo simulation, values for all variables are randomly sampled from a statistical distribution \[[@B20]\]. In our Monte Carlo simulation, we used a triangular distribution. In this distribution, the base-case estimate is entered as the most likely value and values between the high and low value are linearly interpreted. The assumptions of the analyses are listed in Table [1](#T1){ref-type="table"}.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Principal assumptions of the analysis.
:::
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
*Subjects will remain in stated insurance category*. We did not have information on whether subjects continuously or intermittently lacked private supplemental insurance in all datasets.
*Administrative costs associated with private supplemental insurance companies are the only relevant costs*. It was assumed that costs associated with provider and hospital administration of supplemental policies and employer administrative costs would be negligible.
*Private supplemental insurance company administrative costs are proportionate to expenditures*. We calculated administrative costs using an expenditure driven formula \[see [Additional file 1](#S1){ref-type="supplementary-material"}\].
*We included all relevant covariates in our regression models*. It is possible that unmeasured variables explain the observed effects of insurance on costs, mortality, and HRQL.
*Expenditures included in the MEPS survey reflect all relevant costs*. The costs reported in the MEPS do not include the institutionalized population, over-the-counter (OTC) medications, or alternative care.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
:::
The decision analysis models were validated by comparing their outputs with abridged health-adjusted life tables we generated using spreadsheets \[[@B21]\]. Markov modeling allows for a more accurate estimation of year-to-year effects by accounting for the effect of mortality on cost and effectiveness. It also simplifies sensitivity analyses. Abridged life tables allow for clear and simple estimations of changes in life expectancy when different interventions are applied and may thus serve to validate Markov models.
Detailed descriptions of the data sources, regression models, and the decision analysis models are available \[see [Additional file 1](#S1){ref-type="supplementary-material"}\].
Results
=======
Hazard ratio estimates were unstable when the behavioral risk factors were included since including them reduced the sample size. However, behavioral risk factors exerted little effect on differences in mortality by insurance status. For our base-case analysis, we therefore used the larger sample, excluding the behavioral risk factors, but reduced the obtained adjusted hazard ratios by 10% \[see [Additional file 1](#S1){ref-type="supplementary-material"}\].
In 1996 expenditures for persons with private supplemental insurance were higher than for those without private supplemental insurance (\$4915 vs. \$3956). These higher expenditures are reflected in greater utilization and prescription use (Table [2](#T2){ref-type="table"}).
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Relationship between private supplemental insurance, selected utilization, and expenditures in the 1996 Medical Expenditure Panel Survey.
:::
Medicare only Private supplemental insurance
----------------------------------------- --------------- --------------------------------
Physician office visits 6.7 8.3
Physician office visit expenditures \$415 \$745
Prescriptions filled 16.5 19.4
Prescription expenditures \$565 \$722
Out-of-pocket prescription expenditures \$419 \$371
Insurance prescription expenditures \$6 \$333
:::
Because of their lower socioeconomic status and lower health status, the predicted medical expenditures among those provided private supplemental insurance coverage would be higher than those who currently have it (Table [3](#T3){ref-type="table"}). The average cost of supplemental insurance coverage would also be slightly higher (\$1349 vs. \$1270). Our models also predict small improvements in HRQL, and relatively larger improvements in mortality, but both remain worse than those currently with private supplemental insurance (Table [3](#T3){ref-type="table"}).
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Observed and predicted expenditures and health outcomes
:::
Expenditures
-------------------------------------- ----------------------------------------- ------------------------ -------------
Age-group Observed Private supplemental insurance Observed Medicare only Predicted\*
65--74 \$4029 \$3632 \$4141
\>75 \$5942 \$4243 \$5919
Health-related quality of life score
65--74 0.78 0.71 0.73
\>75 0.70 0.62 0.67
Mortality†
65--74 0.043 0.067 0.048
\>75 0.083 0.129 0.092
\*Values for effects of supplemental insurance on those with Medicare only using predictive regression models.
† Proportion dying within 24 months.
:::
Providing lifetime private supplemental insurance to the average 65-year-old would cost society an additional \$22,000 and would produce a gain of 0.88 QALY -- an incremental cost-effectiveness ratio (ICER) of \$24,000 per QALY gained (Table [4](#T4){ref-type="table"}).
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Results of cost-effectiveness analyses at a 3% discount rate.
:::
Strategy Cost (USD) Incremental Cost Effectiveness Incremental Effectiveness\* Incremental Cost-Effectiveness†
--------------------------------------------------- ------------ ------------------ --------------- ----------------------------- ---------------------------------
Cost per quality-adjusted life year (QALY) Gained
Medicare only \$34,000 \- 10.06 QALYs \- \-
Supplemental \$56,000 \$22,000 10.94 QALYs 0.88 QALYs \$24,000
Cost per life year (LY) gained
Medicare only \$34,000 13.26 LYs
Supplemental \$56,000 \$22,000 14.14 LYs 0.88 LYs \$24,000
\*Rounded to 2 decimal places.
†Rounded to nearest thousand dollars.
:::
Sensitivity and sub-analyses
----------------------------
Table [5](#T5){ref-type="table"} lists the results of one-way sensitivity analyses on ICERs. Over a plausible range of high and low values, predicted administrative costs exerted the greatest effect on the incremental cost-effectiveness of supplemental health insurance, resulting in a range of values from \$21,000 per QALY gained to \$41,000 per QALY gained. In the Monte Carlo analyses, 95% confidence interval of the ICER values for private supplemental insurance relative to Medicare only ranged from \$12,000 to \$36,000 per QALY gained. The Markov model and spreadsheet produced similar quality-adjusted life expectancy differentials between groups.
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
Variables included in the decision analysis model, their assigned high, low, and baseline values, and the effect on the incremental cost-effectiveness ratio (ICER) of insurance relative to no insurance.
:::
Used in Model\* Effect on ICER
---------------------- ----------------- ---------------- ------- ---------- ----------
Variable Baseline High Low High Low
Administrative costs 5% 25% 0% \$41,000 \$21,000
Hazard Ratio 1.5 1.75 1.25 \$19,000 \$37,000
Discount rate 3% 5% 0% \$26,000 \$23,000
Error in cost† 0% 125% 0.75% \$39,000 10,000
\* All values represent the percentage change in the baseline value except for the hazard ratio, which was varied from previously published estimates to the high value observed in logistic regression analysis \[see [Additional file 1](#S1){ref-type="supplementary-material"}\].
†Variability in the cost estimate for the supplemental Medicare insurance. The baseline cost is multiplied by an error term ranging from 0.75 to 1.25.
:::
Figure [1](#F1){ref-type="fig"} shows the effects of providing private supplemental insurance at different ages. There is generally a trend toward increasing cost-effectiveness with age.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Changes in the incremental cost-effectiveness (ICER) of providing private supplemental insurance to those with Medicare at different ages.
:::
:::
Finally, to estimate the effect of medical inflation under the assumption that health care becomes no more effective over time, we examined the ICER of private supplemental insurance in 2001 dollars. With effectiveness held constant, inflation has a differential effect of the cost of medical care for the insured relative to the uninsured, increasing the ICER to \$30,000 per QALY gained.
Discussion
==========
Our models predict that the average private supplemental insurance plan provided to Medicare recipients is associated with substantial health benefits. Because health insurance increases the utilization of an array of medical goods and services, the impact of providing private supplemental insurance to the elderly is predicted to be relatively large, improving life expectancy in the population by about 11 months.
The ICER of private supplemental insurance is smaller than many medical interventions employed in day-to-day practice \[[@B22]\]. Moreover, incremental expenditures would be considerably lower than investments in many other well accepted social programs (e.g., airline and automobile safety) per life year gained \[[@B23]\].
The cost-effectiveness of supplemental insurance appears to increase with age (see Figure [1](#F1){ref-type="fig"}). This may reflect the increasing absolute net benefit of medical care interventions with increasing age; that is, as the absolute risk of mortality increases with most conditions with increasing age, so the potential absolute net benefit of treating those conditions also increases. The slight decrease in cost-effectiveness at age 70 is likely due to random error, as the sample sizes in each age-group become progressively smaller. Also, the mean expenditures of those in the sample in the age-group without supplemental insurance was notably higher than that of persons in older and younger age-groups.
Our analyses are susceptible to at least two main contravening biases. First, persons who choose to purchase private supplemental insurance may be more health oriented, and it is that orientation, not insurance, that is associated with improved outcomes. Second, and conversely, persons may purchase private supplemental insurance because they perceive themselves to be at higher health risk than is captured in the covariates we used. To control for health orientation, we included two behavioral risk factors, seatbelt use and smoking status, as covariates. However, risky health behavior can be quite nuanced. For example, there is evidence that persons with greater medical skepticism are less likely to have health insurance, have lower health care utilization, and have higher mortality \[[@B24],[@B25]\]. Thus, the use of a cost-effectiveness ratio potentially mitigates some of these biases; those without supplemental insurance may have worse health habits and thus worse outcomes even if insured, but they may also be less likely to utilize health care even if they are given supplemental insurance -- thereby generating lower costs than predicted \[[@B24],[@B25]\]. It should be noted that in this single observational study one cannot be certain of the extent of the possible contravening biases, so that net bias may remain in the estimates used.
There are other notable limitations to our analyses. First, the data sources we used, though they are the most recent publicly available nationally representative samples, are somewhat dated. While the costs of medical care have outpaced general inflation, the effectiveness of medical interventions has also increased. The net effects of these changes are uncertain. However, even if effectiveness showed no improvement, the ratio would increase to just \$30,000 per QALY gained in 2001 dollars.
Second, as suggested by Table [2](#T2){ref-type="table"}, private supplemental insurance coverage is associated with both higher utilization of ambulatory care and greater prescription use. The analyses reflect the predicted impact of an average private supplemental insurance policy and do not provide information about the effects of specific components. The recent controversy over the decision to provide limited Medicare coverage for lung reduction surgery suggests that insurance coverage decisions will increasingly have to deal more explicitly with both the benefits and costs of specific components of health care \[[@B26]\].
Our analysis rests on two major assumptions. First, private supplemental insurance increases utilization of healthcare. Second, this increased utilization results in improved health. Interestingly, our data suggest that private supplemental insurance produces little improvement in health-related quality of life, with both incremental life expectancy and quality adjusted life expectancy rounding out to 0.88 (Table [4](#T4){ref-type="table"}). It is unlikely that the risks of medical care among the living elderly (e.g., impotence secondary to anti-hypertensive medications) outweigh the benefits conferred by this care (e.g., reduced stroke due to anti-hypertensive medications). This is an area that requires further study.
Summarizing these limitations, we acknowledge that neither this study nor any individual study using observational data may adequately address the problems of endogeneity (in econometric terms) or confounding (in epidemiological terms) or establish causality. Additional studies are needed using different datasets and different approaches. Within the econometric framework, studies employing propensity scores or instrumental variables (if good instruments can be found) may yield less biased estimates of the effects described here. Within an epidemiological framework, studies using a broader array of potential confounders may also be helpful. Finally, other study designs employing quasi-experimental methods may be helpful: for example examining the effects of involuntary loss of insurance coverage, or comparing costs and outcomes for persons with more generous coverage provided as a result of prior employment (rather than current choice) to those with less generous coverage. Until such studies are available, the results reported here should be seen as providing a starting-point to examine the incremental cost-effectiveness of specific proposed additions to the overall Medicare package.
Conclusion
==========
Our analyses are not intended to suggest that expanding private supplemental health insurance plans is the most efficient means of improving health outcomes. A simple expansion of Medicare benefits targeted at covering medication costs and reducing inpatient co-payments may be less expensive. The General Accounting Office and Congressional Budget Office, for instance, has found that government programs may offer substantial savings in administrative costs over private sector plans \[[@B27],[@B28]\]. Our analysis does suggest, however, that expanding supplemental health insurance for the elderly may be relatively cost-effective, with a cost-effectiveness ratio similar to that of younger cohorts \[[@B29]\].
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
PF conducted the regression analyses and made major contributions to the development of the manuscript. PM conducted Markov and sensitivity analyses and made major contributions to the development of the manuscript. MG provided guidance for the study and contributed to the development of the manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6963/5/23/prepub>
Supplementary Material
======================
::: {.caption}
###### Additional File 1
Technical appendix. Is health insurance cost-effective? A technical appendix that describes methodology in more detail and provides additional data that might be of interest to readers.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
We would like to thank all of those who helped us in developing this study. Special thanks to Howard Berliner for advice on the direction of the study. Willard Manning provided inestimable help in suggesting optimal approaches to modeling costs.
|
PubMed Central
|
2024-06-05T03:55:55.281458
|
2005-3-14
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079833/",
"journal": "BMC Health Serv Res. 2005 Mar 14; 5:23",
"authors": [
{
"first": "Peter",
"last": "Franks"
},
{
"first": "Peter",
"last": "Muennig"
},
{
"first": "Marthe",
"last": "Gold"
}
]
}
|
PMC1079834
|
Background
==========
Iran, a country of sixty eight million has a national health service which employs over 70,000 nursing personnel (including operating room technicians) who provide nursing care in general and specialty hospitals. Although the population of nurses is approximately one hundred and twenty thousand, many are unemployed. Among the unemployed are those who choose not to work after marriage. Consequently, Iran like other countries is faced with a nursing shortage. The impact of this nursing shortage lead nurses to work more than their required shift of 192 hours per month; with potentially 150 hours of overtime in some parts of the country. The role of nurses is unclear, for although most of them are employed in hospitals, yet few or none are in the role of Public Health Nursing. Additionally, because the people of Iran have a poor image of nursing, those who choose nursing as a profession do experience low self-esteem. The combination of poor image and questions regarding quality of patient care, one would find it hard to believe that these nurses are graduates with a Baccalaureate Science in Nursing (BSN) from either a nursing school or medical science university.
Historical perspective of nursing in Iran
-----------------------------------------
Historically, records of nursing in Iran before 1915 showed that nursing care was carried out by household women or servants. Hospitalized patients were also cared for by untrained personnel \[[@B1]\]. Because of this history, lack of basic education, low cultural status, and some religious limitations for women, nursing as a profession/career neither gained high standard nor recognition. However, 1915 is noted as the turning point for nursing in Iran. During that year the American Presbyterian Missionary Society (APMS) pioneered the training of a few nurses in a small missionary hospital. Subsequently, in 1916 a three-year nursing school was established in Tabriz. After 1916 there was a gradual increase of nursing schools across the country leading to a growing demand for nurse educators who were already scarce \[[@B2]\]. Because there was a lack of qualified nurse educators in Iran the World Health Organization (WHO) was appointed to assist in these schools. The Iranian faculty shortage was further supplemented with recruits from England and United States of America (USA). A two-year program of study was developed and the entry-level requirement for these schools was a minimum of nine years of general education \[[@B3]\]. This minimum requirement failed to attract a large number of prospective nursing students for nursing being predominantly female and in Iran women were prohibited to engage in any social activities that required close contact with men. Avoiding close contact with men is almost impossible in nursing, for healthcare system is not designed to be all male or female and therefore posed a threat for nursing schools and the recruitment of prospective nursing students.
Following the Second World War, Iran began gaining momentum for advancing nursing as a profession. The Princess Ashraf School of Nursing was established with the appointment of nursing faculty from England and offered a three-year program where the admission requirement was a high school diploma. Then, in 1943 The Iranian Nursing Association (INA) was formed by a group of Iranian nurses who were educated abroad and returned to the country \[[@B2]\]. Next, in 1952, a nursing division was established in the Ministry of Health and this was the first time nursing was officially recognized by the government and became part of its structure. The first university program for obtaining a BSN began in October 1967 at Shiraz University \[[@B2],[@B3]\].
The increasing demand for healthcare in Iran forced the stakeholders of nursing education to develop new initiatives to meet the demand. As a result, in 1975 The Ministry of Sciences approved the Associate Degree of nursing program (ADN). Thereafter, nursing education in Iran was on the move, there were ADN, and BSN programs with a Masters of Science in nursing (MSN) program gradually developing. Although nursing education was growing, the numbers of graduates were still inadequate to meet Iran\'s demand for healthcare. For example, between 1915--1979 a total of 8,546 nurses graduated to provide healthcare services for a 30 million population. Because fully qualified nurses remained few in numbers, the alternative was to complement the nursing care in hospitals with auxiliary nursing personnel.
In 1979 when the Islamic revolution took place, major social and cultural changes occurred. These changes not only impacted the health care system, but nursing services too. Previously, the majority of nurses were female and they were caring for both men and women. However, as a result of the Islamic revolution, the government decided that nursing school admissions constitute 50 percent males, with the belief that men should care for men and that women must be separate from men while they engage in social activities.
Despite all the positive changes, bridging the gap of the nursing shortage continued to be a challenge, for new issues continued to emerge and there were no easy solutions or \"quick fixes\". Some of the issues that emerged were, increased birth rate causing an increased population, and the start of the imposed war between Iraq and Iran which only proliferated Iran\'s nursing shortage. Concurrently, the Iranian government decided to re-structure the Ministry of Health where it was re-named The Ministry of Health and Medical Education. Under this ministry new institutions of medical and nursing education were established and some existing institutions expanded. As a result, after the revolution 11,274 nurses were trained in the first decade, with an addition of 22,000 more nurses over the next 7 years \[[@B1]-[@B3]\]. Also, after the war, nurses\' aides and ADN nurses were trained too. While providing educating the different types of nursing personnel narrowed the gap of the nursing shortage dilemma, it created role ambiguity and masked the criteria for \"who is qualified to practice nursing\".
To bring clarity to \"who is qualified to practice nursing\" the Ministry of Health and nurse leaders collaborated and wrote job descriptions for the different levels of nurses. However, these job descriptions were not fully implemented because of several obstacles such as, vague or to abstract job description, work overload, poor staffing in the hospitals, and lack of continuing education for nurses. Consequently, there were several overlaps among the varying roles for nursing personnel in hospitals.
By this time, admissions to nursing schools soared only to be faced with yet another hurdle to overcome. There were too few faculties and a dwindling number of institutions for clinical placement. As a result, many physicians and inexperienced nurses were assigned to faculty roles in nursing education. The mix of physicians and inexperience nurses drastically changed the philosophy and educational model for nursing education. Nursing education was primarily following the medical model. So nurses are taught the medical management of diseases and to follow physicians\' orders without questioning; this educational model was not only oppressive, it silenced its participants and diminished their self-confidence. All this only exacerbated the difficulty of teaching professional nurses to be assertive and be in control of their practice by giving \"voice\" to their concerns.
Because the government did not recruit enough nurses during the war between Iraq and Iran, nurses had no choice but to work overtime and student nurses became part of the work force in order to compensate for the staffing shortage in hospitals. This shift in nursing trend converted nursing care to tasks and these nursing tasks soon became routine care in hospitals. It quickly became evident that new graduates were inefficient, so the MSN programs expanded with the main goal of preparing nurses for faculty roles. So, nurses were educated for roles according to nursing specialties such as, medical-surgical, psychiatric, community health, pediatric, and management, but even with this expansion roles were still not clearly delineated. Thus, many of the graduates from the MSN programs leaned towards nursing education rather than nursing practice. There seemed to be no end to Iran\'s nursing issues, it was time for a new approach.
So, after 30 years of fighting for state recognition, in 2002 Iranian Nursing Organization (INO) was approved by Iran\'s legislature and the INO established itself the same year. Now, the INO has the legal responsibility to represent all nurses in all sectors of nursing. Some of its key objectives are, improving the quality of patient care and developing standards for nursing practice. Even though this newly established organization could play a significant role in the development and empowerment of nurses, much time is needed to establish themselves as a governing body for nurses and they certainly have a tough job to shape the future of nurses and nursing in Iran. Since empowerment is crucial to the role of nurses, a qualitative study was conducted and aimed at designing a model for empowering nurses in Iran.
Methods
=======
A qualitative study was conducted using a grounded theory approach \[[@B4]\]. The purpose of study was to design a model for empowering nurses in Iran. The term grounded theory reflects the concept that theory emerging from this type of work is grounded in the data \[[@B5]\]. The researcher\'s purpose in using grounded theory is to explain a phenomenon within the social situation and to identify the inherent processes involved \[[@B6]\]. The grounded theory approach has been used in nursing research since 1970 and the studies have focused on nursing practice and education \[[@B7]\]. Empowerment is clearly a process rather than a static factor; therefore, a grounded-theory approach is the preferred methodology. In addition, this approach was selected because nurses\' practice takes place in a multidisciplinary team and grounded theory focuses on identification, description, and explanation of interactional processes between and among individuals or groups within a given social context \[[@B8],[@B9]\]. The research question is which processes influence empowering nurses and what is the appropriate model for empowerment of nurses in Iran. Data was collected by individual interviews which were audiotaped and transcribed and through observations which were recorded in field notes.
Participants and data collection
--------------------------------
The study comprised of 44 participants who are nurses in varying roles and settings, they are 12 nurses, 12 head nurses, two supervisors, three nurse managers (matrons), three nurse educators, three senior nursing directors (SND), two doctors and seven members of the INO. Purposive sampling was used at first and then continued with theoretical sampling according to the codes and categories as they emerged. Criteria for selection were nurses with more than five years of nursing experience who worked full-time in four large hospitals covered by the Ministry of Health and Medical Education in Tehran, Iran. Any nurse who met this requirement was considered a potential participant. Data collection began with staff nurses; after interviewing three nurses and coding the transcripts, the codes and categories that emerged were related to managerial support, organizational variables, and nursing education which led to the decision for the researcher to interview head nurses, supervisors and a few other key informants. Other key informants were higher-level managers, doctors, and nurse educators. Each interview session ranged from twenty minutes to three hours with an average of 115 minutes. Data was collected and analyzed over a six months period in 2003.
Interviews
----------
The main researcher contacted each of the potential participants to explain the objectives and the research questions. If the participant agreed to take part in the research, an interview was scheduled.
Participants were interviewed in a private room at the worksite using an individual semi-structured interview format and this was primarily the main method for data collection.
The interview guide consisted of open-ended questions to allow respondents to fully explain their own opinions, perceptions, and experiences. To start, each participant was asked to describe one of his/her own typical tour of duty; then to explain his/her own perceptions and experiences on \"professional empowerment\" and the \"factors influencing it\". For instance, he/she was asked, in your opinion, what is the meaning of empowerment in nursing?; Do you feel you are an empowered nurse?; can you describe some of your instances when you felt empowered?; which factors enhanced your experiences of empowerment? Tell me what are your thoughts on how can nurses and the nursing profession become empowered. Brief notes were made about the issues raised during the interview. Questions were asked later if these issues had not been spontaneously clarified. The interviews were transcribed verbatim and were analyzed consecutively.
Observation
-----------
The main researcher conducted twelve sessions of participant observation in all four hospitals. Observations were conducted during the different shifts in emergency, medical, surgical and intensive care units and involved not only the nurses interviewed but also the other nurses present during the shift. To conduct the observation the researcher sat in a corner of the ward and either watched or followed individual nurses around. Even though the head nurse requested that the researcher do not participate in direct patient care, minimal assistance was given upon some nurses\' request. The main focus of the participant observation was on nurses\' interactions with their patients, colleagues, head nurses, supervisors and doctors. Emphasis was on nurses\' participation on decisions related to patient care and practice settings. The brief notes taken at the time of observation were written up in detail on the same day. These detailed notes were used as data concurrently with the interviews.
Data analysis
-------------
The data collection and analysis were done simultaneously according to the grounded theory approach. The interviews and observations data were analyzed concurrently using constant comparative method. Each interview was transcribed verbatim and analyzed before the next interview took place, so that each interview provided direction for the next. Following transcription, the tapes were replayed and notes were made onto the transcripts. Notes included comments such as tone of voice, recurrent themes, and the researcher\'s own thoughts and feelings about the nature and significance of the data.
Open, axial and selective coding were applied to data \[[@B6]\]. During open coding, each transcript was reviewed multiple times and codes were generated from the respondent\'s words and the researcher\'s constructs. For example, the code \"managerial support\" was generated by the researcher from a respondents\' comments, \"I think it is very important that we as nurses would be supported by our colleagues and managers. Powerful nursing owes support, it must be supported and the people in charge should support it.\" Codes that were found to be conceptually similar in nature or related in meaning were grouped in categories. The categories and codes from each interview were compared with other interviews in order to identify common links. Categories were related to their subcategories in axial coding. Coding was occurred around the axis of a category, linking categories at the level of properties and dimensions. Analytical tools include asking questions and making comparisons helped in finding the properties of each concept \[[@B6]\]. In this stage the structures of categories were related to the processes. For instance, the factors that contributed to nurses\' feeling of collective empowerment or disempowerment were identified. The process of integrating and refining the theory occurred in selective coding. It is here that the main category \"the culture and structure of the organization\" was ascertained. Lastly, a model was developed to illustrate the important steps which are required for empowering nurses in Iran.
Although a variety of personnel at different levels were interviewed, the themes that arose were consistent across interviews. Interviewing stopped when data saturation occurred. Data saturation occurred when no more codes could be identified and the categories were \"coherent\". After data analysis each participant was given a full transcript of their coded interviews with a summary of the emergent themes to determine whether the codes and themes are true or matched their responses. Five faculty members also did peer checking on approximately 45% of the transcripts. The transcripts of interviews were given to each of the above persons and they followed the same process as above to arrive at core themes. There was a 90% or greater agreement between different raters. Results were also checked with some of the nurses who did not participate in the research and they confirmed the fitness of the results as well.
Sampling strategies allowed for maximum variation to occur and a vast range of views and perspectives to be considered. The researcher attempted to have precise documentation of the direction of the research and decisions made to save the \"auditability\" that will make it easier for other researchers to follow. Prolonged engagement with participants in the research environment allowed the researcher to gain participants\' trust and better understanding of the research environment.
Ethical considerations
----------------------
Permission to conduct this study was through the ethics committee of Tehran University. Other ethical issues in this study involved the assurance of confidentiality and anonymity of the participants and their responses. All participants were informed of the purpose, design of the study, and that their participation was strictly voluntary. Participants signed a written consent to be in the study and permission was obtained from hospital directors and head nurses for the nurses to be interviewed and observed in the work setting.
Results
=======
Three main categories emerged from the data and each has 1--3 subcategories. These categories and their subcategories are representative of the main factors influencing the empowerment of nurses/nursing in Iran.
Category 1: personal empowerment
--------------------------------
According to the participants, personal empowerment is dependent on three variables, these are \"having authority\" and \"professional self-confidence\" for the \"application of professional knowledge and skills.\" According to two nurses, \"A powerful nurse is one who has good knowledge and can use it well\" and \"The power of a nurse depends on his knowledge and skills as well as his self-confidence in application of his knowledge in the provision of care for their clients.\" Furthermore, participants pointed out that the culture and structure of organization negatively impacts nurses\' self-confidence and authority and it is one that emphasizes \"physician centeredness\". Two participants noted that \"I must have the right to do nursing care based on my diagnosis, but I haven\'t this authority now,\" \"a person can be powerful only when he/she can decide on his/her own.\" Others pointed out that \"the public hasn\'t an appropriate view on the nursing profession,\" \"nurses cannot provide their own services to the public directly and people go to the doctor first,\" \"nurses are only expected to do the doctors orders,\" \"their workloads are high,\" and \"there isn\'t any system for nurses\' continuing education.\" Collectively all these variables have a negative effect on nurses\' self-confidence and minimized their ability to exercise authority or power in the practice setting.
Category 2: collective empowerment
----------------------------------
According to the participants \"collective power\" of nurses or the power of the profession of nursing comes from the interaction between two variables \"supportive management\" and \"unity.\"
They considered the unity of nurses and coming together in nursing organizations as an efficient way to increase their professional power. Two participants commented: \"to me the most important factor in the power and influence of a profession is the unity of its members\". \"We could be powerful when we are together and the best way for this is connecting to the nursing association, by this we can develop standards and regulations for our profession and these are the important prerequisites for supporting the nurses as professionals. These are prerequisites for determining our domain of professional authority.\"
The participants were aware that collective power brings them the possibility for obtaining better working condition as well as professional self-regulation. Also, collectively their voices are stronger and more influential. The latter would make it possible for them to develop standards for their profession and clearly define the scope of authority and responsibility of each nurse in his/her specific setting. Thereafter, these nurses will be able to apply their professional power in their practice settings to provide high-quality patient care. Nurses frequently stressed that supportive management will empower them. The participants\' views and experiences on supportive management were categorized under the three subheadings of \"provision of job related supports,\" \"provision of financial support,\" and \"provision of emotional support.\" However, a feeling of being unsupported was dominant among nurses. Shortage in the nursing workforce and lack of supplies such as sheets, dressing equipment and so on, prevented nurses from applying their professional knowledge and skills. Consequently, they were unable to meet their clients\' needs, which give them a feeling of inadequacy and dis-empowerment.
Category 3: the culture and structure of organization
-----------------------------------------------------
The culture and structure of the health care system was another important factor that either facilitated or inhibited nurses\' abilities to feel empowered. As participants indicated, these factors hindered the nurses\' power in the patient care settings. This is true; when nurses are part of an organization that is \"physician centered\" they are almost extinct. One nurse manager noted \"the nurses capabilities are not use appropriately because of some cultural and organizational factors\". Data related to the cultural and organizational variables were categorized under two sub-headings \"the public culture\" and \"organizational culture and structure.\"
The public culture
------------------
Most of the nurses participating in this study mentioned the poor imagine of nursing being held by the Iranian people. Participants felt that the general public did not recognize the professionalism of nurses. One nurse commented in this regard: \"the word nurse generalized to anyone who care for childes, sick or elderly, but not to a person who is expert in professional nursing care.\" Another nurses stated: \"Because of the low social status and a poor public image involved with nursing, I always try to avoid conditions where people ask about my work, so I hate to say anything about it,\" \"We are seen as doctors\' handmaidens, for people goes to the doctors first, but they referred them to the hospitals, where nurses are.\"
The public status of physicians has affected the culture of health care organization. Inter-professional relationships have been affected by this culture and uneven power relations have developed in which the doctors have the last word. One of the participating doctors said: \"the culture of health system doesn\'t regard nurses as professionals. There is an old and unscientific climate that induced a unilateral relationship between doctors and nurses. I think that many physicians don\'t know the real meaning of nursing.\"
Organizational culture and structure
------------------------------------
The belief that public and organizational culture have led to the development of a physician favored structure in the health care system was well articulated by the majority of participants. This is evident in the following quotes: \"the health system is at hands of the physicians,\" \"all of the top managers of health system and also in hospitals are the physicians,\" \"nurses are only considered as tools for carrying out the doctors orders.\" The design of the nursing system was affected by this physician-centered structure. The participants mentioned three parts of this system which included nursing services, nursing education, and nursing research.
The structure of nursing services
---------------------------------
Classifying nursing services as in-patient settings was considered a barrier and limitation to the true potential and capability of the nursing profession. One participant\'s comment was \"there are many nurses with a major in community health but they are not promised for working in the community.\" According to the participants, the root of this problem lies in the ambiguity of scope and standard of practice for nurses in the national healthcare system. One of the senior nurse directors who hold a position in the ministry of health shared: \"we lacked a logic in our health system and also in our nursing system. So, the territory of nursing is not clear,\" Another participant emphasized that: \"we are lacking a defined philosophy for nursing in Iran. This created structural obstacles that limited their power to implement their professional knowledge.\" It is obvious that there is an emergent need to clearly define the scope and standard of nursing practice in Iran, for only then can Iranian nurses practice as empowered professionals across the different areas within nursing. One of the senior nurses stated: \"I believe that defining the mission and territory of nursing is the first step in the restructuring the nursing system.\" He believed that \"we only offer some primary care within the hospitals, while we should integrate our services into the national health system. This is the best way for us to introduce our capabilities to the public.\" Also, this need is articulated by other nurses as evidenced in their statements \"We have not clear professional rules and regulations;\" \"We don\'t really have a clear job description\... one person\'s interpretation of what I should be doing could be different to mine\... I would have some difficulties with my management over decisions that I would have made and they would say \'you should be doing that\', and I would say \'I just can\'t\'.\"
When the researcher asked the participants \"what should be the first step in the process of empowerment of nursing in our country?\" they responded, \"We should define the personal and professional domain of nurses and their roles and position in health system.\" They believed that these should be done by the nursing organization. One of the senior nursing directors voiced \"a group of specialists in the field of nursing from nursing unions, nursing schools, and the nursing office in the ministry of health should define the domain and roles of nursing\...this is after such an important step that we could revise our nursing system and make it empowered so that nurses could implement their professional knowledge and skills for their clients at the level of hospitals and the community.\" The structure of in-patient nursing services was physician centered and routine-oriented. One of the supervisors stated: \"\...nothing could be done based on the nursing process\...it is expected that nurses only obey orders, give the drugs, monitor the blood pressures \...but that they do not intervene independently.\" Inadequate staffing and having to perform non-nursing duties were felt to be disempowering and inimical to the recognition of nursing as a profession. These were highlighted as barriers to provide quality nursing care. \"We\'ve taken on every role. If something needs to be done and nobody else is going to do it I am compelled to it. There will be trouble if I don\'t do it\" one nurse said.
The structure of nursing education
----------------------------------
There was great concern regarding the education system even though this plays an integral role in the process of empowerment. This concern originates from entry into nursing school as supported by the statement of one participant: \"entrance examinations only measure the academic capabilities of volunteers but do not measure their compatibility to the nursing profession.\" Another concern is the curriculum content, it is highly theoretical and one nurse said: \"\...nurse educators think the best nurses are the nurses who have more medical information. They give them an extensive range of disease-related, pharmacological and physiological information, but don\'t spend even ten minutes on the nursing care in a class of two hours.\"
Role models also played a significant role in the weakness of nurses. It seems nurse educators doubt their own confidence, competency, and autonomy and were ineffective role models for students. An experienced nurse educator believed that \"due to inexperience and freshness of most of the nurse educators, they lacked self confidence and could not educate a good new nursing generation.\" A philosophy of nursing education was considered absent and the question at large was \"what are the guidelines for nursing education\"? Therefore, \"there is no relationship between nursing and clinical setting and the clinical setting is inappropriate for students\' clinical placement\" as one senior nurse director said. As a result of all these factors, nursing schools will continue to graduate dis-empowered nurses.
Providing continuing education was judged critical for nurses to maintain competency in the clinical setting and to become life-long learners so they can develop confidence in \"giving voice\" to continually improve nursing practice and build a community of empowered professionals. However, low staffing and lack of staff development resources by \'the Ministry of Health and Medical Education\' only blocked the cycle of empowerment. As one participant pointed out: \"I think continuing education is empowering. People who are educated are more empowered to carry out their work, but the ministry of health doesn\'t support in-service education for nurses\".
The structure of nursing research
---------------------------------
Research utilization or the implementation of evidence-based practice was difficult for nurses for a variety of reasons. These are, the traditional structure of hospitals, poor quality of education, lack of continuing education, heavy workloads, no time, no mentoring and/or training in designing and conducting research, lack of financial resources, poorly defined nursing roles, lack of team work, and no opportunities for interdisciplinary relationships. These barriers are evident in the following quote of a nurse educator: \"the research findings don\'t use in our nursing practice at all. We are two groups in nursing. One group is teachers and mainly teaches in nursing schools, another group is clinical nurses who are very busy and are also not educated for doing research\... some of nurse educators also conduct researches not to be used in practice but only with the purpose of their promotion.\"
Presenting the model of empowerment
===================================
Professional empowerment is a dynamic process, which occurs through interaction at the personal, professional, cultural and organizational levels. On one hand, the existence of competent nurses (those who have a wide range of professional knowledge and skills, authority and self-confidence) is essential for empowerment, as presented in figure [1](#F1){ref-type="fig"}- and on the other hand, the power of the profession and the public\'s image of nursing can affect nurses\' self-confidence for demonstrating their full capabilities.
::: {#F1 .fig}
Figure 1
::: {.caption}
######
Relationship between personal and professional empowerment
:::
:::
Through our own knowledge, action, and behavior personal and professional power is created \[[@B10]\]. Those who build excellent interpersonal relationships by demonstrating their knowledge gain credibility and experience a sense of empowerment \[[@B11]\]. As previously indicated, empowerment of nurses has been negatively affected by a multitude of cultural and structural factors which negatively affected the nurses\' professional self-confidence and superimposed the belief that nurses are subordinates and handmaids to physicians. This caused nurses role to be restricted only to hospital rather than expanding it to the community setting. Other structural factors such as heavy workloads, inadequate staffing, lack of evidence-based nursing, and the task-oriented nature of nursing, have impeded empowerment of nursing as a profession and the role of nurses specifically in the application of their professional knowledge and skills \[[@B12]\].
As our data indicated, there was great concern regarding the education system even though this plays an integral role in the process of empowerment/dis-empowerment. The negative effects of incompetent nurse educators, lack of appropriate educational resources, oppressive methods of education, and using nursing students as part of the work force have also been confirmed by Adib hagbaghery et al.(2004); Espeland and Shanta (2001) and Chun-Heung (1997) \[[@B12]-[@B14]\].
After careful review, the data confirmed that \"the culture and organizational structure\" are the main variables affecting empowerment of nurses. Thus, a model of empowerment was developed and its emphasis is on re-designing the nursing systems. In this model, nursing system has three interrelated sub-systems; these are education, service, and research. Figure [2](#F2){ref-type="fig"} represents the model and its major variables as well as the relationship between those variables influencing empowerment of nursing as profession.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
The model for empowerment of the nursing profession in Iran
:::
:::
The goals of model
------------------
1\. Strengthen the INO as a professional organization to represent nurses in the country
2\. Re-design nursing system in hospital and community
3\. Extend the implementation of comprehensive care in the country
4\. Establish a national policy for nursing education
5\. Revise and update the curriculum for the BSN nursing programs
6\. Establish post-graduate education according to specialty and develop continuing education for nurses
7\. Develop scope and standards to regulate nursing practice
8\. Develop a valid and reliable system to administer licensure examination for nurses
It is imperative that these goals are achieved since it supports professional growth and empowerment of nurses in Iran.
As the model depicted, empowerment is a dynamic process, which results from interrelationship of personal and organizational factors. In other words, the process of empowerment requires changes not only in the structure of the organization but in the nurses\' perception of themselves and their role. These changes will lead to transformation and the development of new perspective by the nurses not only of themselves but their organization too. This transformation will provide new lens through which individuals will seek new knowledge, and develop new problem solving skills, acquire a variety solutions that can be observed in their thinking, and actions. Nurses, then, will replace their subservient behavior and routine based nursing care with evidence based practice to support their nursing care actions, so having nurses to function at this level is necessary for the process of empowerment.
Reforming and the close inter-relationship of the three sub-systems are the cornerstone for restructuring the nursing system. Therefore, nursing education will produce competent and confident nurses; nursing services will support quality patient care and other subsystems; and nursing research will interact with other sub-systems to form an interdisciplinary team which will provide the knowledge and skills for conducting research that will not only identify issues but will provide creative solutions.
Empowerment of nurses/nursing has been considered at both the micro and macro levels (fig. [3](#F3){ref-type="fig"}). At the macro level is the domain of nursing in the health system. As findings in this research have identified that nursing, its mission and its position have not been clearly defined in Iran. So, the first and most important step will be to define nursing, its mission and establish its position in the national healthcare system. Thus, the establishment of a committee named \"the Scientific Committee for Defining Nursing and its Mission\" (SCDNM) is recommended. The members of this committee will include members of the INO, nursing schools and the division of nursing in the Ministry of Health. Presently, these three parts are separate, but they will connect through their research subsystems for the execution of important mission such as:
::: {#F3 .fig}
Figure 3
::: {.caption}
######
Steps in the model for empowerment of the nursing profession in Iran
:::
:::
\- Defining nursing as a profession in the whole body of the national health system
\- Declaring a philosophy of nursing and stating its mission and position within the national health system
\- Designing clear job descriptions for nurses in general as well as in each specific setting
\- Establishing standards, rules and regulations for the profession
\- Establishing salary scale, standard of care, and nurse patient ratio, identifying staff competencies
\- Developing and integrating the community nursing system in the body of the national health system
\- Designing strategies for the development of nursing research and evidence based nursing
A review of the findings showed that, content and methods of nursing education along with the admission procedure need to be revised. Thus, the establishment of a committee named \"the Committee for Revision in Nursing Education\" (CRNE) was recommended. The committee consists of INO members, nursing schools, and the division of nursing in the Ministry of Health. The CRNE will develop an appropriate procedure for admission into nursing schools; evaluate and revise the content and process of nursing education- based on the approved criteria and the mission for the nursing profession in Iran.
The primary purpose of the revision was to improve the quality of nursing education so that when nurses graduate they are both competent and confident in their roles whether they are nurse generalist or specialists; they possess excellent inter-personal, critical thinking, and problem-solving skills which will enable them to identify their clients\' healthcare needs both in hospital and the community. Establishing a system for continuing education and in-services for nurses was highly recommended.
The philosophy, mission, and position of nursing, will be the base for restructuring the nursing service. An important strategy will be to develop and integrate community health nursing in the national healthcare system thereby providing comprehensive health care. This initiative will extend nursing care at the community level and increasing the visibility of nursing. The increased visibility will allow the public to observe the role of nurses and their capabilities so that the public can acquire new perspectives and opinions regarding nurses/nursing and this will help regain nurses\' self-confidence.
Another committee named \"the Human Resources and Restructuring Committee\" (HRRC) is also recommended for implementing the above strategy and restructuring the nursing system at the hospital level. The following tasks will be implemented:
\- Evaluation of nurse-patient ratio and recruitment procedures
\- Establishing standards for staffing levels and dividing them into three levels of Novice, Competent and Expert, according to their education, experience, and expertise
\- Designing specific and clearly defined job descriptions for each level
\- Allocating nurses in positions and on each nursing unit according to their level of expertise
\- Fostering the staff nurses\' professional growth by introducing a council system and developing a participative management style for nursing
These strategies will support nurses\' professional growth but will provide opportunities for nurses to be involved in making decision regarding their professional practice.
Participating in the research and implementing its findings will empower nurses; they will have ownership in the changes they make. Empowered nurses will certainly make the health care system more effective and efficient. According to the results of this study, the traditional, typical structure, and processes of hospitals and healthcare systems prevented conducting and utilizing research in nursing. Hence, conducting nursing research and implementing research utilization will be re-enforced in nursing through the coordination and interaction between researchers and nursing services. They will strengthen the research structures and develop strategies for conducting future studies. The following actions were recommended:
\- Adding nursing research to the curriculum of nursing
\- Establishing a nursing research center in all nursing schools, hospitals, clinical-educational centers, and strengthen the existing centers
\- Providing nurses with facilities to learn and conduct research projects
\- Supporting nursing research by providing financial, informational, technical and management support including the provision of adequate manpower for nursing services
\- Providing facilities for nursing faculties to conduct scientific seminars and conferences in nursing research.
These strategies can promote understanding of the research process and facilitate future nursing research studies. This will help nurses utilize the research findings and enhance the quality of nursing care and practice settings.
Discussion
==========
Nurses in this research frequently complained about lack of authority, some of them implied that they have the authority but do not use it due to some organizational and individual barriers such as heavy workloads, staff shortage, ambiguous job descriptions and the lack of self-confidence. These findings are confirmed with studies carried out by Laschinger et al. (2000), Fulton, (1997), Scott, Matthews and Corbely (2003) \[[@B15]-[@B17]\]. Reviewing the existing organizational structure, taking measures to balance medical, nursing and administration input to strategic planning and decision making, clarifying nurses\' roles in different fields in the hospital and the public setting will all help to set professional boundaries and improve the interdisciplinary relations. These measures will help nurses exercise power and authority and empower them to make use of their knowledge and skills in practice. According to our findings, the culture and the structure of the organization along with lack of support and authority have decreased nurses\' self-confidence. This finding has also confirmed by Madjar (1997) and Fulton (1997) \[[@B18],[@B16]\]. Thus, restructuring the health system, particularly the nursing system based on its philosophy and mission will strengthen nurses\' professional identity and self-confidence. Feeling of self-confidence is an essential factor for self-efficacy \[[@B19]\] and the process of empowerment. Restructuring of organization is an important way for increasing staffs\' professional self-confidence \[[@B20],[@B17]\]. Therefore, restructuring the nursing systems at all levels will empower and enhance nurses\' professional self-confidence and allow them to deliver quality patient care.
As the results of this study indicated, application of professional knowledge and skills and improving the quality of health care are the most prominent outcomes from restructuring the nursing system. The publics\' recognition and its view of nurses and the nursing profession will increase. This will enhance the nurses\' professional identity, their self-confidence, and will certainly empower them for developing and demonstrating their professional capabilities. Extending nursing services at the community level, developing a participative management style, mobilizing active staff involvement in a councilor structure, and involving nurses in the process of decision-making will empower nurses and replace their routine-oriented style of delivering nursing care with decisions based on their professional knowledge and skills \[[@B21],[@B10],[@B12],[@B22]\].
Conclusion
==========
Currently, nurses have demonstrated a strong commitment to change and improvement in the health care services and systems in Iran. Nursing in Iran has been progressive as evidenced by comparing the history of nursing to where they are now. Therefore, empowerment is essential for enhancing nurses\' role, strengthening the professional image, and continuously improving the healthcare system nationally and globally. Restructuring nursing services will eliminate barriers to poor quality nursing care, inadequate educational preparation, role ambiguity and low self-esteem among nurses. The nurses\' experiences and their perceptions regarding empowerment were studied and a model was developed for empowerment of the nursing profession in Iran. This model was sent to the Ministry of Health and approval is pending. Once approved, the model will be implemented and applied in healthcare practice settings. This model will not only benefit Iranian nurse leaders, nurses, and The Ministry of Health, but nurses and leaders globally.
List of abbreviations used
==========================
Baccalaureate Science in Nursing (BSN), American Presbyterian Missionary Society (APMS), World Health Organization (WHO), United States of America (USA), Iranian Nursing Association (INA), Associate Degree of nursing program (ADN), Masters of Science in nursing (MSN), Iranian Nursing Organization (INO), senior nursing directors (SND), Scientific Committee for Defining Nursing and its Mission\" (SCDNM), Committee for Revision in Nursing Education\" (CRNE), Human Resources and Restructuring Committee\" (HRRC),
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
MAH: Initiation and design of the research, collection and analysis of the data and writing the paper. MS: Co-analysis of the data and revision of draft paper.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6963/5/24/prepub>
Acknowledgements
================
The authors would like to acknowledge the assistance of all the nurses, nursing directors, supervisors, head nurses and physicians who participated in this research.
The authors also gratefully acknowledge of Minitie Inder-Maraj from Teachers College Columbia University for linguistic revision of the manuscript.
|
PubMed Central
|
2024-06-05T03:55:55.284277
|
2005-3-16
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079834/",
"journal": "BMC Health Serv Res. 2005 Mar 16; 5:24",
"authors": [
{
"first": "Mohsen",
"last": "Adib Hajbaghery"
},
{
"first": "Mahvash",
"last": "Salsali"
}
]
}
|
PMC1079835
|
Background
==========
Parkinson\'s disease (PD) is a neurodegenerative disease which at present has no cure, and, despite the variety of pharmacological and surgical treatment options \[[@B1]-[@B3]\], it usually results in severe disability. In Europe, age-adjusted prevalence rates of PD have been estimated at 1.6 per 100 population, with a steady increase in older groups, up to 3.5--3.6 in people aged 80 years and older \[[@B4]\]. Similar estimates have been reported for the United States \[[@B5]\]. In view of the increasing number of elderly people in developed countries, the prevalence of PD is expected to increase, as well.
Because PD is a chronic condition, the disease burden increases as the illness progresses, due to the appearance of both disease- and drug-related problems, resulting in the extensive utilisation of both health and community services \[[@B6]-[@B10]\]. The high rate of prevalent comorbid conditions occurring in PD, either associated or unrelated, significantly contributes to the utilisation of healthcare resources \[[@B11]\], which is higher than in subjects without PD \[[@B8]\] and has substantial economic implications \[[@B12]\]. As a consequence, it can be easily predicted that over the next few years more PD patients will use more healthcare resources, with a significant impact on the healthcare systems.
Knowledge of healthcare use patterns and of their determinants may greatly contribute to improve patient care, providing physicians, caregivers and politicians with information useful to estimate patient needs and to plan intervention strategies and allocation of resources. At present only few studies have examined healthcare use in PD patients and its relationship with potentially relevant factors, either related to the patient, such as age, sex and other social and demographic characteristics, or to the disease itself, such as severity, duration and comorbidity \[[@B8],[@B11],[@B13]\].
In the context of a project aimed at investigating healthcare needs, comorbidity, and drug use in PD patients, we previously reported that in PD patients attending a neurological service drug prescribing patterns may be associated with both disease- and patient-related factors, and that analysis of drug prescriptions may help to identify major comorbid conditions \[[@B14]\]. Since it was suggested that PD patients seen primarily by a neurologist may have an increased use of resources \[[@B15]\], as a logical extension of our previous work, we decided to investigate the use of healthcare resources in our cohort of ambulatory PD patients. However, as we were not interested in performing a formal cost-of-illness study, but rather to focus on the practical care of the patient, we directed our attention to medical healthcare resource use, with particular regard to the consequences for drug treatments. Regarding resource data collected, we decided to focus on hospital and emergency room (ER) admissions, and general practitioner (GP) and non-neurologist specialist visits, as these are the circumstances when physicians other than the neurologist take care of PD patients.
Our specific aims were: to assess the reasons for medical healthcare use, to look for possible relationship between the medical healthcare use and the pattern of anti-Parkinson drug (APD) medications, and to assess drug prescriptions issued by other physicians. All these aspects have so far received little attention, despite their straightforward relevance for the care of PD patients.
Methods
=======
The present study is part of a cross-sectional survey of PD outpatients consecutively attending the Centre for PD and Movement Disorders of the Neurological Service of the Ospedale di Circolo of Varese (Italy). The diagnosis of PD was established by neurologists skilled in movement disorders, according to the United Kingdom Parkinson\'s Disease Society Brain Bank Criteria \[[@B16]\]. Parkinsonisms resulting from other degenerative conditions or secondary to drugs, metabolic disorders or exposure to toxins were excluded. All the participants in the study gave informed consent.
Information about medical healthcare use over the previous year were obtained during a standard interview with the patient and/or her/his caregiver(s), by using self-report questionnaires. Medical healthcare resources addressed included: hospital admissions, ER admissions, GP and/or non-neurologist specialist visits (excluding therefore programmed neurological visits directly related to the follow-up of PD). Information was specifically sought about: the number of times each medical healthcare resource was used, the reasons for use, the occurrence on such occasions of modifications of the APD treatment and the prescription of drugs for indications other than PD (non-APDs). Collected information also included: the main patient demographic and clinical data (i.e. sex, age, age at onset and duration of PD), severity of PD, scored according to the Hoehn and Yahr (H&Y) scale \[[@B17]\], information about current drug treatments.
Data were stored in a database for subsequent analysis. To ensure confidentiality, patients\' names were not recorded and a unique personal identification code was used to perform data linkage. Reasons for medical healthcare resource use as well as indications for drug use were coded using the International Classification of Diseases, 9^th^revision, Clinical Modifications (ICD-9-CM) \[[@B18]\], whereas drugs were classified according to the Anatomical, Therapeutic and Chemical (ATC) classification index \[[@B19]\].
Statistics
----------
Results are presented as mean ± SD (unless otherwise stated) with n indicating the number of observations. Analysis of the data was performed on the whole patient population and on groups of patients stratified according to: decades of age and age at onset of PD, years of PD duration, and H&Y stage. Statistical significance of the differences between groups was assessed by use of parametric (Student\'s t test or ANOVA followed by Tukey-Kramer Multiple Comparisons post test) or non-parametric (Kruskal-Wallis analysis followed by Dunn\'s Multiple Comparison test) tests, according to the results of a preliminary normality test. Frequency distribution differences were analysed using contingency tables and the χ^2^test or the Fisher\'s exact test, as appropriate. All the statistical calculations were performed by use of a commercially available statistical software (GraphPad Prism version 3.0 for Windows, GraphPad Software, San Diego California USA, <http://www.graphpad.com>).
Results
=======
Use of medical healthcare resources
-----------------------------------
The study included 130 persons with PD, 69 females and 61 males. Their characteristics are shown in Table [1](#T1){ref-type="table"}. Fifty four patients (41.5% of total patients) had dyskinesia and 88 (67.7%) referred fluctuations of motor performances, including wearing off and on off phenomena, while 42 subjects (32.3%) were non fluctuators. Ten patients (7.7%) were taking antipsychotics for hallucinations, 22 (16.9%) were treated with antidepressants, 38 (29.2%) received anxiolytics. All these drugs were prescribed and managed by the neurologist of the Centre for PD. There was no significant difference between females and males with respect to any of the characteristics considered (data not shown). In the previous year, 92 out of 130 patients, corresponding to the 70.8% of all the patients, used one or more medical healthcare resources (1.9 ± 1.1 resources/patient, range 1--5). There was no significant difference between patients using and not using medical healthcare resources (Table [1](#T1){ref-type="table"}). However, among patients using resources both age at onset of PD and PD duration were significantly different according to the type of resource used (Table [2](#T2){ref-type="table"}). This finding was further supported by stratified analysis of the patients, inasmuch as 35.8% of the patients aged at onset of PD 60 years or more had at least a GP visit vs 12.2% in patients with earlier PD onset (*P*= 0.0040), and 41% of the patients with disease duration longer than 8 years were admitted to ER vs 13.5% of those with shorter disease duration (*P*= 0.0190). Other significant differences were: 75% of the patients aged more than 60 years used any type of resources vs. 50% of the younger ones (*P*= 0.0129), 35.4% of the patients aged 70 years or more and 50% of those in H&Y stage IV had at least a GP visit vs. 18.5% of younger patients (*P*= 0.0471) and 20.6% in those in lower stages (*P*= 0.0191).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Characteristics of the patients
:::
Overall Resource use
---------------------------- ---------------------- --------------------- ----------------------
patients, n (%) 130 (100) 92 (70.8) 38 (29.2)
age (years) 68.6 ± 10.0 (38--87) 69.4 ± 9.1 (45--86) 66.6 ± 11.7 (38--87)
age at onset of PD (years) 61.7 ± 10.2 (36--84) 62.3 ± 9.0 (40--84) 60.1 ± 12.7 (36--82)
PD duration (years) 6.9 ± 4.5 (1--24) 7.1 ± 4.8 (1--24) 6.5 ± 3.8 (1--14)
H&Y stage 2.6 ± 1.0 (I-IV) 2.7 ± 1.0 (I-IV) 2.5 ± 0.9 (I-IV)
Data are means ± SD (min-max), unless otherwise indicated.
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Characteristics of the patients according to the use of medical healthcare resources
:::
Hospital admissions ER admissions Specialist visits GP visits *P*
--------------------------------- --------------------- ----------------------- --------------------- --------------------- -------
resources used, n 29 31 74 42
patients, n (%) 25 (19.2) 29 (22.3) 54 (41.5) 35 (26.9)
resources/patient, mean (range) 1.2 (1--3) 1.1 (1--2) 1.4 (1--3) 1.2 (1--3)
patient characteristics
age (years) 71.0 ± 9.4 (45--86) 68.1 ± 9.0 (47--83) 68.6 ± 9.1 (45--85) 72.9 ± 7.7 (54--85) Ns
age at onset of PD (years) 64.3 ± 9.4 (40--84) 58.7 ± 7.3\* (42--72) 62.0 ± 8.6 (40--80) 64.1 ± 7.5 (50--78) 0.037
PD duration (years) 6.7 ± 3.5 (2--20) 9.4 ± 3.8^\#^(2--20) 6.6 ± 3.7 (1--24) 8.8 ± 4.6 (1--24) 0.004
H&Y stage 2.8 ± 0.9 (I-IV) 3.0 ± 1.0 (I-IV) 2.5 ± 1.0 (I-IV) 3.0 ± 1.0 (I-IV) Ns
Data are means ± SD (min-max), unless otherwise indicated.
\* = *P*\< 0.05 vs GP visits; \# = *P*\< 0.05 vs specialist visits.
:::
As regards patient gender, there were no major differences. However, stratified analysis revealed minor differences, as described hereafter:
\- the frequency of hospital admissions was significantly higher in males than in females in H&Y stage III (28.6% vs 3.8%, *P*= 0.0347) and in the age interval 70--79 years (33.3% vs 6.9%, *P*= 0.0407);
\- among females hospital admissions were more common in patients in H&Y stage II (30.8% vs 7% of patients in other stages, *P*= 0.0405), while among males they were more frequent in H&Y stage IV (47.1% vs 11.4% in lower stages, *P*= 0.0194);
\- in males (but not in females) the frequency of ER admissions and GP visits increased with increasing PD duration (52.6% and 42.1% of the male patients with disease duration longer than 8 years vs 9.5% and 11.9% with shorter disease duration, *P*= 0.0068 and *P*= 0.0048 respectively).
Reasons for medical healthcare resource use
-------------------------------------------
According to the ICD-9-CM classification \[[@B18]\], the most common causes for resource use were represented by **programmed hospital admissions and control visits**, which were included in the category \'supplementary classification of factors influencing health status and contact with health services\'. The next most frequent category were **symptoms, signs and ill-defined conditions**, collecting several complaints which could not be precisely assigned to other more specific categories. **Injuries**were 23 cases in total and included 10 fractures and 11 ER admissions for injuries due to falls. Table [3](#T3){ref-type="table"} collects detailed information about all the reasons for medical healthcare resource use recorded in the present study.
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
ICD-9-CM classification of the reasons for medical healthcare resource use
:::
Hospital admissions ER admissions Specialist visits GP visits Total \%
----------------------------------------------------------------------------------- ------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------------------------------- --------------------------------------------------------------------------------------------------------------------------------------------- -------------------------------------------------------------------------------------------------- ------- -------
**factors influencing health status and contact with health services (V01--V83)** rehabilitation (4) = rehabilitation (5); problems of eye (7), heart (3), diabetes (2), urogenital (2), skin (1); Rx of hip (1) general control (8); urinary cathether substitution (1) 34 19.3
**symptoms, signs, and ill-defined conditions (780--799)** chest pain (2); sleep disturbance (1); epistaxis (1); abdominal pain (1) abdominal pain (2); dyskinesia (1); epistaxis (1); dysphagia (1) abdominal pain (6); hallucination (1); headache (1) abdominal pain (2); hallucination (1); cough (1); nausea, vomiting (1); urinary incontinence (1) 24 13.7
**injury (800--959)** fracture of femur (1), femur plus humerus (1) injuries due to falls (11); fractures: ribs (2), wrist (1), upper limb (1), foot (1), unspecified (2); open wound (1) = fracture unspecified (1); open wound (1) 23 13.1
**diseases of musculo-skeletal system and connective tissue (710--739)** osteoarthrosis (1) pain in joint (1); lumbago (1) intervertebral discopathy (3); lumbago (3); pain in joint (2); osteoarthrosis (2); spondilosis (1); osteomyelitis (1) pain in joint (4); lumbago (2); rheumatism (1) 22 12.5
**diseases of the circulatory system (390--459)** acute hypotension (2); cerebral ischemia (1); aortic aneurysm (1) = hypertension (3); ischemic heart disease (2); atrial fibrillation (1); myocardial degeneration (1); aortic aneurysm (1); varicous veins (1) hypotension (2); hypertenson (1) 16 9.1
**diseases of the nervous system and sense organs (320--389)** cataract (3) hearing loss (1) tinnitus (1); presbyopia (1); cataract (1); hearing loss (2); glaucoma (1); impacted cerumen (1) conjunctivitis (2); vertiginous syndrome (1) 14 7.9
**diseases of the digestive system (520--527)** abdominal hernia (2); gastric ulcer (1); constipation (1); cholelithiasis (1) pulp disease (1); teeth extraction (1) avulsions of teeth (2); periodontal disease (1) constipation (2); intestinal obstruction (1); cholelithiasis (1) 14 7.9
**diseases of respiratory system (460--519)** pneumonia (1) pneumonia (1) asthma (1) influenza (5); acute laringytis (1) 9 5.1
**other\*** 4 1 13 2 20 11.4
**Total** 29 31 74 42 176 100.0
\* = encompasses the following ICD-9-CM categories (which individually accounted for less than 5% of total medical healthcare resource use): neoplasms (140--239), n = 8, 4.5%; diseases of genitourinary system (580--629), 7, 4%; diseases of skin (680--709), 3, 1.7%; infectious and parasitic diseases (001--139), 1, 0.7%; mental diseases (290--319), 1, 0.7%.
:::
Relationship between use of medical healthcare resources and APD treatment patterns
-----------------------------------------------------------------------------------
Patients treated with levodopa alone or associated with other APDs were respectively 62 and 66, and they did not differ in the overall frequency of healthcare resource use (75.8% vs 65.1%). As regards the frequency of use of specific medical healthcare resources, more patients on levodopa alone had GP visits than those on levodopa associated with other APDs (37.1% vs 18.2%, *P*= 0.018). No significant differences were found in the frequency of hospital admissions (21.0% vs 16.7%), ER admissions (43.5% vs 39.4%), or specialist visits (21.0% vs 22.7%). The only two patients on DA agonists alone both used medical healthcare resources: one was admitted to hospital and to ER (on separate occasions), and the other had two different (and separate) specialist visits.
Drug prescription during medical healthcare resource use
--------------------------------------------------------
**APD treatments:**A total of 11 patients (8.5% of total patients) had modifications of their APD treatment, most often during hospital admissions (6 patients), followed by specialist visits (2), GP visits (1 patient) and ER admissions (1 patient). More than one third of hospital admissions (11 out of 29, 37.9%) therefore resulted in changes of APD treatment (6 dose increased, 2 dose decreased, 2 drug withdrawn, 1 new prescription), which on the contrary occurred only occasionally as a consequence of specialist visits (1 dose decreased, 2 new prescriptions) or GP visits (2 drug withdrawn). Levodopa was the drug most frequently changed (10 prescription changes, including 3 dose increased, 2 dose decreased, 2 new prescriptions, and 2 drug withdrawn), while on the whole prescriptions of DA agonists were changed on 7 occasions (3 dose increased, 1 dose decreased, 1 new prescription, and 2 drug withdrawn). On one occasion there were no information concenring the specific type of change. **Non-APD treatments:**Prescriptions of non-APDs were issued to 86 patients (66.1% of total patients), most often during ER admissions and GP visits (37 and 36 patients, respectively), followed by specialist visits (13 patients). Most prescribed drugs were (ATC first level): general antiinfectives (J, 15 patients), musculo-skeletal system drugs (M, 14 patients), drugs for the alimentary tract and metabolism and for the cardiovascular system (A and C, both 11 patients). Information about occasions in which non-APDs were prescribed are given in Table [4](#T4){ref-type="table"}. Data about prescriptions of non APDs during hospital admissions were not analysed since nearly all the patients and/or their caregivers were unable to report affordable information concerning this particular issue.
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Non-APD prescriptions issued on the occasion of medical healthcare resource use
:::
ATC (first level) ER admissions Specialist visits GP visits
------------------------------------------------------------- --------------- ------------------- -----------
A -- alimentary tract and metabolism 2 (2) 3 (2) 9 (7)
B -- blood and blood forming organs 5 (5) 2 (2) =
C -- cardiovascular system 6 (6) = 5 (5)
D -- dermatologicals 2 (2) = 1 (1)
G -- genitourinary system and sex hormones 4 (3) = 1 (1)
H -- systemic hormonal preparations, excluding sex hormones 2 (2) = =
J -- general antiinfectives for systemic use 6 (5) 4 (4) 7 (6)
L -- antineoplastic and immunomodulating agents = 3 (3) =
M -- musculo-skeletal system 6 (5) 5 (4) 7 (5)
N -- nervous system = 1 (1) 5 (5)
P -- antiparasitic products, insecticides and repellents = = =
R -- respiratory system 2 (1) = 5 (4)
S -- sensory organs 2 (2) = 2 (1)
V -- various 1 (1) 1 (1) =
Other 4 (4) = 1 (1)
Total 41 (37) 15 (13) 43 (36)
Data expressed as number of prescriptions and as number of patients (in parentheses).
:::
Discussion
==========
The results of this study allowed to describe the type of and reasons for medical healthcare use in patients with PD referring to a neurological service dedicated to movement disorders, and to examine the role of patient- and disease-related factors, with particular regard to the pattern of APD treatment. In addition, the changes of APD medications and non-APD prescriptions as consequences of medical healthcare use were also specifically analysed.
A few points need to be preliminarily discussed about our data. First, this was a survey of PD outpatients attending a service devoted to movement disorders. But it also means that only persons who were ambulant or able to reach an outpatient service have been considered, and subjects with H&Y score over IV, which means in the very advanced stages of the disease, have not been included. Second, data collection was accomplished by retrospective self-report of events occurring in the previous 12 months, and it is possible that some respondents had difficulty in accurately reporting such information, even if most of these patients are known to carefully keep the documents concerning their health. Being aware of this potential bias, we sought only information that could be expected to be recalled with some degree of accuracy, therefore excluding e.g. drug prescriptions in hospital. Moreover, whenever possible, we cross-compared reported information with patient records in our service (e.g., in the case of medical healthcare resource use for symptoms, signs and ill-defined conditions and for injuries) and concluded that only negligible, if any, bias was introduced by inaccuracy of collected information.
According to our results, the most utilised medical healthcare resource were specialist visits (41.5% of the patients), followed by GP visits (26.9%), ER admissions (22.3%), and hospital admissions (19.2%). Consultations and hospital admissions for reasons related to PD were *a priori*excluded from our survey. However, according to the records of our service during the previous year, only 14 patients (11.5% of total patients) were admitted to the hospital for reasons related to the disease and requiring more careful clinical observation and drug changes, while the mean year frequency of visits for PD was 2.1 (range: 1--7). Other studies reported on average 5.4 medical visits and 2.7 inpatient hospital days in 6 months \[[@B15]\], or 6.0 physician consultations (including neurologist and other physicians) per year, and no hospital days and ER visits \[[@B8]\]. De Boer et al. \[[@B13]\] showed that nearly half of the patients had at least a GP visit in the previous 6 months. In comparison our patients require less consultations by GPs or non-neurological specialists, possibly due to the preferential consultation of the Centre for PD and Movement Disorders of their reference. On the other side, they seem to use more frequently hospital and ER services, although without any apparent relationship with patient and/or disease characteristics. This finding was quite unexpected, since disease severity is usually pointed to as one of the major predictors of increased healthcare use in PD \[[@B13],[@B20]-[@B22]\]. Another study \[[@B10]\] however reported that disease severity better predicted non medical rather than medical needs.
In the present study, differences according to patient and disease characteristics regarded mainly the specific type of resource used (Table [2](#T2){ref-type="table"}). In particular, patients admitted to hospital or consulting GP were slightly older at onset of PD, while those admitted to ER had longer disease duration, suggesting that need for hospital and/or GP care is related mainly to age, while disease duration is rather involved in acute comorbid events leading to ER admission. Further support to this suggestion comes from the observation that the most frequent causes of ER admissions were injuries (19, 61.3% of total ER admissions), mainly fractures (7, 36.8% of total injuries), which are typical complications in the advanced stages of the disease \[[@B23]\]. A recent study \[[@B24]\] comparing PD with old-age versus middle-age onset evidenced greater motor impairment in subjects with old-age onset, pointing to the possible role of age and comorbidities. On the other hand another study \[[@B25]\] describes similar severity and disability in patients with onset of the disease before 50 years and in persons with PD onset after 50 years, giving relevance to social and psychosocial factors in contributing to the impairment of quality of life. A final consideration can concern the comparison of our data with the ones concerning healthcare utilisation by a sample of 5000 UK population aged 65--90, showing that almost 80% of them visited a GP at least once a year and overall, women used more ambulatory care services and men hospitalised more often \[[@B26]\]. Similar data concerning hospitalisation were found in our cases, although we were not able to confirm the tendency of female patients to seek more often GP assistance. Differences in the overall organisation of the National Healthcare System between Italy and UK might however account for such a discrepancy. Indeed, only in males we observed a correlation between PD duration and the frequency of ER admissions and GP visits, possibly in line with the gender differences in the clinical aspects of the disease, with more severe motor impairment and behavioural problems in men \[[@B27]\].
Although it is well established that PD is often complicated by a burden of comorbid conditions, ranging from psychiatric disturbances \[[@B28]\] to bladder dysfunctions \[[@B29]\], relatively few studies exist which address the issue of comorbidity in PD from a comprehensive point of view. In the present study, we selectively targeted only those comorbid conditions which resulted in the use of medical healthcare resources, to address specifically how comorbidity impacts on medical healthcare use in PD patients. As a consequence, our results reflect the pattern of comorbid diseases which are serious enough to warrant medical intervention, and therefore may have a major impact on the general patient care. In this regard, it is of interest that three categories of disease conditions, namely symptoms and signs and ill-defined conditions, injuries, and diseases of the musculo-skeletal system, account for nearly 40% of all reasons for resource use (Table [3](#T3){ref-type="table"}). Symptoms, signs and ill-defined conditions include mainly conditions such as abdominal pain due to constipation, sleep disturbance, hallucination, headache, dyskinesia, dysphagia, all representing likely complications of the course of PD, as also falls and fractures and lumbago and pain in joint \[[@B1],[@B3]\]. It follows that in these persons more than 1 out of 3 reasons for medical healthcare resource use is justified by PD itself together with its burden of complications, while other conditions typically occurring with high frequency in PD patients, such as dementia, diabetes and cardio- and cerebrovascular disease \[[@B11]\]. On the other hand, psychiatric disturbances (hallucinations) in this sample accounted for only 1 specialist and 1 GP consultation, which seem quite low figures, in comparison to the known frequency and severity of such ailment in PD patients \[[@B12],[@B13]\]. We have no exhaustive explanations for this finding, however it cannot be excluded that the management of the patients by a PD-dedicated centre may contribute to minimize the problems related to the disease and its therapy.
In the present study, we found no apparent correlation between resource use and APD treatments, with the only exception of a positive correlation between levodopa medication and GP visits. However, we believe that this is a spurious association, due to the fact that in our cohort PD patients on levodopa in association with other APD drugs or on DA agonists alone were significantly younger and had an earlier onset of PD than patients on levodopa alone \[[@B14]\]. This finding is in line with the study by Diederich et al. \[[@B24]\], which was performed in a similar population of patients referring to a centre for movement disorders. The preferential use of association treatments, or even of therapy with DA agonists alone \[[@B30]\], is thought to prevent the motor complications associated with long-term levodopa treatment \[[@B31]\]. Interestingly however, in the present investigation no significant difference was found in medical healthcare use according to the various patterns of APD medication, i.e. levodopa alone, levodopa associated with other APD (mainly DA agonists), or DA agonists alone. It cannot be excluded that the fact that the present patient sample refers to a centre for PD and movement disorders may account for an optimisation of individual therapeutic schemes, possibly leading to a reduced risk of complications.
An additional objective of our investigation was the study of drug prescriptions on the occasions of medical healthcare resource use. According to our results, APD treatments were modified in 8.5% of the patients, mainly (55% of the cases) during hospital admissions. Levodopa was the drug most frequently modified (59% of the cases). Considering that in the present population APD treatments account for 307 prescriptions \[[@B14]\], the total number of prescriptions changed (17, 5.5% of total prescriptions) may be considered low. It seems thus that APD treatment remains primarily the responsibility of the neurologist. On the contrary, as for non-APD treatments we found that a total of 99 prescriptions were issued to 86 patients (66% of total patients), which look as very high figures. GP (43% of total prescriptions) and ER (41%) were the main responsible for such prescriptions. While prescriptions on the occasions of ER admissions are more likely to reflect acute and intercurrent needs, prescriptions by GP may reflect the need to treat chronic illness, although this speculation should be specifically addressed in ad hoc studies. Taking in mind that in our previous study \[[@B14]\] we found in this cohort of patients 252 non-APD prescriptions, the present results suggest that physicians other than the reference neurologist are responsible for a significant fraction of such prescriptions, and further emphasise the need for an accurate update of the patient files on the occasions of control visits, as well as for a continuous communication between the reference neurologist and other physicians who from time to time may take care of the patient.
Conclusion
==========
In summary, we have described the type of and reasons for medical healthcare use during 12 months in a cohort of patients with PD referring to a neurological service dedicated to movement disorders. According to our results, major determinants of resource use are programmed hospital admissions and visits, injuries and diseases of the musculo-skeletal system; new drug prescriptions (mainly non-APD drugs) are a common occurrence on the occasions of resource use. Optimisation of the global assistance to patients, including prevention of traumatic accidents, could therefore result effective to reduce and rationalise resource use. The need for careful periodical check of drug treatments is also emphasised.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
MC and EM initiated, designed and supervised the study, analysed the results and wrote the manuscript. DM, DC, GR and CP performed the study. GM, GN and SL provided general advice and contributed to data interpretation. All Authors read, discussed, contributed to and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6963/5/26/prepub>
Acknowledgements
================
This work was supported in part by a grant from the IRCCS C. Mondino (PF Min. San. ICS-57.2/RF.94/124). The helpful collaboration of Dr. Elisabetta Corengia, Dr. Laura Godi, Dr. Olivia Leoni, and Dr. Cristina Oria in collecting and analysing the data is gratefully acknowledged. The Authors are also grateful to Dr. Antonietta Citterio, IRCCS C. Mondino (Pavia, Italy) for her useful suggestions in data management.
|
PubMed Central
|
2024-06-05T03:55:55.289283
|
2005-3-24
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079835/",
"journal": "BMC Health Serv Res. 2005 Mar 24; 5:26",
"authors": [
{
"first": "Marco",
"last": "Cosentino"
},
{
"first": "Emilia",
"last": "Martignoni"
},
{
"first": "Donatella",
"last": "Michielotto"
},
{
"first": "Daniela",
"last": "Calandrella"
},
{
"first": "Giulio",
"last": "Riboldazzi"
},
{
"first": "Claudio",
"last": "Pacchetti"
},
{
"first": "Gianmario",
"last": "Frigo"
},
{
"first": "Giuseppe",
"last": "Nappi"
},
{
"first": "Sergio",
"last": "Lecchini"
}
]
}
|
PMC1079836
|
Background
==========
During the last decade, out-of-hours care by general practitioners (GP) in the Netherlands has changed substantially. Formerly, GPs performed out-of-hours care in small locum groups in which they joined a rota system. In recent years, large GP cooperatives have been set up following British and Danish examples \[[@B1],[@B2]\]. Currently, about 124 GP cooperatives are operational in the Netherlands, taking care of more than 90% of the Dutch population during out-of-hours.
In the current Dutch out-of-hours primary care, roughly two types of organisation models can be distinguished; a separated model and an integrated model. In the separated model the GP cooperative is located separate from the hospital\'s accident and emergency (A&E) department, indicating that there is no functional integration of out-of-hours services. In this model patients with a medical problem can choose between attending the GP cooperative or the A&E department, during out-of-hours. In the second organisation model the GP cooperative is integrated with the hospital A&E department. In this model, all patients utilising out-of-hours primary and emergency care without referral are first seen by a GP or practice nurse. It is known that a substantial number of self-referred patients at the A&E department exhibit minor injuries or non-urgent ailments that can be treated by a GP \[[@B3],[@B4]\]. As a consequence, GPs of the integrated model will generally have to handle more patients than GPs at the separated model \[[@B5]\]. Patients with a referral or brought in by ambulance always bypass this system and will be directed to the emergency department without interference of the GP on call.
The initiative of the out-of-hours primary care reorganisation has come mainly from the medical profession itself, motivated by increased dissatisfaction with the former out-of-hours services. However, as indicated by a recently published systematic review, there is little evidence available on current GPs\' satisfaction with out-of-hours services \[[@B6]\]. The authors of this review identified only one study with respect to GPs\' satisfaction with out-of-hours care. That study showed high levels of satisfaction with cooperative based primary care services \[[@B7]\]. However, only a part of the GPs interviewed actually participated within this service. Other studies have reported of beneficial effects to GPs with the introduction of GP cooperatives for out-of-hours care, like improved GPs\' health \[[@B8]\] or decreased levels of stress \[[@B9]\]. We have also identified one Dutch study that showed increased satisfaction after reorganising out-of-hours care from practice based to cooperative based \[[@B10]\]. No further information is available on GPs\' satisfaction with cooperative based out-of-hours care.
There were two important reasons to conduct this study. First, because GP satisfaction has been shown to be an important contributor to quality of care \[[@B11]\]. GP satisfaction, besides patient satisfaction and costs, should be taken into account when evaluating out-of-hours care services. Second, during the time of the study, the integrated GP cooperative was still in its trial period and insight had to be gained in experiences and opinions of GPs working at this cooperative to support the decision whether this way of organising out-of-hours care should be continued.
The purpose of this study is to gain insight in the satisfaction of GPs with out-of-hours primary care organised in cooperatives. In addition, this study investigates potential differences in the relationship of satisfaction and other out-of-hours care related opinions between GPs working in an integrated model and GPs working in a separated model.
Methods
=======
This study investigates two specific elements: GPs\' satisfaction with the organisation of two types out-of-hours care and the GPs\' opinions related to working at either of two GP cooperatives. Two differently organised out-of-hours cooperatives are involved; a separated model and an integrated out-of-hours care model in two adjacent geographic regions in the Netherlands.
Setting
-------
The separated cooperative is located in the centre of the city of Heerlen, the Netherlands, about 5 km and 9 km away from the only two A&E departments in this region. This cooperative was first set up in 1999, and covered at that time a population of approximately 100,000. In 2001, more GPs joined the cooperative and the population was increased to 278,000. The number of participating GPs increased to 120. In this system patients are stimulated to make a phone call before attending the GP cooperative. This allows the GP cooperative to triage patients at urgency levels of their medical complaints in order to prioritise treatment. During out-of-hours, patients with a medical problem can choose which out-of-hours service to attend, i.e. the GP cooperative or the hospital A&E department.
The integrated GP cooperative is located in the city of Maastricht, the Netherlands, at the region\'s only A&E department of the University Hospital Maastricht. This cooperative was set up in January 2000. During the first one and half year, this cooperative covered only the population of the city of Maastricht (approximately 120,000). In August 2001, the surrounding area of Maastricht also joined the cooperative, increasing the coverage area to 190,000 inhabitants. In total, 83 GPs participate in the integrated GP cooperative. At this GP cooperative patients are allowed to attend the cooperative without an appointment, although it is preferred that they make a phone call first. All patients attending the integrated out-of-hours care facility without referral are first seen by a GP, who refers, if necessary, the patient to the A&E department.
At both GP cooperatives, telephone triage is performed by doctor\'s assistants who are supported by guidelines and protocols, and are supervised by a GP. GPs of these cooperatives perform telephone consultations, consultations at the cooperative, and home visits. Regarding home visits a chauffeured care is at their disposal. Both regions comprise rural as well as urban areas.
Development of the questionnaire
--------------------------------
Topics relevant to out-of-hours primary care were identified in interviews with three GPs participating in the two GP cooperatives under study. We have developed a set of items to enable us to measure and test multi-item scales. The items are related to relevant themes with respect to working at a GP cooperative. In total the questionnaire consisted of 86 items. (Some items are excluded from the analysis because they are only of local interest.) We investigated opinions on: overall satisfaction with the GP cooperative for out-of-hours, reorganisation of out-of-hours care, perceived workload, out-of-hours care as being an essential part of primary care, anonymity of care, gatekeeper function, availability of patient dossiers, cooperation with medical specialists during out-of-hours, and safety. We used a Likert five point scale (strongly agree, agree, neutral, disagree, strongly disagree) to record responses.
Sample
------
In both GP cooperatives (separated and integrated) a random sample of 50 GPs was taken. In case one of these GPs was not able or refused to participate, we had a substitution list of 25 GPs for each cooperative. This list was a random sample of the remaining GPs who were not selected by the first sampling.
The questionnaire was administered by telephone to ensure high response rates. Two research assistants administered the questionnaire and received instructions, prior to the study, by FN. The study was conducted from November 2001 to February 2002.
Statistics
----------
Beforehand, the 86 items of the questionnaire were divided into four blocks. These blocks represented \'satisfaction with out-of-hours care organisation\', \'perceptions and subjective evaluations on working conditions in the present organisation\', \'opinions and beliefs on professional philosophy\', and \'evaluation of the cooperation with medical specialists at the local hospital\'. The most important block concerns the one in which satisfaction with out-of-hours care organisation was measured: this was operationalised with 12 items. Principal component analysis with oblimin rotation was performed on the items of this block and after removal of items with weak factor loadings (lower than +0.60 or -0.60) and/or ambiguously loading items (on more than one factor) two factors remained in analysis. Oblimin rotation can be clarified as an oblique axis-rotation technique in finding the proper factor solution within data. By using this method we wished to come to a more \'natural\' solution. With oblimin rotation no orthogonal assumptions are made to the correlation(s) between factors. In this study we strove for unidimensionality as a prerequisite for independently measured scales. Four items measured satisfaction with the current cooperative (scale 1: response variable) and three items measured satisfaction with the state of affairs before the cooperative was set up (out-of-hours care in a rota system). Next, per intended scale the test stability of each factor was measured by Cronbach\'s alpha, and again items could be removed from this scale, if this did increase the value of the alpha coefficient. In constructing the scale \"overall satisfaction with the GP cooperative\" one of the four items was removed following inspection of the Cronbach\'s alpha, while all four had high loadings on the same factor in the principal component analysis. The item was left out of table [2](#T2){ref-type="table"} and concerned the notion that the current organisation should be restructured. Alpha went up from 0.852 on four items to 0.893 on three items. In constructing the other scales, items which were loading unambiguously high in the principal component analysis were all included after being tested on reliability by Cronbach\'s alpha.
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
GP questionnaire. Description of scales and items. (Original items are in Dutch\*)
:::
--------------------------------------------------------------------------------------------------------------------------------------
**BLOCK 1. Satisfaction with out-of-hours care organisation**
**Scale 1**. Overall satisfaction with GP cooperative (Cronbach\'s α = 0.90; mean (SD) = 65.0 (21.6))
I am very satisfied about the functioning and the structure of the GP cooperative (+)
I am pleased with the current GP cooperative (+)
I am very satisfied with the current organisation of out-of-hours primary care (+)
**Scale 2**. Current out-of-hours care is better organised than formerly (Cronbach\'s α = 0.92; mean (SD) = 89.8 (19.5))
Out-of-hours care was better organised before the establishment of GP cooperatives (+)
I prefer the former organisation of out-of-hours primary care (-)
The current organisation of out-of-hours care is not an improvement compared to the former organisation (-)
**BLOCK 2. Perceptions and subjective evaluations on working conditions in the present organisation**
**Scale 3**. Experienced a high workload (Cronbach\'s α = 0.87; mean (SD) = 65.5 (18.4))
The workload at the GP cooperative is too high (+)
Out-of-hours care during daytime on Saturday and Sunday is very aggravating (+)
Usually, out-of-hours service is much too aggravating (+)
I do not experience such a high workload at the GP cooperative (-)
Performing out-of-hours care is absolutely not aggravating (-)
Out-of-hours care during daytime in the weekends is not too high (-)
**Scale 4**. One feels safe at the cooperative (Cronbach\'s α = 0.87; mean (SD) = 76.9 (18.7))
Sometimes, I feel unsafe at the GP cooperative during out-of-hours (-)
During my shifts at the GP cooperative, I never feel unsafe (+)
Regularly, I feel unsafe at the GP cooperative during out-of-hours (-)
**Scale 5**. One feels safe during home visits (Cronbach\'s α = 0.84; mean (SD) = 76.0 (18.4))
Regularly, I feel unsafe when performing home visits during out-of-hours (-)
Sometimes, I feel unsafe when performing home visits during out-of-hours (-)
Usually, I feel safe when performing home visits during out-of-hours (+)
**BLOCK 3. Opinions and beliefs on professional philosophy**
**Scale 6**. Out-of-hours care is an essential part of primary care (Cronbach\'s α = 0.97; mean (SD) = 60.3 (31.7))
These days, out-of-hours care should no longer be an essential part of primary care (-)
Out-of-hours care is definitely an essential part of primary care (+)
Out-of-hours care should always be a part of general practice (+)
There is no place anymore for out-of-hours care in general practice (-)
**Scale 7**. Anonymity of care is a problem (Cronbach\'s α = 0.90; mean (SD) = 32.9 (21.8))
Because the GP of the cooperative and the patient are not familiar with each other, there is a risk for inadequate treatment (+)
Because of anonymity of care there is a risk that diagnostics and treatment are not adequately adjusted for the patient\'s needs (+)
One of the big disadvantages of the GP cooperative is the anonymity of care, because the GP is not familiar with the patient (+)
Because the GP of the cooperative and the patient are not familiar with each other, there is a risk for inadequacy of care (+)
**Scale 8**. Gatekeeper function is well guaranteed (Cronbach\'s α = 0.74; mean (SD) = 64.5 (16.8))
I think the GP gatekeeper function at the GP cooperative is well guaranteed (+)
I am afraid that the GP gatekeeper function during out-of-hours will disappear (-)
I believe that in the near future the GP gatekeeper function will be put under too much pressure (-)
I think the GP gatekeeper function at the GP cooperative is well protected (+)
**Scale 9**. Availability of patient dossiers is important (Cronbach\'s α = 0.93; mean (SD) = 57.4 (24.1))
As far as I am concerned, I do not need the availability of my colleagues\' patient dossiers (-)
I think it is a serious problem that my colleagues\' patient dossiers are not at my disposal (+)
The availability of patient dossiers of the other participating GPs during out-of-hours is absolutely unnecessary (-)
During out-of-hours I am hindered in my practice, because of lack of information about my colleagues\' patients (+)
**BLOCK 4. Evaluation of the cooperation with medical specialists at the local hospital**
**Scale 10**. Cooperation with medical specialists is good (Cronbach\'s α = 0.94; mean (SD) = 62.7 (23.0))
Sometimes, the cooperation between GPs and medical specialists is not so good (-)
I think that during out-of-hours the understanding between GPs and medical specialists is sometimes pretty bad (-)
I think that during out-of-hours the cooperation between GPs and medical specialists is always fine (+)
Generally, the cooperation between myself and the medical specialists of the hospital is good (+)
--------------------------------------------------------------------------------------------------------------------------------------
\* The provisional translation into English is meant to inform the reader of the content of the scales and cannot be seen as a definite one.
:::
Scale constructions were performed under specific rules for missing item data: in summating to a total for each case, scores had to be valid on at least half of the items, if the number of items was even, and on at least half of the items plus one half, if the number of items was uneven. Otherwise, scale scores were set to \'missing\'. Finally, a transformation of the total scale score to a 0--100 score was made \[[@B12]\]. After that, the remaining three blocks were analysed in a similar way. In total this procedure produced ten scales.
The relationship between individual scales and overall satisfaction (scale 1) was analysed using multiple regression analysis. Upon finding the final \'direct effects\' model for both regions of interest we have tested all possible interactions of pairs of significant predictors for statistical significance by the forward selection technique. Next to this, interactions were computed by multiplying predictors significant within the \'direct effects\' model by the dummy 0--1 (Heerlen or Maastricht GP cooperative) coding of region. In case of missing data, listwise deletion of missing cases was applied. To test differences between GPs from either two GP cooperatives we performed independent Student\'s t-tests per scale. In case of non-normality, which was assessed visually by histogram analysis and by the Kolmogorov-Smirnov test, a Mann-Whitney test was used. A *P*-level of less than 0.05, was considered to be statistically significant. All data were analysed using SPSS-pc, version 10.0.5.
Results
=======
In total 100 GPs participated; 50 GPs per each cooperative. One respondent of the Maastricht GP cooperative (integrated model) refused to participate and was substituted by a GP from the reserve list. The mean duration of the interviews was 22 (± 6.6) minutes. The characteristics of the respondents of both models do not differ statistically (Table [1](#T1){ref-type="table"}).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Characteristics of respondents.
:::
GP cooperative Heerlen (n = 50) GP cooperative Maastricht (n = 50)
------------------------ -------------- --------------------------------- ------------------------------------
Age 48.0 ± 7.5 47.3 ± 6.6
Gender Male 42 43
Female 8 7
Employed Part-time 11 16
Fulltime 39 34
Size of practice (GPs) Mean (range) 2.5 (1 -- 6) 2.0 (1 -- 7)
Participation in GP
cooperative Fully 37 37
Partly 13 13
:::
We tested ten scales related to aspects of current out-of-hours primary care (Table [3](#T3){ref-type="table"}.). Internal reliability of these scales was considered appropriate; Cronbach\'s alpha\'s ranging from 0.74 to 0.97. An overview of all scales and related items is presented in Table [2](#T2){ref-type="table"}.).
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Scales scores for GPs opinions on different aspects of out-of-hours primary care.
:::
Separated Model Integrated Model
------------------------------------------------------------------ --------------------- --------------------- --------------
(Heerlen) (Maastricht)
Scale score Scale score Significance
Scale Mean (95% CI) Mean (95% CI)
Overall satisfaction 70.0 (64.0 -- 76.0) 60.0 (54.0 -- 66.0) 0.020
Current out-of-hours care is better organised than formerly^b,†^ 90.1 (84.7 -- 95.5) 89.5 (83.7 -- 95.3) 0.569\*\*
Experience a high workload^b^ 63.4 (58.1 -- 68.8) 67.6 (62.5 -- 72.7) 0.256
One feels safe at the cooperative^b^ 76.5 (71.1 -- 81.9) 77.3 (72.0 -- 82.7) 0.825
One feels safe during home visits^b,‡^ 74.8 (68.7 -- 81.0) 77.2 (73.1 -- 81.4) 0.532
Out-of-hours care is an essential part of primary care^b^ 52.9 (43.7 -- 62.1) 67.8 (59.4 -- 76.1) 0.018
Anonymity of care is a problem^b^ 31.8 (25.1 -- 38.4) 34.0 (28.3 -- 39.7) 0.609
Gatekeeper function is well guaranteed^b,†,‡^ 66.8 (62.2 -- 71.2) 62.2 (57.1 -- 67.3) 0.180
Availability of patient dossiers is important^b^ 55.4 (48.4 -- 62.4) 59.3 (52.6 -- 66.0) 0.422
Cooperation with specialists is good^b,†^ 48.5 (42.1 -- 54.8) 76.6 (72.9 -- 80.4) \<0.001
\* Scale score ranges from 0 to 100 points
^b^100 points represents strong agreement
^†^One case missing at the separated model, ^‡^one case missing at the integrated model
\*\* Because of non-normal distribution, the Mann-Whitney test was used
:::
GPs\' overall satisfaction score with the current organisation of out-of-hours care was 65 points (95%CI: 60.7 -- 69.3) on a scale from 0 (absolutely not satisfied) to 100 (highly satisfied). However, GPs from the separated model were more satisfied compared to their colleagues of the integrated model (scale score 70.0 vs. 60.0; *P*= 0.020). GPs from both cooperatives reported that the new organisation of out-of-hours primary care is better compared to the former practice-based out-of-hours care (mean scale score 89.8). Most GPs experience a high workload (mean scale score 65.5). A minority of all interviewed GPs think that the anonymity of patient care -- many patients are not known to the GP because care is organised on large-scale -- endangers adequacy of out-of-hours primary care (mean scale score 32.9). Furthermore, a small majority feels that the patient\'s medical file should be available at the cooperative (mean scale score 57.4). In both cooperative models (integrated and separated) of out-of-hours care, GPs think that their gatekeeper\'s role to secondary care is sufficiently guaranteed (mean scale score 64.0). Most GPs feel relatively safe at the cooperative or during out-of-hours home visits (mean scale score 76.9 and 76.0 respectively).
GPs from the separated model were neutral about out-of-hours care as being an essential part of their job as a GP, in contrast with the integrated model GPs who were more convinced that out-of-hours care is an important part of their job (scale score 52.9 vs. 67.8; *P*= 0.018). GPs at the integrated model experience a better cooperation with medical specialists during out-of-hours care (scale score 76.6 vs. 48.5; *P*\< 0.001).
The regression analysis identified three scales that are significantly related to overall satisfaction (Table [4](#T4){ref-type="table"}.). Effects on satisfaction for two of these scales, experienced workload and whether the GP thinks that his gatekeeper function is well guaranteed during out-of-hours, are different for both cooperatives. Experienced workload is mainly related to overall satisfaction of GPs from the integrated model. Whereas, the GP\'s opinion about the gatekeeper function is mainly related to overall satisfaction of GPs from the separated model. Increased experienced workload will lead to a decreased overall satisfaction, and the better the GPs valued the guarantee of their gatekeeper function during out-of-hours the higher their overall satisfaction will be.
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Regression analysis with overall satisfaction with the organisation of out-of-hours care as dependent variable (0 = not satisfied, 100 = very satisfied) (n = 98; R^2^= 0.36).
:::
Unstandardised coefficients Standardised coefficient
-------------------------------------------------------- ----------------------------- -------- -------------------------- -------- --------------
Scales B SD Beta t Significance
Constant 106.352 22.154 4.801
Age -0.034 0.262 -0.012 -0.130 0.897
Gender ^b^ -2.188 5.228 -0.037 -0.418 0.677
Cooperative ^a^ -75.980 26.073 -1.840 -2.914 (0.005)
Gatekeeper function 0.101 0.149 0.081 0.677 (0.500)
Out-of-hours care is an essential part of primary care -0.140 0.063 -0.212 -2.216 0.029 ^c^
Experienced workload -0.597 0.152 -0.518 -3.915 (\< 0.001)
Gatekeeper \* cooperative 0.737 0.247 1.253 2.980 0.004^c^
Workload \* cooperative 0.487 0.219 0.799 2.225 0.029^c^
^a^Cooperative: Maastricht = 0; Heerlen = 1; ^b^gender: male = 0, female = 1 ^c^Only effects that are interpretable
Gatekeeper function (0 = not guaranteed, 100 = highly guaranteed)
Out-of-hours essential part (0 = not essential, 100 = highly essential)
Experienced workload (0 = very low, 100 = very high)
:::
The third scale that is significantly related to overall satisfaction is the GP\'s opinion about out-of-hours care as being an essential part of his task as a GP. GPs who indicated that they believed that out-of-hours care was an essential part of their job as a primary care physician had a lower overall satisfaction with respect to current out-of-hours care. Neither gender nor age was significantly related to overall satisfaction. The regression model explained 36% of the variation in overall satisfaction.
Subgroup regression analysis for GPs of the integrated and separated model separately (see Table [5](#T5){ref-type="table"}), showed that for the GPs of the integrated model workload was the main factor that influenced overall satisfaction (variance explained: 34%). With respect to the GPs of the separated model, the guarantee of the gatekeeper function was of great importance to the overall satisfaction (variance explained: 35%).
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
Regression analysis results on overall satisfaction with the organisation of out-of-hours care as dependent variable for both GP cooperatives separately (0 = not satisfied, 100 = very satisfied).
:::
Unstandardised coefficients Standardised coefficient
-------------------------------------------------------- ----------------------------- -------- -------------------------- -------- --------------
B SD Beta t Significance
**Separate GP cooperative (R^2^= 0.35) n = 49**
Constant 21.823 20.437 1.068 0.291
Age -0.122 0.366 -0.043 -0.333 0.741
Gender ^a^ -0.110 7.475 -0.002 -0.015 0.988
Gatekeeper function 0.956 0.201 0.697 4.755 \<0.001
Out-of-hours care is an essential part of primary care -0.182 0.098 -0.273 -1.851 0.071
**Integrated GP cooperative (R^2^= 0.28) n = 50**
Constant 101.692 23.204 4.383 \<0.001
Age 0.073 0.398 0.023 0.182 0.856
Gender ^a^ -9.375 7.335 -0.156 -1.278 0.208
Experienced workload -0.648 0.144 -0.551 -4.496 \<0.001
^a^Gender: male = 0, female = 1
Gatekeeper function (0 = not guaranteed, 100 = highly guaranteed)
Out-of-hours essential part (0 = not essential, 100 = highly essential)
Experienced workload (0 = very low, 100 = very high)
:::
Adding the scale of experienced workload to the regression equation of the Separate GP cooperative in Table [5](#T5){ref-type="table"} will make variance explained in overall satisfaction only higher by 0.007 to 0.359 (F ratio of the change = 0.50 by 1 and 43 df., p = 0.482). Adding the scales of gatekeeper function and out-of-hours care as an essential part of primary care to the regression equation of the Integrated GP cooperative will make variance explained in overall satisfaction only higher by 0.036 to 0.317 (F ratio of the change = 1.13 by 2 and 43 df., p = 0.332).
Discussion
==========
GPs in this study are generally satisfied with the way out-of-hours primary care is currently organised. However, GPs from the separated cooperative are more satisfied than GPs working at the integrated cooperative. Mainly three factors are related to overall satisfaction. These are: experienced workload, guarantee of the gatekeeper function, and attitude towards out-of-hours care as being an essential part in general practice.
To our best knowledge, this is the first study to investigate GPs\' satisfaction with out-of-hours care as organised in separated and integrated primary care cooperatives. One British study and one Dutch study have looked into GPs satisfaction with out-of-hours care\[[@B7],[@B10]\]. However, these studies solely focused on separated out-of-hours care models. The Dutch study showed that 70% of the GPs were satisfied with cooperative based out-of-hours care \[[@B10]\], and the British study found that 92% of the GPs were satisfied with the way out-of-hours care was arranged \[[@B7]\]. Since these studies used different ways to measure satisfaction it is difficult to compare them with our results.
The results of this study indicate a difference in satisfaction between GPs from the separated and integrated cooperative. A possible explanation for this difference could be the fact that the integrated cooperative has to deal with a larger number of patients compared to the separated cooperative \[[@B5]\]. In this study however, experienced workload of GPs from the integrated cooperative did not differ from that of the GPs of the separated one. Obviously workload is also dependent on staffing of the cooperative. We presume that the difference in satisfaction might well be explained by other factors, which we have not investigated in this study. At the time of the study the integrated model was still in its experimental phase; housing in the integrated cooperative was generally not considered to be optimal. In this phase of the experiment the waiting room was very small and quickly overcrowded. Also, the space in the doctor\'s offices was quite limited and contained only room for one bed and no desk. In addition, at this time also patients\' and GPs\' privacy were not as sufficiently guaranteed as in the separated model. These factors may have had an effect on GPs\' overall satisfaction with out-of-hours services.
Three opinions were found to be significantly related to GP satisfaction with the organisation of out-of-hours care. The two opinions that weighted most heavily on satisfaction were experienced workload and gatekeeper function.
We found that the GPs\' opinion on the gatekeeper function during out-of-hours was related to satisfaction with the organisation of out-of-hours care specifically for GPs of the separated GP cooperative. The fact that this opinion is not related to satisfaction with the organisation in the integrated cooperative is probably due to the fact that this is not an issue at this cooperative, because the GP\'s gatekeeper function is fully guaranteed; all patients entering the out-of-hours centre are screened by a GP and if necessary referred to a medical specialist. In the separated model however, the patient can still bypass the GP and attend the emergency department of the hospital without a GP\'s referral. GPs who feel to have too little grip on these self-referring patients appear to be less satisfied with their arrangements of out-of-hours care.
We have not been able to investigate GPs\' satisfaction prior to the reorganisation from practice based out-of-hours care to cooperative based out-of-hours care, whilst GPs\' dissatisfaction with practice based out-of-hours care was one of the important reasons why primary care in the Netherlands was reorganised. Nevertheless, this study shows that GPs feel that current out-of-hours primary care is better organised compared to former practice-based out-of-hours care. These results are in line with previous research \[[@B10]\]. However, this is not surprising considering the effort that has gone into reorganising the out-of-hours services and the prior dissatisfaction. All those who were in favour of the change of the out-of-hours system will obviously be satisfied with the fact that out-of-hours care has been reorganised, and feel that the new system is better than formerly.
A distinct feature of the integrated model is the close cooperation between primary and hospital emergency care. This offers possibilities to improve communication and to exchange expertise. This is reflected by the high satisfaction score of GPs from the integrated model with the cooperation with the medical specialists of the hospital. Because GPs and medical specialists now work at the same site, it is easier to consult each other. Furthermore, GPs who have referred a patient to one of the medical specialists have access to feedback, i.e. they can check on the patient a few minutes later to see if they were right in their diagnosis. Nevertheless, region-specific differences may also have accounted for this difference, because in the region with the integrated cooperative there is a longer tradition in cooperation between primary and secondary care.
We investigated GPs\' opinions on working at two contrasting models of out-of-hours primary care, i.e. a separated and an integrated GP cooperative, in order to gain insight in GPs\' preferences for either one of these models. Until November 2001, the Maastricht GP cooperative for out-of-hours care was the only cooperative in the Netherlands that was integrated with a hospital A&E department. Consequently, the Maastricht out-of-hours care organisation was the only service at the time of the study that could be used as an example of integrated out-of-hours care.
There are limitations to generalise the results of the study to other regions. First, results of the study reflect the opinions of GPs at only two cooperatives in the South of the Netherlands. Second, the integrated GP cooperative was still in its trial phase and may therefore have not been a good representative of a well-established GP cooperative. Nevertheless, this is the first study to address GP satisfaction with an integrated GP cooperative and may therefore, despite the limited generalisability, give some indication of relevant aspects of integrated out-of-hours care for further research and care development.
Currently, three regions in the Netherlands are working according to an integrated out-of-hours care system. However, at the moment GPs in other regions consider adopting this organisational structure. Furthermore, the current Dutch minister of health care has stated to be in favour of an intensive collaboration between primary and emergency care, for this will probably reduce costs \[[@B13]\]. The results of this study can support the current discussion in the Netherlands on the organisation of out-of-hours primary care.
Future research should focus on the economic efficiency of both models and patient preference with respect to the organisation of out-of-hours primary care, because these are important features to take into account when developing out-of-hours care.
Conclusion
==========
GPs in this study appear to be generally satisfied with the organisation of GP cooperatives for out-of-hours care. Furthermore, GPs working at the separated cooperative seem to be more satisfied compared to GPs working at the integrated cooperative.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
CvU, FN, GV, RW, GW, and HC participated in the design of the study. CvU, FN and GV developed the questionnaire. CvU and FN performed statistical analysis. CvU drafted the manuscript. FN, GV, RW, GW, and HC provided critical edits to this manuscript. HC supervised the study. All authors have read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6963/5/27/prepub>
Acknowledgements
================
We would like to thank all GPs who participated in this study. This study was financially supported by a grant from the Dutch Health Insurance Board (registered as VAZ/CVZ project nr. 00103). The funding organisation had no role in the design, data collection, data analysis, interpretation of the data, and in the preparation, review, or approval of the manuscript.
|
PubMed Central
|
2024-06-05T03:55:55.293126
|
2005-3-31
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079836/",
"journal": "BMC Health Serv Res. 2005 Mar 31; 5:27",
"authors": [
{
"first": "Caro JT",
"last": "van Uden"
},
{
"first": "Fred HM",
"last": "Nieman"
},
{
"first": "Gemma BWE",
"last": "Voss"
},
{
"first": "Geertjan",
"last": "Wesseling"
},
{
"first": "Ron AG",
"last": "Winkens"
},
{
"first": "Harry FJM",
"last": "Crebolder"
}
]
}
|
PMC1079837
|
Background
==========
Leprosy, or Hansen\'s disease (HD), is a chronic infectious disease caused by *Mycobacterium leprae*\[[@B1]\]. The disease has been known since ancient times: The origin of the word leprosy dates back to Greek and Latin\[[@B2]\]. Over a long time period, the disease can be disfiguring, and societies have stigmatized victims of the disease. This attribute is deeply discrediting since the stigmatized individual is disqualified from full social acceptance. Leprosy was thus dreaded, not because it killed, but because it left one alive with no hope\[[@B3],[@B4]\]. After Armauer Hansen\'s discovery of the bacteria in 1873, the disease became feared as a contagion, and segregation was recommended for prevention. The advent of effective drugs in the 1940s drastically changed the course of disease, and in many countries, patient isolation was deemed unnecessary and terminated. However, some countries that practiced patient isolation experienced difficulties in changing their policies, abolishing leprosaria and reintegrating the patients into the community\[[@B5]\]. As a result, prolonged policies caused the institutionalization of patients for a period longer than scientific knowledge justified.
Public policy termination is often perceived as the final outcome of a political, but highly rational, policy process. When a policy\'s objectives are reached and maintained, its relevance and applicability should be reconsidered and, if found redundant, outmoded, or dysfunctional, the policy should be terminated\[[@B6]\]. Termination might occur with either a bang or a whimper: public programs and policies may end suddenly, perhaps after a lengthy resistance, or they may end slowly after a long-term decline in the resources with which they are \[[@B7],[@B8]\]sustained. In either case, when termination does not take place properly, the policy may cause harmful effects, either materially or ideologically, rather than beneficial ones\[[@B9]\]. Adopting a policy to terminate a law is always a struggle presenting a number of obstacles, especially. Those posed by agents with vested interests\[[@B10]\]. To overcome various political hurdles, a set of effective political strategies is sometimes required, as well as political maneuvering by those skilled at policy \[[@B11]-[@B13]\]termination.
There seems to be a general expectation that the introduction, maintenance and termination of health and medical policies are deliberate, meaning both rational and scientifically justifiable. However, the relationship between science and policy is actually quite complex. It is rarely a linear and rational one in which science finds and establishes the facts and then policymakers incorporate them to solve social problems. This study examines the history of leprosy control policy in the US and Japan, focusing especially on the national legislative changes concerning patient isolation. It seeks to examine the functions of science in policy change, in this case the abolition of the long-standing patient isolation policy. Importance of favorable political context, along with skillful policy entrepreneurs, is then addressed.
Methods
=======
Materials
---------
To examine the process of policy making concerning the isolation of patients with leprosy / HD, a comprehensive search and collection of historical documents were conducted both in the US and Japan. After a thorough search of archives of public records, both printed and on-line, using the keywords leprosy and Hansen\'s disease, relevant documents were exhaustively collected from libraries, leprosaria/sanatoria, offices of patients\' organizations, and the museum of Hansen\'s disease (Carville and Tokyo) in each country. Sources include the Research Library of the National Archives and Record Administration (Washington)\[[@B14],[@B15]\], the libraries at Louisiana State University (Baton Rouge) and Harvard University (Cambridge and Boston), and the National Diet Library (Tokyo). Databases of medical articles, such as MEDLINE\[[@B16]\] and ICHUSHI\[[@B17]\], were utilized to search for publications on leprosy/HD. Other types of documents and books, such as theses, essays and recollections, were also gathered. Supplementary interviews were then conducted during 2000--2004 with leprosaria directors and physicians, administrators, and patients, both past and present, in the US and Japan.
The Garbage Can model of policy making
--------------------------------------
The observation of politics involves making observations conform to sets of assumptions, which are called models. These models help delineate the relationships among conditions and patterns in political life. The Garbage Can model of policy making was proposed by Cohen et al\[[@B18]\], and further developed by Kingdon\[[@B19]\] as a contrast to linear, comprehensive, and rational models\[[@B20]\]. This model assumes that policy windows open only when the process streams of problems, policies and politics converge, separately and independent of each other. In such a convergence, problems are brought to the attention of people in and around government by systemic indicators, by focusing events like crisis and disasters, and by feedback on the operation of current programs. Alternate policy proposals are developed from the many possible ideas floated by those both in and out of government. The proposals that survive to achieve serious consideration must meet several criteria, including technical feasibility, a fit with dominant values and the current national mood, budgetary workability and political support. Political support may be affected by a variety of influences including swings of national mood, administrative or legislative turnover, interest group campaigns, and social movements.
In this context, policy windows are the opportunities to push forward one\'s preferred proposal or conception of the problem. These windows of opportunity in policy making can open either as a result of happenings in the political stream or by the development of compelling problems. In the Garbage Can model, solutions and problems have equal status as separate streams in the system, and the popularity of a given solution at a given point in time often influences which problems are focused on. In other words, collections of choices are looking for problems, issues and feelings are looking for decision situations in which they might be aired, solutions are looking for issues to which they might be the answers, and decision makers are looking for work. Agendas are set by problems or politics and alternatives are generated in the policy stream. So, outcomes are heavily dependent on the coupling of the streams. Sometimes in this process, policy entrepreneurs -- people who invest their resources in pushing their pet proposals or problems -- are responsible, not only for prompting important people to pay attention, but also for coupling solutions to problems and for setting both problems and solutions in a functional political context.
This Garbage Can model was sometimes employed to explain the dynamics of policy making in the areas other than health care, and proved to be a powerful analytical tool\[[@B21]\]. In this study, the courses of events were analyzed through this model, and important factors for and obstacles to linking science and policy were examined.
With respect to leprosy policy, history after WWII is divided into three periods. Period I covers the time period of the late 1940s through 1950s, when the advent of new drugs could have opened the policy windows in both countries. Related events in Japan seem to have lagged about five years behind those in the US, presumably due to the delayed introduction of those drugs. Period II represents the time in the 1960s and 1970s, when true legislative policy change did not occur, but incremental administrative actions were taken. Finally, Period III encompasses the 1980s and 1990s, when the policy of patient isolation was legislatively abolished in both the US and Japan. Relevant facts preceding Period I are also succinctly presented.
Results
=======
Isolation of leprosy patients in the United States
--------------------------------------------------
### US background
In 1865, Hawaii established a policy for the quarantine and isolation of HD patients. A leper settlement was established at Kalaupapa on Molokai island, and the Kalihi Receiving Station was opened on Honolulu to diagnose cases and provide emergency treatment\[[@B22]\]. In 1873, when *Mycobacterium leprae*was discovered by Hansen and leprosy was thus proved to be a contagion, the seclusion of patients in their homes or in hospitals was advocated for disease prevention\[[@B23]\]. By 1894, Louisiana had established a leprosarium, and several other states, such as New York and California, had also introduced similar isolation policies for HD patients\[[@B24]\]. The First International Congress on Leprosy held in Berlin in 1897 recommended isolation as the appropriate policy measure against leprosy. In 1909, the U.S. government built another leper colony in the Philippines on the Island of Culion.
After several national and state surveys of HD patients, the need for a national policy to control HD and the desirability of patient isolation were discussed repeatedly during the 1890s. The first legislation, passed in 1898, authorized a thorough investigation of leprosy. A commission was appointed, and based on its recommendation, legislation was introduced that allowed for the establishment of a national leprosarium for the segregation of lepers with the goal of preventing the spread of leprosy in the United States\[[@B25]\]. It is worth noting that scientists were never unanimous about the necessity of isolation. Scientific publications at the time described HD as a relatively non-contagious disease\[[@B26]\].
After succeeding reports and debates, legislation was passed in 1917 that provided for purchase of a site and receipt of any person afflicted with leprosy who presented himself or herself for care, detention, and treatment, or who might be apprehended under authority of the United States Quarantine Acts\[[@B27]\]. The American Academy of Medicine, the American Medical Association, and the American Dermatological Association expressed their support for the isolation policy, arguing that patient isolation serves for disease control and patient treatment, as well as protection of patients from social stigma and ostracism. The 1917 legislation also directed that public health employees assigned to the institution be paid one and one-half times the regular pay due to the disagreeable and dangerous nature of the service\[[@B28]\].
Having failed to locate a new site, in 1921 the U.S. government took over the Louisiana Leper Home in Carville, Louisiana, and the United States Public Health Service (PHS) opened the US Marine Hospital Number 66 as a National Leprosarium\[[@B29]\]. The rules \"Regulations governing the care of lepers: regulations for the government of leprosaria and for the apprehension, detention, treatment, and release of lepers\" were drawn up by the Surgeon General in 1922 to implement the 1917 legislation. The 1922 rules reiterated the admission options (voluntary or involuntary), prohibited patients from leaving the sanitarium grounds, and provided that patients shall not \"hold communication\" with patients of the opposite sex. The rules also outlined a procedure for discharge of patients, which included a period of one year of special observation for any signs of leprotic retrogression after medical tests showed no signs of the disease. After this year, the patients were requested to appear before a board of three or more medical officers who might pronounce the patient to be no longer a menace to public health\[[@B30]\].
The institution was gradually expanded in the year 1933, having 65 rooms and outpatient facilities in 1941. By 1945, there were 369 inpatients\[[@B31]\]. During those years, there was no really effective treatment of the disease. Most of those at the facility were involuntarily admitted (only 15% of admissions between 1930 and 1945 were reportedly voluntary)\[[@B32]\].
### US period I (1940 through 1950))
In 1941, Dr. Guy Faget at the National Leprosarium started a clinical trial of Promin, a sulfone drug, for the treatment of HD. Its beneficial effects became rapidly known and were reported within a few years in leading medical journals, such as Public Health Reports and Journal of the American Medical Association (JAMA)\[[@B33]\]. Other forms of sulfone drugs, Diasone and Promazole, were also developed and replaced the traditional HD treatment of applying chaulmoogra oil\[[@B34]\]. The international academic community paid close attention, and the 1946 Conference on Leprosy in Rio de Janeiro reported significant advances from the introduction of the sulfone group of drugs\[[@B35]\].
The enactment of the Public Health Services Act of 1944\[[@B36]\] could have provided a forum for discussion and revision of the US HD policy. Parts D and G of the Act, however, literally duplicated the 1917 Act, apparently without discussion. As before, its Section 331 states -- \"that the Service (PHS) shall, in accordance with regulations, receive into any hospital of the Service suitable for his accommodation any person afflicted with leprosy who presents himself for care, detention, or treatment, or who may be apprehended under section 332 (regulation) or 361 (quarantine) of this Act, and any person afflicted with leprosy duly consigned to the care of the Service by the proper health authority of any State, Territory, or the District of Columbia. The Surgeon General is authorized, upon the request of any health authority, to send for any person within the jurisdiction of such authority who is afflicted with leprosy and to convey such person to the appropriate hospital for detention and treatment.\"
Only a year later, in 1945, the American Public Health Association advised against isolating people with leprosy. It was by then well established that the majority of the population had a natural immunity to leprosy, and that it was only mildly contagious to the rest. Since the relative infectiousness of the two types of leprosy had been measured by epidemiological studies, which demonstrated the very low infectiousness of the tuberculoid type, experts insisted that only open (infectious) cases, if any, required isolation\[[@B37]\]. However, at that point, scientists were again not unanimous concerning patient isolation policy. Dr. Faget, who should have known the effects of sulfones better than anyone else, wrote in Public Health Reports that the only reliable means to eradicate leprosy is isolation. Dr. G. W. McCoy, an ex-official of the PHS and Dean of the Louisiana State University Medical School, on the other hand, recommended that since infectivity varies depending on clinical types, the universal isolation policy should be abandoned. He stressed that when there is room to question infectiousness, it should be used to serve patients (meaning to protect patients\' liberty)\[[@B38]\].
In response to these debates, the Federal Security Agency (FSA) set up under the Surgeon General the National Advisory Council on Leprosy. The Council discussed the issues raised by the United Patients\' Committee for Social Improvement and Rehabilitation -- composed of the Patients\' Federation, Veterans/Legionnaires, and the staff of *The Star*(a monthly periodical published by the HD patients) -- which had expressed its objection to the Public Health Services Act. Abolition of forced isolation and detention and the broadening of outpatient treatment services were discussed, but only the latter was supported\[[@B39]\]. After a while, an experimental project for outpatient services was established in New Orleans.
Efforts to revise the law began soon after the enactment of Public Health Services Act. In 1948, G. H. Rarey (US Army -- retired Colonel, national vice president of the American Federation of the Physically Handicapped) and Paul Strachan (president, American Federation for the Disabled) drafted the National Leprosy Bill, and asked Congressman Charles J. Kersten (R-Wisconsin) to submit it to Congress\[[@B40]\]. In the following year, a similar bill, the National Leprosy Act was submitted by Senator Claude D. Pepper. Both of these draft legislation intended drastic changes in leprosy control policy, including: dissemination of pertinent facts concerning leprosy (to promote an enlightened public opinion, a new and more accurate understanding of leprosy); treatment of leprosy patients (to establish leprosy treatment principles, methods, and regulations for administering leprosy treatment more in harmony with the customs of society as applied to the care and treatment of persons afflicted with other diseases; to arrange treatment centers, both inpatient and outpatient services); the National Advisory Council on Leprosy; rehabilitation and reemployment of leprosy patients; financial assistance for leprosy patients and their dependents; compensation for disability incident to leprosy; and expansion of leprosy research.
At the Congressional hearings on the bills, the experts\' testimony took different positions\[[@B41]\]. As a drafter of the bill, Colonel Rarey, who had served in the Philippine Islands and encountered many leprosy patients, argued insistently against the segregation of leprosy patients, claiming that it was not effective, could destroy patients\' social lives, and might hinder patients from seeking proper treatment. He stated, \"Many cases of this disease are not communicable. The remainders are classified by leprologists as feebly communicable. The status and low degree of communicability of leprosy in the US does not justify continuance of the present compulsory segregation policies.\" He also criticized the PHS, arguing that the FSA\'s favoring of segregation erred on the ultraconservative side of this question.
On the other side, J. Donald Kingsley, Acting Administrator of the FSA, presented opposition to the bill. Although he indicated sympathy with the basic objectives of the bill, the necessity for law revision was denied. He argued that most of the objectives of the bill could be more effectively achieved through intensification of activities already authorized by the existing law. The bill was criticized as defective in attempting to divide responsibility for the treatment and care of patients not hospitalized in PHS hospitals, between the PHS and the various federal, state, and local governmental agencies. Especially stressed was the point that the proposed Act does not meet squarely the problems of forced detention of leprosy patients: It was insisted that the existing authority of the PHS should not be repealed or cast in doubt without careful consideration and provision for alternative methods of meeting the problem, should need arise. The FSA, while recognizing the accumulating evidence showing the effectiveness of the sulfone drugs, thus retained a conservative position.
The arguments of the FSA were accepted, and as a consequence, those bills, though submitted repeatedly through 1954, were never enacted. The only leprosy-related legislation passed in this period was a bill enacted in 1951, which dealt with transportation costs for released patients\[[@B42],[@B43]\]. At the administrative level, however, a series of changes was introduced. In 1946, patients were given the right to vote. In 1947, the PHS removed HD from the list of quarantinable diseases which required a travel permit. In 1948, the barbed wire fence was removed from the facility, and the first active patient was released to the care of her private physician. In the same year, the PHS officially recognized the name Hansen\'s disease as a replacement for the term leprosy. In the 1950s, patients were allowed to marry. Twelve negative monthly tests were no longer required for discharge; instead those who qualified for discharge were required to be financially stable and assure authorities that there were no children living where they intended to go\[[@B44]\].
### US period II (1950 until 1980)
In the early 1950s, patient isolation was criticized more and more and eventually condemned at the International Congress on Leprosy and at other international conventions \[[@B45]\]. Following advice that the isolation of patients should be limited to infectious cases, the abolition of compulsory isolation was repeatedly recommended also by WHO\[[@B46]\] and UNICEF\[[@B47]\]. Definition of the cases suitable for temporary isolation gradually became delineated\[[@B48]\].
In 1952, a decision was made that no administrator should stay at the institution for more than three years. Accordingly in 1953, the government appointed a new director, Dr. Edward M. Gordon, a former director of the PHS hospital in Chicago. Soon after his arrival, he announced that those who did not need to be medically hospitalized should leave the institution\[[@B49]\]. His policy was that those able-bodied residents diagnosed as arrested should be discharged; that discharge should be recommended to those who were arrested and partially disabled; that those who were disabled (blind or physically handicapped) could stay in the institution, but could leave if their relatives or friends offered appropriate care. He also passed a rule prohibiting able-bodied residents from working in the institution. Fearful of losing their homes and jobs, some of the patients sent lawyers to Washington to lobby elected officials\[[@B50]\]. After three years, Dr. Gordon left for Fort Monroe in Virginia, and Dr. Edgar B. Johnwick was appointed as the new director. In 1956, on the third day of his duty, Dr. Johnwick declared before the patients that no one would be discharged from the hospital against his/her will and that no one would be kept in the hospital against his/her will\[[@B51]\].
During this period, scientists, while acknowledging the clinical effects of sulfones and the distinction between infectious types, were not unanimous in negating the utility of patient isolation. Dr. L. F. Badger of Leprosy Control Unit, the Communicable Disease Center (later, renamed as Center for Disease Control), stressed in 1956 the importance of isolation of infectious patients, as well as the need for patient followup and continued treatment of arrested cases\[[@B52]\]. Conversely, in 1958, Dr. Meyer, Medical Director at Carville, insisted that as disease control measure, isolation is not necessary in many cases and has serious defects\[[@B53]\]. No overt controversies were found after that in the US literature. Meanwhile in 1960, the World Health Organization advised against isolation and in 1963 at the 8th International Leprosy Congress in Rio de Janeiro, the International Leprosy Association recommended against isolation and furthermore recommended that leprosy needed no special care beyond that afforded other communicable diseases\[[@B54]\].
The last compulsory isolation was reportedly enforced in 1960, and Part 32 of the Federal Code of Regulations \"Medical Care for Persons with Hansen\'s Disease and Other Persons in Emergencies\" dropped the term \"detention\" in 1975\[[@B55]\]. It should be noted that all these policy changes were made through administrative action, not by changing statutory law. Law revision was not visible on the agenda as these policy changes proceeded. Even in 1974 when Hawaii ruled that all future new cases of the disease would be treated as outpatients, no debate took place on the federal level\[[@B56]\]. Compulsory isolation was terminated as the practice of involuntary admission and that of forced institutionalization ended.
### US period III (Post 1980 through 2000)
The driving force to terminate the isolation policy legislatively and abolish the leprosarium came not from new medical findings but from arguments for economic efficiency and budget cutting. In 1981, the Reagan administration and Congress agreed that it no longer made sense for the government to run public health hospitals, since Medicare, Medicaid, and the general oversupply of hospitals made specialized hospitals obsolete. A congressional subcommittee, backed by the Omnibus Budget and Reconciliation Act, thus targeted the nation\'s nine public health hospitals, including the National Leprosarium at Carville\[[@B57]\]. However, lobbied by the several groups which had developed over the years to oppose termination -- including the community of Carville, the patients, a society of Legionnaires and veterans -, Congressman Gillis Long fought for an exemption for Carville. Consequently, the leprosarium survived the challenge based on the argument that HD patients need special care and considerations (The institution was renamed as Gillis Long Hansen\'s Disease Center later in 1986)\[[@B58]\].
To the Congress, after 1981, Congressmen Henry A. Waxman and Gillis W. Long instead proposed a bill to change HD policy. Avoiding the issue of closing down the Carville institution, the bill was intended to permit treatment of HD outside PHS facilities and to eliminate additional pay for personnel treating HD. In the report attached to the bill\[[@B59]\], Richard Ashbaugh (Acting Director, Bureau of Medical Services, PHS) indicated the PHS\'s support for the bill. He stated, \"Leprosy pay \... was instituted because of a mistaken belief that workers were at significant risk in contracting Hansen\'s disease \... Not only is leprosy pay an unjustifiable expense, it actually increases the stigma associated with Hansen\'s disease.\" He went on to say, \"The PHS is strongly committed to the least restrictive possible treatment of Hansen\'s disease \... \" A similar act was submitted in the subsequent year, but also not enacted at least in part because beneficiaries of the center\'s payroll, including the patients working at the facility, made a strong objection to the bill.
Without legislative support, the PHS had been subsidizing ambulatory care centers for HD patients, outside the HD Center since 1981. The PHS National Outreach Program started 11 regional outpatient services nationwide\[[@B60]\]. When the PHS in 1983 did a formal review of the Center, it issued a report which highlighted the Center\'s economic inefficiency, making that the focal point of the discussion. The Center at that time housed only 200 patients but comprised 98 buildings on 337 acres. The staff included 317 civil service and PHS workers and 125 part-time patient employees. The PHS recommended that the Center continues its custodial care and research programs, but asked for more outpatient clinics and a study of the elimination of residential care. In the following year, the PHS set up a utilization review committee in Carville to review periodically the conditions of patients who had been hospitalized more than two years, for their possible discharge ability\[[@B61]\].
Finally, legislation was made. In 1985, PL99-117 was enacted which stated the PHS \... \"shall provide care and treatment (including outpatient care) without charge \... to any person suffering from Hansen\'s Disease who needs and requests care and treatment for that disease\"\[[@B62],[@B63]\]. The term \"detention\" was deleted from the Act, though the provision of residential care was not completely negated. The hazardous duty salary supplements, while had been reduced to one and one-quarter time, were discontinued for new hires.
In 1988, three years after the legislation, the PHS published its report \"Strategic Plan, National Hansen\'s Disease Center (NHDC)\[[@B64]\].\" The plan reiterated the situation of the NHDC, and evaluated whether it would be cost effective and feasible to contract out the patient care activities of the Center and transfer the research activities elsewhere. The report recommended expanding the Center\'s mission to include other nerve-desensitizing diseases, contracting out long-term patient care, and moving the Center\'s research facilities to Baton Rouge. It also argued the necessity of maintaining the Center so that patients who had lived at Carville for decades could be cared for compassionately in familiar surroundings. The report reached the definitive conclusion that there should be no new resident admitted to the Center, so that over time the population would dwindle to nothing and the entire facility could be closed. Accordingly in that same year, Dr. John Duffy, a new director, began the move of the Center\'s acute care, research and educational functions to Baton Rouge, while allowing the current residents to stay in Carville\[[@B65]\]. Thus, in practice, both voluntary and involuntary isolation and hospitalization ended.
Nonetheless, despite these developments, closure of the leprosarium required more than a decade, thwarting both a 1990 plan to transform the institution into a geriatric prison and a 1996 plan to make it a federal prison for minimum security geriatric patients. The efficiency argument made by the PHS was always counterposed by the equity and civil rights arguments offered by the patients and their allies. The patients, though demanding that their basic human rights and freedom be restored, claimed a right to lifelong care by the government\[[@B66]\]. Only the research branch was moved in 1992 to a location at Louisiana State University in Baton Rouge.
Then, agenda was set again by the economic efficiency argument, but this time with a new set of policy alternatives. Representative Richard H. Baker proposed that the site be used as a job training school, and that patients be given an annual living subsidy if they departed\[[@B67]\]. Baker, US District Judge Frank Polozola, and Jim Mitchell (PHS) visited Carville to discuss with patients and staff members the idea of closing the Center. This proposal was quashed once when Congressman Cleo Field, who represented the district from 1992 to 1996, objected. However, when the 1996 election returned the district of the Center (District 6 of Louisiana) to Baker, he resubmitted the bill with the proposal that the site be transferred to the State of Louisiana\[[@B68]\]. This time his plan was adopted.
PL 105-78, \"Relocation of Gillis W. Long Hansen\'s Disease Center\" was signed into law in 1997\[[@B69]\]. The legislation returned the physical facility to the State of Louisiana without charge, though it specified that the Carville site must be used for health or educational purposes for 30 years. It offered a voluntary separation incentive payment to civil service employees. It directed that at or through the Center, the Secretary of the Department of Health and Human Services (DHHS) must provide short-term care and treatment without charge, including outpatient care, for Hansen\'s disease and related complications. The Secretary, however, was not permitted to provide long-term care for any disease or complication through the Center. For long-term-care patients, the Secretary was instructed to provide for long-term care without charge for the remainder of the life of the patient. The bill also directed the relocation of the patients at the Center to Baton Rouge within three years unless such relocation was not feasible, and it reassured the remainder that they could stay at Carville as long as they were able to live independently. It further offered a \$33,000 annual stipend to any patient who chose to leave the institution, and this irrevocable option could be chosen at anytime by a patient. In August 1999, the federal government transferred the leprosarium to the State of Louisiana for use as a Job Corps training site. The Louisiana National Guard initiated its Youth Challenge program shortly after that\[[@B70]\]. Some patients left Carville going either to the Baton Rouge facility or to the other places, but quite a few remained. Today, the NHDC in Baton Rouge admits about 180 HD patients, and treats several hundred patients annually on an outpatient basis\[[@B71]\].
Isolation of leprosy patients in Japan
--------------------------------------
### Japan background
Since ancient times, historical records indicate, many HD patients lacked permanent homes, leaving them to wander around living in both towns and rural areas, and sometimes creating their own colonies. Leprosy had been understood to be a hereditary disease, but since the mid 19th century, the view that it was the result of a contagion became known gradually, first among the medical experts, and then among the public\[[@B72]\]. The recommendation of the first International Conference on Leprosy held in Berlin in 1897 led Japanese experts to support patient segregation, although not always unanimously. A national survey was conducted about leprosy, and Law No. 11, \"The Act on Leprosy Prevention,\" was passed in 1907. At that time, the Ministry of Internal Affairs (MIA) explained that despite the mild infectiousness of leprosy, Law No. 11 was required for the sake of vagrant lepers, those without means of support\[[@B73]\]. Five public leprosaria were established by local governments in 1909. Voluntary organizations and local governments launched the No Leprosy Movement in 1924, which tried to find all leprosy patients and send them to the leprosaria.
Subsequently, in 1931, the first national leprosarium was opened, and at that time, Law No. 11 was revised to the Leprosy Prevention Law, which allowed all patients, whether they could be cared for at home or not, to be hospitalized without any financial burden levied on their families\[[@B74]\]. The Ministry of Health and Welfare (MHW), the successor of the MIA, nationalized all of the existing leprosaria so as to coordinate their activities. Ministerial officials issued manifests that leprosy should be eradicated from Japan through absolute segregation\[[@B75]\]. In accordance with the increasing institutional capacity of leprosaria, the number of patients sent to leprosaria increased considerably.
### Japan period I (1940 through 1950s)
The sulfone drug Promin came to Japan in 1946 shortly after WWII ended. Its significantly beneficial effects were reported shortly at the Congress of the Japan Leprosy Association\[[@B76]\]. However, many physicians remained unconvinced of the real efficacy of the drug: They argued that the disease could relapse, even though it might show initial response\[[@B77],[@B78]\]. The new drugs were difficult to obtain either by import or by domestic production, but patients were desperate to try them. They organized the Federation of National Leprosarium Patients (FNLP) and launched lobbying activities to acquire the drug. In response, the Diet (the legislative body in Japan) approved in 1950 a budget for sulfones. The drugs were delivered to leprosaria, and, a year later, the first 35 patients were officially discharged as a result of their improvement\[[@B79]\].
Meanwhile after WWII, during the late 1940s and early 1950s, a new government was established in Japan, and legal systems were thoroughly reviewed and revised in compliance with the new Constitution. As part of that process, leprosy wound up on the agenda in the Diet in the early 1950s during discussions of social security systems. Debate went on until the revision of the existing law of 1931.
On February 15, 1950, leprosarium directors Drs. Kensuke Mitsuda, Yoshinobu Hayashi, and Ryoichi Yajima testified at the Health and Welfare Committee of the House of Representatives, along with the Director of the Medical Affairs Bureau, the MHW\[[@B80]\]. They insisted on the necessity of continued isolation of leprosy patients, argued for the desirability of the government subsidy for Promin, and recommended the expansion of leprosaria. On October 10, 1951, the Health and Welfare Committee of the House of Councilors heard testimony from five experts, including three leprosarium directors, Drs. Hayashi (then also President of the Japan Leprosy Association), Mitsuda and Miyazaki. Dr. Hayashi stressed the fact that there were approximately 9000 institutionalized patients, and 6000 outside the leprosaria who posed a threat of infection. He argued that leprosaria should be expanded to institutionalize all of the latter and that hazardous pay (about 5--10% at that time) should be increased as in the US. Dr. Mitsuda suggested that a more stringent law should be implemented to require isolation in order to prevent spread of the disease. He contended that though the necessity of isolation is sometimes refuted in the US, leprosy, regardless of its types, must be isolated to prevent infection and that sterilization should be recommended to patients for this purpose. He stressed that the new drugs, although promising in treatment, could not be expected to eradicate leprosy. Dr. Miyazaki essentially agreed with Dr. Mitsuda, adding that patients, many of them with deformities and disabilities, could not be released from leprosaria while they remained unaccepted by society and unable to live outside. He suggested that public enlightenment efforts would be advisable. The other experts argued for expansion of research into leprosy\[[@B81]\].
The bill to revise the Leprosy Prevention Law was submitted as a cabinet bill in 1953. The Minister of Health and Welfare explained to the House members that leprosy is hard to cure, that patient isolation is the only measure that prevents leprosy infection, and that the welfare of patients and their families should be protected by the government. Masayoshi Yamaguchi, Director of Public Health Bureau, the MHW, stated that patient isolation is the sole measure for leprosy prevention and protection of the public welfare, and that compulsory isolation should be warranted by law as a last resort. He suggested that patients might be discharged when leprosarium directors consider their isolation unnecessary and recommended that temporary leave also be allowed at the discretion of directors\[[@B82]\]. Facing the possibility of more stringent laws, the Patients\' Federation expressed concern that the prospective law under consideration was focused too narrowly on social protection, while ignoring patients\' rights\[[@B83]\].
The revised Leprosy Prevention Law of 1953 was in essence a reflection of the arguments of the experts and maintained a legal basis for compulsory isolation of patients proven to have bacilli and for prohibition of leave without permission. In response to patients\' pleas, nine supplementary resolutions were adopted, including provisions for living stipends and patient work, improvement in living conditions, promotion of research, and installation of rehabilitation facilities. The MHW did initiate programs for patient rehabilitation, but the Ministry was still devising a plan to expand leprosaria capacity to hospitalize leprosy patients. The Notice of the Vice-Minister, issued soon after the promulgation of the Law, still described isolation as the only reliable means of prevention of the spread of the disease. A tentative standard for discharge was officially detailed by the MHW in 1956, but it was stated as not being intended to facilitate discharge\[[@B84]\]. In practice, about 500 patients were admitted or readmitted to leprosaria in 1956, while fewer than 100 were discharged.
### Japan period II (1960s through 1980s)
In 1961, the legislature of the government of the Ryukyu Islands, then under the auspices of the US army, took a major step when it passed the Hansen\'s Disease Prevention Act. It provided that the Chief Executive may advise hospitalization and may also order an improved patient to leave the hospital\[[@B85]\]. Inspired by the policies of the Ryukyu Islands and acquainted with the international recommendation that most leprosy patients should be treated on an outpatient basis, thus abolishing compulsory isolation, the Patients\' Federation (FNLP) repeatedly voiced support for revision of the Leprosy Prevention Law of 1953\[[@B86]\]. They repeatedly insisted that the Law lacked a scientific basis and violated human rights, and argued that they were victims of enforced segregation and of social stigmas fostered by the Law\[[@B87]\]. In 1963, the Patients\' Federation filed its petition for the revision of the Law, sending more than 200 patients to pressure the Diet and the MHW.
Despite these efforts, the Diet did not add revision of the Leprosy Prevention Law to its agenda. The MHW officials argued that the existing law legitimized compulsory isolation of leprosy patients, which in turn constituted the legal basis for the government\'s responsibility to provide them with care and a comfortable living environment. Revision of the Law might eliminate the leprosaria, making it impossible for the patients to live on public support. Overt transition of leprosy control to an outpatient basis could also lead to a similar outcome. It was also argued that the transformation of leprosaria from medical facilities to rehabilitation facilities would be difficult because of budget constraints\[[@B88]\]. Concerned about the future of leprosaria, patients did not have the unanimity necessary for law revision. A 1965 survey disclosed that only 16% of patients had an intention to be discharged in the future while the rest felt that they were unable or unwilling to leave their leprosaria\[[@B89]\].
Leprosarium directors in Japan gradually relaxed the implementation of the law. Patients became increasingly free to leave their leprosaria, and abscondments were not punished after the 1960s. In addition, treatment began to be offered outside leprosaria, although on an informal outpatient basis\[[@B90]\]. Concerned that law revision could eventually harm patients\' welfare, patients, as well as the officials in charge, came to pay more attention to improving the leprosaria and providing other benefits for patients. The preservation of the Law was implicitly justified as a way to guarantee patients\' welfare.
In May 1972, the Ryukyu Islands, where leprosy treatment was based on outpatient services, were returned to Japan. This return again focused the attention of both patients and medical professionals on policy issues. In 1976, when the Federation of Leprosarium Directors drafted a revision proposal which incorporated specific discharge codes, and then submitted it to patients for their consideration, the Patients\' Federation objected to the proposal based on the concern that the adoption of specific discharge codes could result in forced discharge\[[@B91]\]. As was seen in the 1960s, many patients remained hesitant to address the issue of law revision, and their movements focused more on compensation for low-wage patient labor and redressing the long-term compulsory segregation of the past. A decade later, in 1984, the Patients\' Federation itself created a committee to examine revision of the Law, but continued to be concerned about the fate of their adopted homes, the leprosaria\[[@B92]\]. Consequently, the Federation of Leprosarium Directors only directed that leprosaria should not be abruptly discontinued as many of their residents were already too old to leave.
The MHW endeavored to improve living conditions in national leprosaria in response to repeated pleas filed by patients, as well as discussions in the Diet. Medical staffing was increased, patient labor was gradually reduced or made more rewarding, and government allowances were increased for leprosaria and their patients\[[@B93]\]. Officials did warn patients of the possibility that the numbers and sizes of leprosaria could be reduced as part of administrative reform if revisions to the Law were implemented too quickly.
### Japan period III (1980s through 2000)
When Fujio Otani became the Director General of the Tofu Kyokai Foundation, which had been established to serve patients with leprosy, the stalemate in policy change began to resolve. Otani was a medical officer in the MHW, and had been Chief of the Section of National Hospitals and Sanatoria in the 1970s, and Director General of Medical Affairs Bureau in the early 1980s. Recruited to the Foundation after his retirement, he also served as the Chairman of the Advisory Council on Public Health in the MHW. Since the recommendation in the mid 1980s of the UN Commission on Human Rights to improve psychiatric treatment in Japan, Otani had been inspired to act for the protection and restoration of human rights among patients\[[@B94]\]. In 1989, when the Patients\' Federation consulted with him on its draft petition on the law, he began to commit himself to the issue of law revision.
In 1992, encouraged by Otani, the Patients\' Federation filed a petition with the Minister of Health and Welfare. Although many patients still worried about the fate of their leprosaria, Otani promised them that their homes would be maintained\[[@B95]\]. In the following year, the Tofu Kyokai Foundation was officially consulted by the Ministry on the prevention policy for leprosy. A committee was established, composed of leprologists, medical experts, lawyers, media representatives, bureau officials and patients, and presided over by Otani. He tried to appeal to public opinion by establishing the Museum of Hansen\'s Disease (MHD) and hosting a series of symposia about the policies on HD\[[@B96]\]. In 1994 at the 67th Congress of the Japan Leprosy Association (JLA), Otani gave a special lecture and stated that the existing Leprosy Prevention Law should be abolished since it was not scientifically justifiable and violated patients\' human rights. Six months later, as a result, the Federation of Leprosarium Directors and the JLA publicly confirmed his opinion that the existing law should be abolished\[[@B97]\].
Upon the release of the report by the MHW committee in 1995, the MHW organized an internal panel of experts on the abolishment of the Leprosy Prevention Law. This panel, again chaired by Otani, subsequently submitted its report which recommended abolition of the Law, continued provision of public support for existing patients, and the use of the term \"Hansen\'s disease\" in place of \"leprosy\" in laws and regulations. Finally in 1996, the \"Act to Abolish the Leprosy Prevention Law,\" drafted by the MHW, was passed. It abolished the 1953 Law and at the same time codified the government\'s responsibility to provide existing HD patients with continuous medical and other social services. It was also provided that patients could either leave the leprosaria or stay there as long as they wished, and that those who decided to come back to the leprosaria after their discharge could be readmitted.
At the time of the abolition of the old law, there were 5,413 patients in leprosaria, whose average length of stay was more than 40 years and whose average age was 72 years old. Only six of these patients actually left their leprosaria in the following two years\[[@B98]\]. In response to lawsuits filed in the late 1990s, in 2001 the court awarded financial settlements to those who had been isolated under the earlier laws\[[@B99]\]. The awards amounted to between \$65,000 and \$114,000 per person depending on the degree of suffering. Additional pay to the workers at leprosaria was preserved.
Discussion
==========
This study demonstrates that the isolation of leprosy patients was introduced and made rigorous in both countries around the time when it became known that the disease was contagious. Patient isolation policies and leprosaria were maintained long after it became known that isolation is not necessary in the majority of cases. Remarkable was the stagnation of policy change in the post-war period. The preservation of the isolation policy provided patients with some social support, but continuously deprived them of their civil liberties. Furthermore, the policy as an authoritative statement on the disease may have fostered the social stigma associated with a belief that the disease is a dreadful contagion, thereby maintaining a hurdle to patients\' reintegration into society. Evidently, the policy\'s abolition was not easily accomplished nor was achieved solely by advances in scientific knowledge.
Garbage Can model of policy making
----------------------------------
Kingdon\'s Garbage Can model is very useful in understanding the failure of \'appropriate\' policy termination, as well as its delayed termination, observed in both countries: Any policy has an inertia and efforts to abolish the patient isolation policy did not bear fruit until three streams, i.e., problem, policy, and politics, converged after the mid-1980s. Science, although presumably a potent factor, was not in itself a major driving force in opening a policy window.
In the early post-WWII period (late 1940s through 1950s) both in the US and Japan, scientific developments such as learning that the infectiousness of leprosy is usually feeble and uneven across disease types and the development of sulfone drugs effective against the disease might have been expected to be the major factor in reconsideration of the traditional isolation and institutional policy for leprosy management. In reality, however, the laws were revised, not as a result of the scientific developments, but as a spillover from other policy agendas: the enactment of the Public Health Services Act in the US, and the revision of social security systems in Japan, without substantial policy changes. Once legislated, they could hardly be challenged.
Most actors then participating in the policy discussion were well aware of the new drugs, and patients were anxious to reap the fruits of these drugs and achieve the restoration of their civil liberties, sometimes participating in political mobilization for law revision. Scientific and medical advances could have altered the face of issues in the problem stream, helped generate alternative methods of disease management in the policy stream, and changed the opinion of the public and experts in the political stream. However, many influential experts, who had been engaged in the establishment and/or expansion of leprosaria were determined to keep the policy and the institutions and played on remaining public fear of leprosy\[[@B100]\]. Those conservative \"elites\", primarily bureaucratic agency representatives in the US and medical professionals in Japan, manifested opposition to the proposed law revision, and their arguments were not critically reviewed and more or less accepted by the other actors. Consequently, science and medicine were unable to play a major role in changing the political environment. For these reasons, a valuable opportunity for legislative change was lost in this period when patients were still young and had not become so dependent on their sanatoria in both countries.
After the unsuccessful exploitation of these incomplete policy windows, the three streams of problems, policies and politics did not converge for several decades. By the late 1960s, it could be said that a near consensus had developed in the research community that isolation was an unreasonable solution. Moreover, the infringement of civil liberties was perceived to be a problem, a perception which increased in importance as academics and international health communities repeatedly recommended the abolition of isolation policy as scientifically unjustifiable.
In the problem stream, focusing events were created by patients, administrators and medical professionals with the goal of encouraging the public and politicians to put the issue on the agenda. Most obviously, patients published periodicals, invited media persons to their meetings, petitioned bureaucracies, and lobbied legislators in both countries. In Japan, patients even conducted sit-ins around the Diet and the Ministerial buildings. The repeated recommendations of international organizations, and the occasional official reports recommending outsourcing and outpatient services, also served as focuses. In the political stream, there were some changes as well. As the public\'s fear of leprosy dwindled (or at least became latent as they forget about the disease), so did the associated stigma. Improved public attitudes toward the disease meant that the political climate was more favorable to returning the patients to society, terminating the isolation policy, and abolishing leprosaria. In both countries, there were periodic turnovers of leprosaria directors, sometimes recruited from outside the leprosaria. The appointment of leprosaria directors from other public hospitals and academic institutions, who had the intention to discharge the long-term residents, certainly could have opened a window.
Every time the issue reached the agenda, however, the proposed policy alternatives raised concerns among patients. Treating leprosy as an ordinary disease and integrating treatment into general medical care and public health measures threatened the \'special\' status of leprosy. Many patients, who had become dependent on the leprosaria, feared losing their homes and privileges. If from a scientific perspective the disease no longer required isolation, patients sought to justify their continued privileges either as government duty to provide disease victims with compassionate care or as government compensation for past erroneous policies. Such patients\' claims were not easily accepted. Given the lack of a public consensus on the moral underpinnings for patient support, the belief that long-term residents are entitled to generous government \"compensation\" was a hurdle. To surmount that hurdle, administrators and medical professionals would have had to acknowledge their wrongdoing and determine when the isolation policy could and should have been judged as obsolete. Additionally, public opinion, and of course the views of elected officials, would have needed to be nurtured to recognize and sympathize with patients\' adversaries, allowing for generous government payments. As a result, the position of the patients was not really unified, and therefore they could not fully mobilize themselves politically. Consequently, this policy community more or less shared the view that the patients\' disadvantages caused by the law were a necessary cost of its even greater benefits, namely the continuation of leprosaria\[[@B101]\].
When it proved impossible to accomplish decisive legislative changes, straightforward efforts to discharge long-term and medically unwarranted residents were also reduced. Instead policy adjustments were made by incremental administrative actions. Slow, limited and sometimes informal efforts were made for patients\' discharge and for the provision of outpatient services. As one might expect\[[@B102],[@B103]\], these disjointed and informal increments of change had limited success in modifying the policy\'s original effect. Since the policy under the existing law was administered on the premise that leprosy requires confinement and special care, it was limited in its capacity to compel patients to leave the leprosaria. As many patients chose to remain in their leprosaria, isolation ostensibly continued in the form of long-term residential care. In this way, the main focus of leprosy prevention policy gradually changed from social protection and patient care to the mere provision of residential places for patients and ex-patients (and later, rehabilitative services). These bureaucratic satisficing behaviors, as reported in other cases\[[@B104],[@B105]\], could not raise again the fundamental question of law revision, and as a consequence many engaged themselves in pork barrel politics to preserve and improve life in the leprosaria. For many years, policy (change) focused on discrete, short-term outputs rather than broader, long-range outcomes\[[@B106]\].
In the 1980s, broad policy revision became possible again though the agendas were set by factors again other than science, specifically the discussion on economic inefficiency in the US and the issue of human rights violation in Japan. In the former, large government expenditure was a big social and political issue, and there was an active search for potentials for spending cuts. In the latter, AIDS as a source of social stigmas and the forced institutionalization of psychiatric patients were both prominent in the media, calling for the protection of the human rights of the diseased. Furthermore, as Bardach suggested\[[@B107]\], increased political competition and Cabinet turnover at that time may also have increased the likelihood that elected officials would take up potential social issues to be remedied. In the US, Representatives Baker and Field fought for a House seat, while in Japan the two major parties competed vigorously for government office.
An indispensable key to policy change was the development of policy alternatives. Interestingly, they were quite common in both countries. Evidently, policy change occurred only after the leprosaria were recognized as legitimate homes for some patients and outpatient services were officially established. Patients were assured of their residences, but had the option of leaving the institutions. The patient\'s privilege of remaining in a leprosarium was then justified as partial compensation for the past compulsory segregation. In both countries, hazard duty payments were also maintained for ongoing employees, who could have otherwise opposed to policy changes. The primary difference, as reflective of the key arguments against the old policy, i.e., economic inefficiency and human rights violation, was that in Japan a public apology was offered by the government for its long-term negligence, though the apology was later used by the patients to win further government compensation through lawsuits. Another point is that in the US, large stipends were awarded to patients as an economic incentive to reduce the number of institutional residents.
The functions of science and scientists in the policy process
-------------------------------------------------------------
From the perspective of the Garbage Can model, problems, policies and politics typically evolve in separate, unconnected streams. When science is considered as a fourth factor, it is seen to influence all the streams, changing perception of problems, helping develop policies, and transforming political environments\[[@B108]\]. Thus, science can be seen as approving or disapproving certain (existing) policies. However, it does not automatically create those effects by itself, nor in a political vacuum. Even when science indicates that an existing policy is no longer relevant, as this case study shows, people may make objections to its abolition for a variety of other reasons. In comparison with basic research, furthermore, applied research, including public health and epidemiological research, is more likely to follow an agenda driven by forces other than science\[[@B109]\]. This means that the imminence of an issue, or its agenda status, can affect the production and use of scientific knowledge.
Furthermore, scientific disputes will not be always resolved during the time the scientific issues are considered relevant to policy discussions\[[@B110]\]. Looking back at Period I, Promin was first mentioned in Leprosy Review in 1945, but it was not until 1952 that sulfone therapy became accepted among academics\[[@B111],[@B112]\]. Controversy on the effectiveness of sulfone drugs certainly still lingered at the time when the legislative changes were on the agenda, namely, around the time of the enactment of the Public Health Services Act and subsequent policy proposals in the US and at the time of the revision of Leprosy Prevention Law in Japan. The development and acceptance of scientific knowledge might have been slowed by the attitudes of authoritative and powerful experts in Japan. Also, in the US, quite a few experts continued to justify the isolation of infectious cases for many years. Arguments that did not support their convictions were sometimes screened out by disqualifying them. Authoritative positions of medical and administrative elites in society also hindered the public\'s critical appraisal of their arguments, as did the importance then attached to public health\[[@B113]\]. The strength of the conservatives\' arguments possibly came also from the systemic bias in their position as defenders of status quo: they had to simply attack a proposer\'s case as insufficient\[[@B114]\].
In an adversarial process of rule-making, knowledge claims are sometimes deconstructed, exposing areas of weakness or uncertainty. These revealed weaknesses provide justification for political decision makers to assert that they have a right to engage in interpreting science, especially in areas that are controversial. Again, uncertainty within science itself is also a subject for negotiation, decision, and argument in policy process, since the quality and amount of knowledge are the objects of social negotiation\[[@B115]\]. This partial transfer of cognitive authority to the legal and political arena may be seen as the way of assuring that the interpretation of (indeterminate) facts reflects the public values embodied in the legislation as well as the norms of the scientific community\[[@B116]\]. The connections between given scientific data, expert interpretation of these data and policy content are like chains of linked arguments and beliefs\[[@B117]\]. This process might be termed co-evolution or the process of mutual validation between policy and science\[[@B118]\]. In a case where science and policy are co-constructed through processes which occur in tandem, it becomes difficult to explain the one by using the other\[[@B119]\]. From the vantage point of history, the very absence or scarcity of critical review among legislators and the public in reviewing scientists\' statements on the risk of disease spread might reflect and indicate their own fear of the disease and indifference to patients\' human rights. When scientific knowledge is used in policy development, scientists and/or policymakers may choose to err either on the side of public safety or on that of patients\' liberty and dignity.
For a long portion of Period II, incomplete knowledge was not necessarily the primary reason why science did not play a major role. Scientists, experts and legislators could have acted strategically as policy entrepreneurs to take advantage of occasional opportunities, provided by several focusing events, to open the windows of opportunity. In other words, scientific opposition, or dissensus if any, was not a major hindrance to their actions. Scientists, however, preferred not to encroach upon policy debates in an official and overt fashion, thus avoiding direct questions about justification and continuation of the institutions. In both countries during the 1960s and 1970s, there was a near consensus, albeit implicit among the interested parties, including scientists, physicians, bureaucrats, patients and politicians, that the leprosaria should be preserved or at least not immediately abolished. Though scientific assessment appeared conclusive and potent enough if effectively presented, consideration for the possible negative consequences of policy outcomes constrained its use. In the resultant incremental decision making, the demand for science was minimal.
Later in period III, when development of policy alternatives was definitely the key to changing policies, the process was not fueled by (new) scientific knowledge or discussion, but rather propelled by political skills which crafted both the policies and the favorable environments. A scientific consensus had already existed before the issue was taken up at this point, which illustrates the secondary importance of science in the emergence of policy windows. At most, the known arguments were reiterated and pushed to the fore. Policy change was not really contested and disputed in terms of science, simply because it did not have to be once attractive policy alternatives were proposed.
From this vantage in all the periods, science remained instrumental in the sense that scientific rationality was not the primary objective for which policy change was discussed or intended, and in the sense that scientific arguments were not the principal driving force for policy change. To review the course of events: Some initiating event, not directly associated science, leads to a policy issue that must be decided on; a debate ensues on the possible policy options and on possible scientific views of the issue; a scientific assessment of the policy issue yields a rationale for a policy choice\[[@B120],[@B121]\]. Thus, science is used either to legitimate policies developed for nonscientific reasons or is ignored if the consensus contradicts policy or there is scientific \"dissensus.\" To the extent which science can be regarded as instrumental, namely that final policy decision was rendered to lawmakers, the role and responsibility of science could decrease\[[@B122]\]. The scientific base and the political will to translate it to policy are ingredients necessary to induce policy changes\[[@B123]-[@B125]\]. Inertia and/or political considerations certainly exerted huge influence\[[@B126]\].
Policy entrepreneurs and leadership
-----------------------------------
Although many contextual conditions could facilitate changes in health policies, they do not in themselves activate actions by either legislators or administrators\[[@B127]-[@B129]\]. As is the case with policy adoption\[[@B130],[@B131]\], successful termination might need a political leader, a terminator, to trigger and manage effectively the process of policy termination. He must cultivate support from both influential interest groups and from the general public and choose the arena for policy discussions\[[@B132]-[@B134]\]. Then, a person skilled in policy termination should get the termination on the agenda, orchestrate advocacy coalitions, negate the survival tactics of anti-termination coalitions, and manage the administrative details\[[@B135]-[@B138]\]. With respect to HD policy, for a long period, intellectual reluctance, the tendency to institutional permanence, dynamic conservatism, efficacious anti-termination coalitions, and legal obstacles certainly posed difficulties for policy changes in both countries\[[@B139],[@B140]\]. After a set of these contextual conditions were met, however, entrepreneurs had to accomplish their tasks. A crucial step for policy termination, namely the translation of ideas into action by coupling concrete and acceptable policy proposals with problem situations and political opportunities, was accomplished by a policy termination expert\[[@B141],[@B142]\].
When things proceed as the Garbage Can model suggests, policy change requires a good luck and/or a skillful policy entrepreneur, who can induce the convergence of streams and open the windows of opportunity. This study highlighted an essential role assumed by a skillful terminator, Baker in the US and Otani in Japan. In the US, Representative Baker, along with the PHS officials, held meetings with the patients and staffs at the institution so that all interested parties could reach agreement on his proposals. The PHS publicized its own review of the institution and its activities highlighting their economic inefficiency. In this way, Baker and the PHS maneuvered the political process to successful legislation enactment. In Japan, Otani dramatized the evils of the old policy, stressing the benefits of termination, while mollifying the worried patients by reassuring of their continued residence in the sanatoria. He thus successfully engineered a consensus among key interest groups, publicizing a thorough justification for termination and devising an acceptable plan. Through political leadership and skill, the issue was finally brought to the forefront, alternatives supplied, obstacles cleared, and the policy abolished. Both Baker and Otani conducted a series of bandwagon activities making effective use of the media.
Conclusion
==========
Quarantine measures aim to protect the health and safety of the public, and consequently individual liberty is subordinate to the common good\[[@B143]\]. Parens patriae, the obligation of the state to act as parent of the country in caring for those who cannot care for themselves, is the other side of the duty of the state to protect the rest of the community from infected individuals\[[@B144]\]. The human rights or civil liberties of patients must be weighed the effectiveness and efficacy of the isolation measure. It is inarguable that science could and should play a key part.
The utilization of scientific knowledge contributes to increased rationality of health policy, which is desirable especially when incongruence could produce adverse effects. However, when policy changes occur as a result of the convergence of three largely independent streams of problems, policies and politics, we cannot assume a rational and linear process which automatically incorporates science into policy. As discussed, a variety of dynamics influence the utilization of scientific research in policymaking\[[@B145]-[@B147]\]. For policies to be reflective of science, these models suggest, the interactions among researchers and policy makers are critical. Research is more likely to be utilized in a significant way when effective networks and mechanisms are established at the interfaces among researchers and policy makers\[[@B148]\]. However, even that may not be sufficient.
Events in the various streams should be made readily understandable and visible, either in hopes of a serendipitous convergence or so that policy entrepreneurs can assist in their convergence, in order to open the window of opportunity. Problems should be foregrounded, political factors should be elucidated, and feasible and attractive policy alternatives should be envisioned. Experts should be mobilized to assume authoritative roles in linking science and policy, though it should always be noted that they too could be biased in one way or another. Political empowerment of the people who suffer from the negative policy outcomes, and therefore might be most attentive to them, could also help serve the purpose.
Policy is always developed both from scientific and nonscientific considerations, as this study indicated. The Possibilities for the instrumental use and/or disuse of science should be made explicit to the public. Mobilization of different social views and values, in comprehensive consideration of direct and indirect policy impacts, could lead to a more explicit and formal social process, which could eventually secure and propel the process of muddling through\[[@B149]\]. Public governance over these processes, along with public\'s enlightenment and mobilization, might facilitate, check and complement the functions fulfilled by the experts. Analysis of policy and politics is the indispensable key to these processes.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
All the authors (HS, JF) fully participated in the planning of research, the collection of historical materials, and the carrying out of interviews in both the US and Japan. After thorough discussion, HS drafted the article, and both authors read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-698X/5/3/prepub>
Acknowledgements
================
This study was partially supported by the health science research grant from the Japan Ministry of Health, Labor and Welfare (2000--2001) for HS, and the grant from the US Social Science Research Council (2002--2003) for JF. Cooperation provided by the officials and ex-officials in both countries are greatly appreciated.
|
PubMed Central
|
2024-06-05T03:55:55.296981
|
2005-3-16
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079837/",
"journal": "BMC Int Health Hum Rights. 2005 Mar 16; 5:3",
"authors": [
{
"first": "Hajime",
"last": "Sato"
},
{
"first": "Janet E",
"last": "Frantz"
}
]
}
|
PMC1079838
|
Background
==========
Rheumatoid arthritis (RA) is an autoimmune disease characterized by chronic inflammation of the synovial membrane that leads to the destruction of cartilage and bone tissues. The major cellular components of the inflammatory infiltrate in RA synovium are T cells, macrophages, B cells, and dendritic cells (DCs) \[[@B1]\]. Tissue-infiltrating lymphocytes are often arranged in follicles which exhibit variable degrees of germinal center (GC) reactions \[[@B2]-[@B4]\]. These highly organized structures are similar to secondary lymphoid organs, and the process is referred to as lymphoid neogenesis \[[@B5],[@B6]\]. In RA, this ectopic formation of lymphoid structures correlates with the expression of lymphotoxin-β (LT-β) and homing chemokines such as BLC (CXCL13) and SLC (CCL21), recapitulating the developmental process of lymphoid organogenesis \[[@B7]\].
The recruitment of cellular elements to secondary lymphoid organs relies on interactions between circulating leukocytes and the specialized endothelium of high endothelial venules (HEV), which are exclusive to these organs under physiological conditions. The interaction of L-selectin on lymphocytes with sulfated glycoprotein ligands on HEV of lymph nodes results in lymphocyte rolling on endothelium which represents the initial step in lymphocyte homing \[[@B8]\]. HEV-ligands for L-selectin consist of a set of heavily O-glycosylated glycoproteins, which include GlyCAM-1 and CD34 in the mouse and podocalyxin and CD34 in the human \[[@B9]\]. Recognition of these ligands by L-selectin requires sialylation, fucosylation and sulfation of their mucin-like domains \[[@B10]\]. The minimal recognition epitope appears to be comprised of a capping group known as 6-sulfo sialyl Lewis x (6-sulfo sLex) in which the C-6 position of GlcNAc is esterified with sulfate. The MECA-79 mAb recognises a critical sulfation-dependent determinant on these ligands that overlaps with the L-selectin recognition epitope. Expression of GlcNAc6ST-2 (commonly referred to as HEC-GlcNAc6ST or LSST), an HEV-localized sulfotransferase, is essential for the elaboration of functional ligands within lymph nodes, as well as the generation of the MECA-79 epitope \[[@B11]-[@B13]\]. Another sulfotransferase known as GlcNAc6ST-1 also contributes to this epitope in HEV \[[@B14]\]. MECA-79 recognizes a set of sialomucins, including the aforementioned L-selectin ligands, together with another recently identified sialomucin called endomucin \[[@B15],[@B16]\]. The complex is referred to as peripheral node addressin (PNAd) \[[@B13]\]. MECA-79 blocks the attachment of lymphocytes to peripheral lymph node (PN) HEV in vitro and inhibits lymphocyte homing to PN in vivo \[[@B17]\].
The role of GlcNAc6ST-2 in the generation of L-selectin ligands in lymphoid organs has been clearly established in mice genetically deficient in this enzyme \[[@B18]-[@B20]\]. Lymphocyte homing to PN is reduced in the *GlcNAc6ST-2*-/- mice as reflected by a smaller size of the PN and a 60% decrease in the number of total lymphocytes within this organ. In addition to lymphoid organs, induction of the MECA-79 epitope has been identified in several chronically inflamed tissues where it is proposed to participate in leukocyte recruitment \[[@B21]\]. Inflamed synovium contains high-walled vessels, resembling HEV, which express the MECA-79 epitope and are capable of binding L-selectin+ lymphocytes in *ex vivo*assays \[[@B21]-[@B25]\].
Recently, the use of GlcNAc6ST-2 specific antibodies has demonstrated de novo induction of this enzyme in HEV-like vessels and its correlation with the presence of the MECA-79 epitope in several mouse models of chronic inflammation \[[@B26]\]. In this study using an antibody to human GlcNAc6ST-2, we analyse whether this enzyme is expressed in RA and in cultured endothelial cells and its correlation with the expression of MECA-79 epitope.
Results
=======
Expression of MECA-79 epitope and GlcNAc6ST-2 in synovial tissues
-----------------------------------------------------------------
We produced an anti-human GlcNAc6ST-2 antibody and tested it on human lymph node and tonsil sections. A vascular pattern of labelling was observed which colocalized to MECA-79 positive vessels of HEV morphology (Figure [1a](#F1){ref-type="fig"}). A few mononuclear cells were also labelled with the anti GlcNAc6ST-2 antibody. Expression of the MECA-79 epitope was detected in 7 of 14 RA and none of the 9 OA synovial tissues. Most MECA-79^+^vessels showed a HEV morphology but in some cases, small vessels with a flat endothelium were also labelled. Five of the 7 MECA-79^+^tissues contained abundant lymphocytic infiltrates, which in 4 cases displayed a follicular pattern, and in 2 cases, MECA-79^+^vessels were observed in the absence of significant infiltration (Table [1](#T1){ref-type="table"}). Three MECA-79 negative RA tissues contained diffuse lymphocytic infiltrates and in one case, a single follicular infiltrate.
Expression of GlcNAc6ST-2 was not detected in any of the OA synovial tissues. In contrast, 10 of 14 RA tissues showed GlcNAc6ST-2 by immunoperoxidase staining (Figure [2](#F2){ref-type="fig"}). Staining was located to endothelial cells of HEV vessels in the sublining and in some samples, also in small vessels with flat endothelium. In some sections, a few perivascular mononuclear cells were also stained. Double MECA-79 and GlcNAc6ST-2 immunofluorescence showed that all MECA-79 positive vessels also displayed GlcNAc6ST-2 (Figure [1b](#F1){ref-type="fig"}). However, in a low proportion of GlcNAc6ST-2^+^vessels, MECA-79 was not detected.
Specificity of the immunoperoxidase GlcNAc6ST-2 labelling was confirmed by substituting normal rabbit IgG for the specific antibody. There was no detectable staining (data not shown). Preincubation of the anti-GlcNAc6ST-2 antibody with the peptide immunogen abrogated the staining (Figure [2b](#F2){ref-type="fig"}). We also immunostained with a polyclonal antibody raised against murine GlcNAc6ST-2 \[[@B26]\] and observed a similar pattern of staining in small vessels (Figure [2c](#F2){ref-type="fig"}). This antibody reacted with smooth muscle cells in the media of larger vessels which we attribute to a spurious cross-reactivity (data not shown). Immunostaining with this antibody was also abrogated by preincubation with murine GlcNAc6ST-2 peptide (Figure [2d](#F2){ref-type="fig"}).
Expression of GlcNAc6ST-2 by cultured HUVEC
-------------------------------------------
To analyze the potential role of LT-αβ and TNF-α, two cytokines abundantly expressed in RA synovium and associated with lymphoid neogenesis and GlcNAc6ST-2 expression in transgenic models \[[@B3]-[@B7],[@B26],[@B28]\], we examined the effects of the exposure to these cytokines on GlcNAc6ST-2 expression in human endothelial cells lacking the HEC phenotype (HUVEC).
By RT-PCR, treatment with LT-αβ or TNF-α for 8 h induced a significant increase in the expression of *GlcNAc6ST-2*mRNA in HUVEC (Figure [3a](#F3){ref-type="fig"}). A synergistic effect of treatment with both cytokines was not detected (data not shown). Treatment with an unrelated cytokine (interferon-γ) did not significantly modify *GlcNAc6ST-2*mRNA expression levels.
By western blot analysis with the anti-human GlcNAc6ST-2 antibody, we detected a single 50 kD band in HUVEC protein extracts, consistent with the expected MW for human GlcNAc6ST-2 \[[@B11],[@B12]\]. GlcNAc6ST-2 protein expression was significantly increased after 24 h of TNF-α or LT-αβ treatment in HUVEC (Figure [3b](#F3){ref-type="fig"}). Immunofluorescent labelling of HUVEC cultures with anti-human GlcNAc6ST-2 antibody showed a weak, but clearly detectable signal above the non-immune rabbit IgG control. Pretreatment with either TNF-α or LT-αβ induced a clear increase of immunofluorescence with a perinuclear distribution which is characteristic of the Golgi apparatus (Figure [3c](#F3){ref-type="fig"}). Neither TNF-α nor LT-αβ treatment induced expression of the MECA-79 epitope in HUVEC (data not shown).
Discussion
==========
The presence of MECA-79^+^vessels has been previously reported in human inflammatory lesions, often associated with the development of lymphoid neogenesis \[[@B21],[@B26]\]. In a variety of unrelated inflammatory conditions (rejecting heart and kidney allografts, peribronchial specimens from asthmatics, thyroiditis, psoriasis, and *H. pylori*infection) induction of MECA-79^+^vessels in association with inflammatory infiltrates has been demonstrated \[[@B29],[@B32]\]. In some cases, MECA-79^+^vessels also display HEV-like morphology although flat endothelium of inflammatory vessels can also express the MECA-79 epitope. In chronic arthritis synovial tissues, the presence of HEV and MECA-79^+^vessels has been previously demonstrated \[[@B22],[@B23]\]. Recently, the pathophysiological significance of MECA-79^+^vessels has been directly demonstrated in a sheep model of asthma. Pretreatment of asthmatic sheep with intravenously administered MECA-79 blunted both late-phase airway responses and airway hyperresponsiveness induced by airway allergen challenge \[[@B33]\].
In murine models of lymphoid neogenesis, including spontaneous models (AKR mouse and NOD mouse), and transgenic models (RIP-BLC and RIP-LT-αβ), expression of the MECA-79 epitope is closely associated with the induction of GlcNAc6ST-2 \[[@B12],[@B26],[@B28]\]. Taking advantage of a newly-generated antibody to human GlcNAc6ST-2, our present results are the first to show this association in a chronic inflammatory disease in humans. Several of our observations suggest that the induction of GlcNAc6ST-2 represents an early event in the process of lymphoid neogenesis within RA synovium. In the first place, GlcNAc6ST-2 was expressed in flat and high endothelial vessels which exceeded the number of MECA-79^+^vessels in double-labelled sections. Secondly, GlcNAc6ST-2 and MECA-79 reactivity were detected in some vessels which were not associated with lymphoid aggregates, and in a few cases, even in the absence of significant mononuclear cell infiltration. The latter observation also suggests that MECA-79^+^vessels may contribute to lymphoid recruitment, but additional factors are necessary. This is consistent with the multistep model of leukocyte transendothelial migration to lymphoid organs, where posting of chemokines on endothelium and integrin-mediated adhesion of rolling leukocytes are required \[[@B8]\]. Lack of expression of MECA-79 epitope and GlcNAc6ST-2 in OA samples suggests that the scanty infiltration of OA tissues is not sufficient to induce these features and that RA-associated factors are needed.
The development of lymphoid neogenesis in RA synovium has been suggested to facilitate local autoimmune responses permitting the development of B cells and local production of autoantibodies and immune complex formation \[[@B34]-[@B36]\]. The recent development of B-cell depleting therapy demonstrates an important role for these cells in RA \[[@B37]\]\]. The recruitment of L-selectin^+^dendritic cells, B-cells and CD45RA T-cells to RA synovium correlates with the expression of homing chemokines such as CXCL13, CCL19 and CCL21, and the latter partially colocalizes to MECA-79 positive HEV \[[@B3],[@B38],[@B39]\]. We have also detected expression of CXCL12, a homing chemokine abundantly present in RA tissues in HEV-like vessels \[[@B40]\].
Local expression of LT-β and TNF-α are important factors in the pathogenesis of RA, and LT-β expression correlates with the expression of homing chemokines and lymphoid neogenesis \[[@B1],[@B3]\]. The role of LT-αβ and homing chemokines in the induction of HEV has been suggested by murine studies in RIP-LT-αβ and RIP-BLC (CXCL13) mice which spontaneously develop lymphoid aggregates and GlcNAc6ST-2 expression in MECA-79^+^vessels \[[@B26],[@B28],[@B29]\]. The role of this enzyme in lymphoid recruitment has also been studied in RIP-BLC/ *GlcNAc6ST-2*-/- mice which exhibit a reduction in lymphoid infiltration \[[@B26]\].
We demonstrate that de novo induction of GlcNAc6ST-2 expression also occurs upon LT-αβ or TNF-α stimulation in cultured HUVEC, further supporting the role of LT-αβ and TNF-α as major inducers of extranodal lymphogenesis. Enforced LT-αβ expression in murine models of lymphoid neogenesis induces the expression of homing chemokines, making it difficult to dissect the role of LT-αβ and chemokines in GlcNAc6ST-2 expression and HEV-development \[[@B41]\]. A role for LT-αβ, which is downstream of chemokines, has been observed in a transgenic model of BLC expression by genetic or functional targeting of LT-αβ \[[@B42]\]. In murine models of extranodal lymphoid neogenesis, a non redundant role for LT-αβ signaling through the alternative NF-κB pathway has been demonstrated, whereas the role for TNF-α seems dispensable and ectopic LT-α does not induce GlcNAc6ST-2 expression (13, 28). Our data suggest that in human cultured EC TNF-α may contribute to this process by inducing GlcNAc6ST-2 although its relevance *in vivo*remains to be determined. Therefore, our data suggest that both LT-αβ and TNF-α can directly induce GlcNAc6ST-2 in cultured endothelial cells, although additional factors seem to be required for the full development of the HEC phenotype *in vitro*. One such factor may be the core 1 extension enzyme, known as Core1-β 3GlcNAcT, which is required for the formation of the MECA-79 epitope on O-glycans \[[@B29],[@B43]\]. Our results further establish carbohydrate modifying enzymes as an important group of transcriptional targets for cytokines during inflammation. Increased carbohydrate sulfation of several glycoproteins, including heparan sulfate proteoglycans, CD44 and airway mucins, has previously been demonstrated in response to TNF-α \[[@B44]-[@B46]\].
In RA, TNF-α plays a major and multifunctional role in the pathogenesis of the disease whereas the role of LT-αβ is currently being explored \[[@B47]\]. The potential contribution of these factors to the development of HEV expands the spectrum of possible pathogenetic roles for these cytokines.
Conclusion
==========
MECA-79 defined glycoproteins serve as ligands for L-selectin during the process of lymphocyte homing to lymph nodes. This class of ligands is also critical for inflammatory leukocyte trafficking and the associated disease, as recently demonstrated in a sheep model of asthma. Previous experiments in the mouse have shown the requirement for GlcNAc6ST-2 in generating the MECA-79 epitope and the associated L-selectin ligand activity in HEV and HEV-like vessels. Here, we demonstrate that MECA-79^+^vessels within human RA synovium also express GlcNAc6ST-2, strongly implying that this enzyme contributes to the ligands on these vessels. Establishing the direct participation of GlcNAc6ST-2 in the formation of L-selectin ligands on RA HEV would identify this enzyme as a potential therapeutic target for inhibiting leukocyte recruitment to the joints. In the same context, our finding that TNF-α induces GlcNAc6ST-2 in cultured endothelial cells may help to rationalize the efficacy of anti TNF-α therapy in the treatment of RA and other inflammatory diseases. Moreover, the activity of LT-αβ in inducing GlcNAc6ST-2 in HUVEC is likely related to the established requirement for this cytokine in the formation of HEV during lymphoid organ development. Many opportunities for future investigation are clearly indicated.
Methods
=======
Preparation of anti-human GlcNAc6ST-2 antibody
----------------------------------------------
A polyclonal antibody against human GlcNAc6ST-2 was prepared following the same basic procedures used for mouse GlcNAc6ST-2 \[[@B26]\]. The peptide chosen, RGKGMGDHAFHTNC, differed from that of the corresponding mouse sequence by one amino acid (D replacing Q) with the C-terminal C added to facilitate conjugation. The peptide was coupled to KLH and injected into New Zealand White rabbits by Covance Research Products (Denver, PA). 500 μg of conjugated peptide was mixed with 0.5 ml complete Freund\'s adjuvant and injected intradermally, followed by three boosts, every three weeks, with subcutaneous injections of 250 μg conjugated peptide in 0.5 ml incomplete Freund\'s adjuvant. The subsequent antiserum was subjected to purification on the immunogen peptide coupled to agarose using the SulfoLink Kit (Pierce Biotechnology, Rockford, IL), following the manufacturer\'s recommendations. The affinity purified antibody was washed into Dulbecco\'s PBS on a Centricon 30 (Millipore, Billerica, MA) and concentrated to 0.125 mg/ml.
GlcNAc6ST-2 and MECA-79 immunolabeling
--------------------------------------
Synovial tissues were obtained from 14 RA and 9 control (OA) patients at the time of knee prosthetic replacement surgery. All patients gave informed consent, and the study was approved by the ethics committee of the Hospital 12 de Octubre.
Tissues were snap frozen in OCT and stored at -80°C. Human tonsil and lymph nodes frozen sections were used as controls for MECA-79 and GlcNAc6ST-2 immunolabeling.
Frozen sections (8 μm) were fixed for 10 minutes in cold acetone. Double GlcNAc6ST-2 and MECA-79 immunolabeling was performed by blocking for 30 minutes in 5% normal goat serum PBS, followed by incubation of sections with 1.4 μg/ml MECA-79 mAb (rat IgM) and 0.3 μg/ml rabbit anti-human GlcNAc6ST-2 at room temperature for 1 hour. After washing in PBS, a secondary biotinylated goat anti-rabbit IgG at 1.3 μg/ml was added (Jackson Immunoresearch Laboratories, West Grove, PA). Fluorescence was developed with 1.5 μg/ml Cy3-conjugated goat anti-rat IgM and 1.8 μg/ml Cy2-conjugated streptavidin (Jackson Immunoresearch Laboratories).
Immunoperoxidase staining was performed by overnight incubation at 4°C with either anti-human or anti-mouse GlcNAc6ST-2. Endogenous peroxidase was blocked in 0.3 % H~2~O~2~in methanol. Immunoperoxidase detection was performed by ABC method according to the manufacturer (Vectastain^®^, Vector Laboratories, Burlingame, CA). Peroxidase activity was developed by diaminobenzidine substrate and slides were counterstained in Gills hematoxylin. Controls with non-immune rabbit IgG at the same concentration as anti-human or anti-mouse GlcNAc6ST-2 antibodies were included. Additional controls were performed by preincubating sections with human or mouse GlcNAc6ST-2 peptide at 10 μg/ml. To ensure identical conditions, RA and OA sections were mounted on the same slide and all slides were processed in parallel.
GlcNAc6ST-2 expression in HUVEC cultures
----------------------------------------
Human umbilical vein endothelial cells (HUVEC) were prepared from umbilical cords by collagenase digestion and were propagated in medium 199 (Life Technologies, Paisley, Scotland) with 20% FCS. Cultured cells were characterized as EC by flow cytometry with anti-endothelial cell antibody P1H12 (Chemicon International, Temecula, CA). The presence of HEC in these cultures was excluded by flow cytometry with MECA-79 mAb.
HUVEC cultures were stimulated with either 25 ng/ml of TNF-α, 20 ng/ml of LT-α1/β2, or 200 ng/ml of interferon-γ (R&D Systems, Inc., Abingdon, UK) for 8--24 h and total RNA or protein were extracted. For immunofluorescent detection of GlcNAc6ST-2, HUVEC were cultured on glass coverslips and similarly treated. Coverslips were fixed with 4% paraformaldehyde and immunofluorescent detection of GlcNAc6ST-2 was performed as described above for tissue sections.
RT-PCR was performed on cDNA synthesized from 1 μg of total RNA of EC cultures using the following *GlcNAc6ST-2*oligonucleotides: upstream 5\'-GCAGCATGAGCA AAAACTCAAG-3\' and downstream 5\'-TCCAGGTAGACAGAAGATCCAG-3\', and β-actin oligonucleotides upstream 5\'-CTACCTCATGAAGATCCTCAC-3\' and downstream 5\'-GTCCACGTCACACTTCATGATG-3\'. Real-time monitoring of PCR reactions was performed on a Roche LightCycler instrument, using SYBR Green I kit (Roche Diagnostics, Mannheim, Germany). The specificity of the amplicons was checked by electrophoresis which demonstrated single transcripts of 449 bp (*GlcNAc6ST-2*) and 303 bp (*β-actin*) as expected. For relative quantification, we calculated the n-fold differential expression by the ΔC~t~method (Ct denoting the threshold cycle of PCR amplification at which product is first detected by fluorescence) that compares the amount of target gene amplification, normalized to the β-actin endogenous reference as previously described \[[@B27]\].
Western blot detection of GlcNAc6ST-2 was performed on HUVEC protein extracts (50 μg) electrophoresed on a 10% polyacrylamide gel and electrophoretically transferred to nitrocellulose filters. Membranes were incubated overnight with either rabbit anti-human GlcNAc6ST-2 or control anti-β actin (clone AC-15, Sigma-Aldrich Química, Spain) antibodies in 5% non-fat dried milk-TBST. The filters were washed and incubated for 1 h with secondary antibodies linked to peroxidase at 0.08 μg/ml (Santa Cruz Biotechnology, Santa Cruz, CA). Bands were visualized by an enhanced chemiluminiscence system (Pierce, Rockford, IL).
Authors\' contributions
=======================
JLP carried out the immunodetection studies and participated in the study design and coordination. BS performed the studies in cultured endothelial cells. MG and GP participated in the selection of patients, and the collection and preparation of synovial tissues. DT and MSS prepared the antibodies and developed the immunodetection techniques. SDR participated in the design and coordination of the study. All authors contributed to drafting the manuscript and have read and approved the final version.
Acknowledgements
================
This work was supported by grants 02/0057 and G03/152 from Fondo de Investigación Sanitaria (Spain) and by grants from the NIH to SDR (R37GM23547 and R01GM57411). G. Palao was supported by Fondo de Investigación Sanitaria. B. Santiago was supported by a grant from Abbott Laboratories to Fundación Española de Reumatología.
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Double immunofluorescent detection of the MECA-79 epitope and GlcNAc6ST-2 in human lymph node and RA synovial tissues.**Sections were simultaneously labelled with MECA-79 (red) and anti-GlcNAc6ST-2 (green) antibodies and photographed under appropriate filters. On the right panels, a merged image shows colocalization of both markers (yellow). (a) Lymph node. (b) RA synovium; Two MECA-79 negative and GlcNAc6ST-2 positive vessels are marked by arrows. Original magnification (a) 200×. (b) 150×. Selected areas within squared insets are shown on the lower right corner insets with higher magnification (630×).
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Immunoperoxidase staining of RA and OA synovial tissue with anti-GlcNAc6ST-2 antibodies.**RA (a-d) or OA (e, f) synovial tissue sections were immunostained with anti-human GlcNAc6ST-2 (a, b, e) or anti-mouse GlcNAc6ST-2 (c, d, f) antibody. Color was developed with DAB peroxidase substrate (brown color) and sections were counterstained with hematoxylin. Sections b and d are serial to a and c respectively, and were preincubated with human or mouse GlcNAc6ST-2 peptide to demonstrate the specificity of immunostaining. Original magnification: 400×.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Expression of GlcNAc6ST-2 in cultured HUVEC.**(a) Real time RT-PCR quantification of *GlcNAc6ST-2*mRNA in cultured HUVEC. HUVEC cultures were stimulated for 8 h with LT-αβ (LT), TNF-α (TNF), or interferon-γ (IFN) and mRNA obtained from stimulated and non-stimulated cells. Basal (non-stimulated) expression was set to 1 and the *GlcNAc6ST-2/β-actin*ratio in stimulated cells is shown (mean ± SD of three independent experiments, each including triplicate HUVEC cultures). (\*) p \< 0.05. (b) Western blot analysis of GlcNAc6ST-2 in cultured HUVEC stimulated for 24 h with LT-αβ (LT) or TNF-α (TNF), (data are representative of three independent experiments with different HUVEC lines).(c) Immunofluorescent detection of GlcNAc6ST-2 protein in cultured HUVEC endothelial cells non-stimulated (Basal) or stimulated for 24 h with LT-αβ (LT) or TNF-α (TNF). In control (CTRL) panel, anti-GlcNAc6ST-2 antibody was replaced by non-immune rabbit IgG (images are representative of three independent experiments).
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Summary of immmunodetection of the MECA-79 epitope and GlcNAc6ST-2 in RA tissues.
:::
Tissues Lymphocytic infiltration MECA-79 GlcNAc6ST-2
--------- -------------------------- --------- -------------
RA1 Follicular \+ \+
RA2 Follicular \+ \+
RA3 Follicular \+ \+
RA4 Follicular \+ \+
RA5 Diffuse \+ \+
RA6 Negative \+ \+
RA7 Negative \+ \+
RA8 Follicular \- \+
RA9 Diffuse \- \+
RA10 Negative \- \+
RA11 Diffuse \- \-
RA12 Negative \- \-
RA13 Negative \- \-
RA14 Negative \- \-
:::
|
PubMed Central
|
2024-06-05T03:55:55.304893
|
2005-3-7
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079838/",
"journal": "BMC Immunol. 2005 Mar 7; 6:6",
"authors": [
{
"first": "José L",
"last": "Pablos"
},
{
"first": "Begoña",
"last": "Santiago"
},
{
"first": "Durwin",
"last": "Tsay"
},
{
"first": "Mark S",
"last": "Singer"
},
{
"first": "Guillermo",
"last": "Palao"
},
{
"first": "María",
"last": "Galindo"
},
{
"first": "Steven D",
"last": "Rosen"
}
]
}
|
PMC1079839
|
Background
==========
The colonization of the urogenital tract of humans by the protozoan parasite *Trichomonas vaginalis*is responsible for trichomonosis \[[@B1]\], the most prevalent, non-viral sexually transmitted infection worldwide. Despite an estimated 8 million new cases per year in the United States alone \[[@B2]\], this health disparities disease \[[@B3]\] remains poorly studied. *T. vaginalis*infection is associated with adverse health consequences to both men and women, including infertility \[[@B4],[@B5]\], atypical pelvic inflammatory disease \[[@B6]\], and increased HIV transmission \[[@B7],[@B8]\]. Trichomonosis is also associated with preterm birth, low birth weight infants \[[@B9]\], predisposition to development of cervical neoplasia \[[@B10]\] in women and non-gonococcal urethritis \[[@B11]\] and chronic prostatitis \[[@B12]\] in men. *T. vaginalis*adherence to host VECs, a step preparatory to infection \[[@B13]\], is complex and involves four surface protein adhesins. Three adhesins studied to date at the molecular level share identity with metabolic enzymes of the hydrogenosome organelle \[[@B14],[@B15]\]. Thus, the proteins exhibit functional diversity based on cellular location. AP65 is a prominent *T. vaginalis*adhesin that is targeted both to the surface and hydrogenosomes, and this compartmentalization is in part modulated by iron \[[@B15]\]. Further, the surface placement of AP65 is increased upon contact with host cells \[[@B15]\]. Recently, silencing of *ap65*gene expression in *T. vaginalis*with antisense reinforced the role of AP65 as a major adhesion \[[@B16]\].
The related *Tritrichomonas foetus*is a causal agent of bovine fetal wastage \[[@B17],[@B18]\]. Adherence by *T. foetus*to bovine VECs appears equally complex as that seen for *T. vaginalis*except that in this case, adherence involves glycoconjugates \[[@B19],[@B20]\]. Interestingly, the interaction between *T. foetus*organisms and VECs is both cell- and species-specific \[[@B21],[@B22]\]. For example, *T. foetus*attaches to bovine VECs, MDCK and HeLa epithelial cells, which have been used previously for studies on the host-parasite interaction. However, the bovine trichomonads have low background levels of adherence to human VECs. This may suggest that the phylogenetically-related *T. vaginalis*and *T. foetus*trichomonads differ in mechanisms of host parasitism and cytopathogenicity. We hypothesized that, given the different mechanisms of host VEC adherence by *T. foetus*and *T. vaginalis*, the heterologous expression in *T. foetus*of the *T. vaginalis*AP65 could be an alternative approach to confirm the function of AP65 as an adhesin.
Importantly, no cross-hybridization and immuno-crossreactivity with *ap65*and mAbs to AP65, respectively, were evident in *T. foetus*. We now show the stable expression in *T. foetus*of a single copy of the *T. vaginalis ap65*gene encoding an \~65-kDa protein \[[@B23]\]. The stable transfectants expressed functional AP65 on the *T. foetus*surface. These findings importantly illustrate the utility of a stable transfection system for heterologus expression in *T. foetus*of *T. vaginalis*virulence genes.
Results
=======
Low levels of adherence to immortalized human MS-74 VECs and lack of ap65 by *T. foetus*
----------------------------------------------------------------------------------------
We wanted to perform comparative binding experiments between *T. vaginalis*and *T. foetus*to epithelial cells used in adherence assays \[[@B15]\]. Given the fact that *T. foetus*exhibits host tissue- and cell-specificity \[[@B24]\], it was not surprising to find the lower levels of adherence by *T. foetus*to HeLa epithelial cells and immortalized human MS-74 VECs. As shown in representative experiment in Figure [1](#F1){ref-type="fig"}, at best only 40% levels of adherence were seen for *T. foetus*compared to *T. vaginalis*.
We next wanted to know if *T. foetus*had mRNA that hybridized to *ap65*used as a probe. As shown in Figure [1B](#F1){ref-type="fig"}, no RNA band was detected by Northern analysis for *T. foetus*(lane labeled Tf). In contrast, a band of \~2-kb was readily hybridized by total RNA of *T. vaginalis*. MR100, the drug-resistant *T. vaginalis*that is unable to adhere to host cells as shown recently also had no detectable *ap65*transcript, as a control \[[@B15]\]. The equivalent amounts of rRNA bands seen in Figure [1C](#F1){ref-type="fig"} illustrate that similar amounts of RNA were added to the lanes used in the Northern blots (Figure [1B](#F1){ref-type="fig"}).
Episomal expression in *T. foetus* of *ap65*
--------------------------------------------
The expression plasmid pBS-*ap65-neo*contains the neomycin (*neo*) gene, and as done recently, the transfectants were selected for Geneticin resistance at 100μg/ml \[[@B16]\]. As shown in Figure [2A](#F2){ref-type="fig"}, the presence of plasmid in transfectants was verified by PCR amplification of the *neo*coding region. A band of 750-bp was detected in the transfectants corresponding to pBS-*ap65-neo*(lane3) and pBS-*neo*(lane 4). Not unexpectedly, no PCR products were obtained using *T. vaginalis*(lane 1) and *T. foetus*(lane 2).
We then checked for the episomal gene expression by means of RT-PCR. The PCR products following reverse transcription were separated on 1% agarose gels. Figure [2B](#F2){ref-type="fig"} confirms the presence of an *ap65*band corresponding to 580-bp amplified from the transfectants with pBS-*ap65-neo*(lane 3), confirming the expression of *ap65*from the plasmid. The same band is also seen in *T. vaginalis*(lane 1). As an internal control in these experiments, the common band of 650-bp seen in all lanes corresponds to the *a-tubulin*gene. This control was important to show the similar amounts of RNA used for RT-PCR in all samples. These results demonstrate that the episomal plasmid with the *T. vaginalis ap65*gene is successfully expressed in transfected *T. foetus*.
*T. vaginalis* AP65 and AP65-HA are expressed in *T. foetus*, and episomal AP65 has function
--------------------------------------------------------------------------------------------
We next wanted to confirm that the episomal *ap65*transcript seen in Figure [2](#F2){ref-type="fig"} was translated. A 65-kDa band was readily detected by mAb 12G4 to AP65 in immunoblots of extracts of total proteins from *T. foetus*transfected with pBS *neo-ap65*(Figure [3A](#F3){ref-type="fig"}, lane 2), and the protein had the same size as AP65 detected in total protein blots of *T. vaginalis*(lane 4). There were no detectable bands in either wild type *T. foetus*(lane 1) or *T. foetus*transfected with the control plasmid (lane 3). This result also confirms that the 12G4 mAb did not cross-react with any protein in blots of *T. foetus*.
Importantly, we further tested for the ability of episomal AP65 in *T. foetus*to bind to host cells using a ligand assay \[[@B15],[@B25]\]. Figure [3B](#F3){ref-type="fig"} (lane 4) shows a typical result of a ligand assay using *T. vaginalis*, and mAb 12G4 used as a probe detected AP65 on blots of proteins that bound MS-74 VECs that were solubilized for electrophoresis and blotting. Likewise, AP65 from transfected parasites bound to MS-74 VECs (lane 2) and gave a band with the same M~r~. There were no detectable protein bands for wild-type *T. foetus*(lane 1) and *T. foetus*transfected with control plasmid (lane 3) handled identically in the ligand assay. These results show the ability of *T. foetus*transfectants to stably express functional AP65. Further, the lack of any immuno-crossreactivity with mAb is consistent with the absence of transcript (Figure [1B](#F1){ref-type="fig"}) and illustrates the absence of an AP65-like protein reactive with mAbs in *T. foetus*.
AP65 expressed in *T. foetus* is compartmentalized to the surface and hydrogenosomes
------------------------------------------------------------------------------------
Next, we performed cell membrane fractionation experiments to determine whether episomally-expressed AP65 in *T. foetus*transfectants localized to the surface membranes. Fractions enriched for membranes and hydrogenosome organelles were obtained by differential centrifugation, and the fractions were used for monitoring the presence of AP65 using a ligand assay. Figure [4A](#F4){ref-type="fig"} illustrates AP65 detected by mAb 12G4 in blots of total proteins from membranes and hydrogenosomes of *T. vaginalis*(lanes 2 and 5) and *T. foetus*transfected with the pBS-*ap65-neo*plasmid (lanes 3 and 6). As above, no bands were visualized on blots with total proteins of wild type *T. foetus*(lanes 1 and 4). We then used these fractions to perform a ligand assay, and as seen in Figure. [4B](#F4){ref-type="fig"}, both membrane and hydrogenosome fractions of *T. vaginalis*(lanes 2 and 5) and transfected *T. foetus*(lanes 3 and 6) had AP65 that bound to MS-74 VECs. Wild type *T. foetus*had no protein detected by mAb from the ligand assay (lanes 1 and 4).
Immunofluorescence detects AP65 in transfectants on the surface and in hydrogenosomes
-------------------------------------------------------------------------------------
It was then important to examine the possible placement and accessibility of episomal AP65 on the surface of *T. foetus*. Figure [5](#F5){ref-type="fig"} presents immunofluorescence data with *T. vaginalis*showing mAb 12G4 detecting surface AP65 (A4) and mAb F11 reacting with hydrogenosomal AP65 (decarboxylating malic enzyme) (B4), data consistent with that shown in a recent report by us \[[@B16]\]. Importantly, as with the fractionation experiments performed above, mAb 12G4 detected AP65 on the surface of non-permeabilized transfected *T. foetus*(Figure [5A2](#F5){ref-type="fig"}), and mAb F11 was reactive with AP65 within hydrogenosomes of permeabilized trichomonads (Figure [5B2](#F5){ref-type="fig"}). The absence of any fluorescence for wild type *T. foetus*(A1 and B1) and *T. foetus*with control plasmid (A3 and B3) reinforced the specific binding of mAbs with episomal AP65 and reaffirmed results from Figures [3](#F3){ref-type="fig"} and [4](#F4){ref-type="fig"} above. The brightfield pictures are included to show the integrity of the trichomonads in the fluorescence experiments. These data further confirm that episomal AP65 is sorted both to the parasite surface and to hydrogenosome organelles, suggesting that the trafficking signals are present within the AP65 open reading frame and that *T. foetus*also possesses the protein compartmentalization machinery similar to that of *T. vaginalis*.
Episomal expression of fusion AP65-HA and sorting to the parasite surface
-------------------------------------------------------------------------
Having established surface AP65 on transfected *T. foetus*, we next wanted to confirm surface placement using a different protein construct that would be detected by a distinct mAb. We, therefore, expressed the AP65-HA fusion protein, and immunoblot of total proteins of transfected *T. foetus*shows a band detected by the anti-HA mAb (Figure [6A](#F6){ref-type="fig"}, lane 5). As expected, the mAb 12G4 to AP65 detected a protein band in blots of total proteins of *T. vaginalis*(lane 1) and the pBS-*ap65-HA-neo*transfected *T. foetus*(lane 2). The M~r~of the fusion protein was higher than AP65, as expected. Also, no bands were apparent with any mAbs in *T. foetus*with control pBS-*neo*plasmid (lanes 3 and 6).
Finally, as shown in Figure [6B](#F6){ref-type="fig"}, immunofluorescence microscopy gave a positive reaction with anti-HA mAb with AP65-HA protein on the surface using non-permeabilized and in hydrogenosomes with permeabilized transfected (Tf-pBS-*ap65-neo*) trichomonads. The *T. foetus*transfected with the control plasmid (Tf-pBS-*neo*) was unreactive with anti-HA mAb and gave results identical to those shown for the wild type *T. foetus*. Finally, in data not shown, no fluorescence with anti-HA mAb was obtained using *T. vaginalis*organisms as controls under identical experimental conditions. These data using a different mAb to the fusion protein construct independently demonstrated the surface placement of episomal AP65 and AP65-HA on *T. foetus*.
Episomally-expressed AP65 in *T. foetus* is related to enhanced adherence to VECs
---------------------------------------------------------------------------------
Finally, we wanted to perform a functional adherence assay to determine if episomal, surface-expressed and host cell-binding AP65 on *T. foetus*, as shown collectively above, would enhance adherence of parasites to immortalized human MS-74 VECs. Figure [7](#F7){ref-type="fig"} presents representative data showing that transfected parasites with pBS-*ap65-neo*were 40% more adherent to VECs when compared to the background level of adherence of wild type *T. foetus*(Tf). *T. foetus*with control plasmid gave lower adherence levels equal to wild type parasites (not shown). Importantly, the enhanced adherence was inhibited by 36 % in the presence of rabbit polyclonal anti-AP65 IgG antibody (striped bar), indicating specific AP65-mediated adherence, as before for *T. vaginalis*\[[@B15],[@B16]\]. Control normal rabbit serum IgG did not inhibit adherence of the pBS-*ap65-neo*transfected *T. foetus*(shaded bar). Levels of adherence were compared with those of *T. vaginalis*(Tv) and normalized to 100%.
Discussion
==========
We have shown that trichomonads possess surface adhesins with functional diversity \[[@B26]\]. The adhesins are also enzymes in hydrogenosome organelles involved in energy generation. Although complex and multifactorial, the process of *T. vaginalis*adherence to human VECs is mediated in part by AP65, the surface protein that plays a major role in adherence \[[@B15],[@B16]\]. In this study, we established the utility of the *T. foetus*bovine trichomonad as a heterologous expression model system. We feel that among the noteworthy findings of this report are the following: 1) We show the stable episomal expression of the *T. vaginalis*prominent *ap65*gene (Figure [2](#F2){ref-type="fig"}). 2) The transcript is translated to form functional AP65 adhesin (Figure [3](#F3){ref-type="fig"}). 3) Episomal AP65 and fusion AP65-HA are compartmentalized to the surface and hydrogenosomes and are accessible to recognition by specific anti-AP65 and anti-HA mAbs (Figures [4](#F4){ref-type="fig"} through [6](#F6){ref-type="fig"}). 4) AP65 on transfected *T. foetus*increased adherence to VECs compared to parasites transfected with control plasmid (Figure [7](#F7){ref-type="fig"}), and the enhanced binding was reduced with anti-AP65 antibodies, showing specific AP65 mediated VEC attachment.
The stable expression of *T. vaginalis*AP65 in transfected *T. foetus*allowed us to characterize via cell fractionation experiments the cellular location of the episomal protein. Remarkably, AP65 fractionated to the plasma membrane and hydrogenosomes (Figure [4](#F4){ref-type="fig"}). This fractionation data was reaffirmed by fluorescence experiments (Figure [5](#F5){ref-type="fig"}), and the ability to detect both surface AP65 and surface AP65-HA using distinct mAbs provides strong evidence for the trafficking of *T. vaginalis*proteins to distinct cellular compartments in the bovine trichomonad. These results indicate that machinery for recognition of the surface and organelle targeting sequences is similar among these phylogenetically-related trichomonad species. Importantly, the expression of AP65-HA now will permit future subclone analysis for identification of the protein region within AP65 directing surface channeling.
That surface AP65 on transfected *T. foetus*elevated levels of adherence to human VECs (Figure [7](#F7){ref-type="fig"}) is significant because human VECs are not the natural host cell for *T. foetus*\[[@B27],[@B28]\]. The net increase in binding levels above *T. foetus*wild type controls were specific, as evidenced by the fact that anti-AP65 IgG reduced adherence to original control values. It is not inconceivable that the experimental conditions used here would not increase adherence by transfected *T. foetus*to those seen by *T. vaginalis*. One possible explanation is the fact that efficient and optimal adherence to VECs requires at least three additional adhesins, something that has been experimentally verified \[[@B15],[@B16]\]. Alternatively, the copy number of episomal AP65 on *T. foetus*may not be optimal for production of amounts of adhesin needed for elevated adherence levels. Such a difference in copy number between AP65 on *T. vaginalis*and *T. foetus*(Fig. [3](#F3){ref-type="fig"}) may not be obvious based solely on fluorescence intensity, as this is not quantitative. In this regard, it may be that the equivalent protein of *T. foetus*decarboxylating malic enzyme is also surface expressed thereby interfering with sequestration of sufficient molecules of episomal AP65 for adherence. The future availability of specific antibodies to the *T. foetus*decarboxylating malic enzyme will be useful for testing this possibility. It is noteworthy that the *ap65*gene and mAbs to AP65 did not cross-hybridize and immunoreact, respectively, with the equivalent decarboxylating malic enzyme gene and protein known to reside in hydrogenosomes of *T. foetus*\[[@B29],[@B30]\]. These data also now indicate that there is significant nucleotide and amino acid sequence divergence between *T. vaginalis*and *T. foetus*in these equivalent genes and proteins. Importantly, except for the N-terminal sequence submitted to the GenBank (Accession number AAK55143.1), to our knowledge the entire decarboxylating malic enzyme gene of *T. foetus*has not been characterized \[[@B31]\]. Finally, not all episomal, surface-expressed AP65 molecules may be in the proper orientation to have adhesive function. Nonetheless, the significant increase in adherence attributable to episomal AP65 by transfected *T. foetus*once again underscores the role of AP65 in adherence by *T. vaginalis*.
This paper now provides evidence for the functional expression of a heterologous virulence factor in a related trichomonad species. This ability to express *T. vaginalis*or other foreign proteins using *T. foetus*as a model is particularly noteworthy, especially if an identical and/or equivalent virulence factor and property are not detectable within the bovine trichomonad. We feel that this approach is innovative and represents a significant advance for understanding the contribution of specific *T. vaginalis*virulence factors to overall mechanisms of trichomonal pathogenesis.
Methods
=======
Parasite and cell cultures
--------------------------
*Trichomonas vaginalis*isolate T016 and *Tritrichomonas foetus*(02--97) were grown in Trypticase-yeast extract-maltose (TYM) medium supplemented with 10% heat-inactivated horse serum \[[@B32]\]. Immortalized human MS-74 VECs were grown as detailed recently by us \[[@B15],[@B16]\] in D-MEM supplemented with 10% fetal bovine serum. VECs were grown at 37°C in a 5% CO~2~atmosphere.
Plasmid construction
--------------------
The plasmid pBS-*ap65-neo*was constructed by cloning the coding region of AP65-3 in pBS-*FdHAHA-neo*(forward primer 5\' GTCCAGCATATGATGCTCGCATCTTCA GTC-3\' and reverse primer 5\'-GTCCACGGTACCTTAGTAGAGTTGCTCGTATTC-3\'). The parent plasmid pBS-*FdHAHA-neo*\[[@B33]\] was used by us recently \[[@B16]\]. The parent plasmid was partially digested and the 1.7-kb *ap65-3*gene was cloned into the *NdeI*and *Asp718*sites giving rise to pBS-*ap65-neo*. In order to generate AP65 with the HA tag, the stop codon of AP65 was mutated and cloned into pBS-*FdHAHA-neo*resulting in pBS-*ap65-HA-neo*. The new plasmids were confirmed by sequencing. Plasmid DNA for transfection was purified using maxi prep columns (Qiagen, Inc., Valencia, CA).
Stable transfection and selection for G418 resistance
-----------------------------------------------------
Transfection of *T. foetus*cells was carried out by electroporation \[[@B34]\]. Parasites at early logarithmic phase of growth were used for transfection. Briefly, 4 × 10^7^parasites were centrifuged at 1,800 rpm at 4°C, and the pellet was resuspended in 400 μl fresh TYM before transferring into a 4-mm gap cuvette (BTX^®^, Genetronics, Inc., San Diego, CA) with 25 μg of plasmid DNA. Electroporation was performed at 320 V, 1000 microfarads and 725 ohms using the ECM 630 Electro cell manipulator (BTX^®^). Following the pulse, cells were placed on ice for 10 min and transferred into two T25 flasks with 50 ml of fresh TYM-serum medium. The cells were grown free of drug for 24 h followed by the addition of Geneticin (Invitrogen-Life Technologies, Carlsbad, CA) at 100 μg ml^-1^. Single cells were cloned using soft-agar plates \[[@B35]\]. The DNA was isolated from single cell cultures using DNAZol (Invitrogen) and further purified by phenol-chloroform extraction. The presence of plasmid in single cell clones was confirmed by PCR amplification of the *neo*gene. The sense 5\' GATCGGTACCATGATTGATTGAACA AGATGGATTG-3\', and antisense 5\'-CTTTAGACCAAGTTCGTGTCAGAAGAACT CGTCAAG-3\' primers were used in the PCR reaction.
RNA isolation and RT-PCR analysis
---------------------------------
Total RNA was isolated using the Trizol reagent (Invitrogen). Briefly, 1 μg of total RNA was reverse transcribed using SuperScript II RNase H^-^Reverse Transcriptase (Invitrogen) in a reaction volume of 20 μl. One μl of the reverse transcribed cDNA was used as template for the PCR amplification. The *ap65*primer sequences used for amplifications were *ap65*-sense 5\'-CAGTCAGTCGACCAGTTAGATATGGGTACAGAC-3\' and *ap65*-antisense 5\'-GTGACAGGATCCCGCTCGCAGTTAGCGCATGTAG-3\'. The *α-tubulin*gene primers were sense 5\'-ACTCTGCTGCCTCGAGCACGGTATC-3\' and antisense 5\'-GAAATGACTGGTGCATAAGAGC-3\'. The common annealing temperature for the primers of both genes allowed for the *ap65*and *α-tubulin*primers to be included in the same PCR reaction.
Immunoblot detection of episomally-expressed AP65
-------------------------------------------------
Total protein from parasites was obtained as before using trichloroacetic acid (TCA) precipitation \[[@B36]\] and separated on sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) \[[@B37]\] prior to blotting onto Hybond-P membranes (Amersham Pharmacia Biotech, Pisscataway, NJ) for immunoblot detection with mAb 12G4 to AP65 and anti-HA mAb. The mAbs and epitope reactivity were described recently \[[@B15]\]. Following blot reactivity with the mAbs, the bands were visualized by the chemiluminescence assay using horseradish peroxidase as the color developer (BioRad Laboratories, Hercules, CA).
Sub-cellular fractionation
--------------------------
The hydrogenosomes were isolated as recently detailed \[[@B15]\]. Briefly, 1 × 10^8^trichomonads were homogenized in buffer (250 mM sucrose, 20 mM KCL, 10 mM KH~2~PO~4~, 5 mM MgCl~2~and 20 mM Tris-HCl, pH 7.0). The homogenate was centrifuged to sediment nuclei and costae at 1,500 g for 10 min at 4°C. The supernatant was then transferred to a new tube and centrifuged at 12,000 × g for 10 min. The sediment was then washed once more with homogenization buffer and was highly enriched for hydrogenosomes, which was confirmed biochemically as recently described \[[@B16]\]. The sediment was resuspended in the buffer and stored at -70°C. Membrane isolation was also carried out as described recently \[[@B38]\]. *T. foetus*(1 × 10^8^) were resuspended in phosphate-buffered saline (PBS), sonicated, and subjected to low-speed centrifugation at 500 × g for 15 min at 4°C. The resulting supernatant was further fractionated by sequential differential centrifugations of 1,500 × g, 10,000 × g, and 100,000 × g for 30 min at 4°C. The supernatant was used for protein analysis.
Ligand assay to determine the amounts of adhesins on the surface
----------------------------------------------------------------
The ligand assay to detect adhesins that bind to host cell surfaces was carried out as before \[[@B25]\]. Briefly, after fixation of the MS-74 VECs with glutaraldehyde, 10^6^epithelial cells were incubated with a trichomonal detergent extract derived from 2 × 10^7^solubilized parasites. After incubation, cells were vigorously washed to remove unbound and loosely-associated trichomonad proteins. Cells were boiled in electrophoresis dissolving buffer to elute the proteins bound to the MS-74 VEC followed by SDS-PAGE. The gels were further stained with Coomassie Brilliant blue for visualization, and duplicate gels were blotted onto Hybond-P membranes for immunoblot analysis using the mAbs as probes.
Immunoprecipitation studies
---------------------------
AP65 from the membranes and hydrogenosomes derived from the fractionation experiments was immunoprecipitated using mAb DM116 \[[@B15]\]. A standard immunoprecipitation protocol was used \[[@B39]\]. Briefly, the fractions (0.5 mg) prepared as mentioned above, were mixed with 5.0 μg of mAb DM116 for 4 h at 4°C. The extract and mAb were mixed on a rotating platform for 3 h with Protein A-Sepharose beads. The beads were washed three times with PBS, and samples were eluted by boiling for 10 min in SDS dissolving buffer. Eluted proteins were resolved by SDS-PAGE, and immunoblot analyses and ligand assays were performed as described above.
Immunofluorescence detection of AP65 on the surface
---------------------------------------------------
Immunofluorescence of AP65 on the surface and in hydrogenosomes of trichomonads was carried out using a modification of a recently-described procedure \[[@B15]\]. Briefly, 1 × 10^6^logarithmic-phase organisms were washed twice with cold PBS and fixed with 4 % paraformaldehyde for 10 min at RT. Fixed cells were washed in PBS and permeabilized with 1% NP-40 for 45 min at RT. Trichomonads were then blocked with 5% BSA for 1 h at RT prior to incubation for an additional 1 h at RT with hybridoma supernatants of mAb 12G4 (1:100) and anti-HA mAb F11 (1:1000). Parasites were washed with PBS and incubated for 1 h at 37°C with fluorescein isothiocyanate-conjugated anti-mouse IgG (Sigma) diluted 1:100. Finally, parasites were washed twice with PBS and observed under 1000× magnification using the Olympus BX41 microscope.
Adherence assay
---------------
Immortalized human MS-74 VECs \[[@B40]\] were used for the adherence assay as recently described \[[@B15]\]. Briefly, 2 × 10^4^MS-74 VECs were seeded onto individual wells of 96-well Costar flat bottom plates (Corning Inc., Corning, NY) and grown for 24 h in DMEM supplemented with 10 % fetal bovine serum (FBS). VECs were then washed twice with a medium mixture of DMEM:TYM (2:1; v/v) without serum. Trichomonads were labeled with \[^3^H\] thymidine for 18 h, washed three times with DMEM-TYM and resuspended. Then, 4 × 10^5^tritium-labeled parasites were added to the individual wells of the 96-well plate with a confluent MS-74 VEC monolayer and incubated for 30 min at 37°C. Cells were then washed thoroughly with the DMEM-TYM. Individual wells were placed in mini-scintillation vials and radioactivity was measured. The assay was performed with quadruplicate samples.
Reproducibility of experiments
------------------------------
Unless otherwise stated in the text, all experiments were performed numerous times and no less than on four different occasions.
Abbreviations
=============
AP65, adhesin protein Mr 65-kDa; DMEM, Dulbecco\'s minimum essential medium; HA, hemagglutinin; kb, kilobase; M~r~, relative molecular weight; pBS-*ap65-neo*, plasmid construct with the *ap65*and *neomycin*genes; Tf, *Tritrichomonas foetus*; Tv, *Trichomonas vaginalis*; TYM, trypticase-yeast extract-maltose; VECs, vaginal epithelial cells.
Authors\' contributions
=======================
ASK carried out the design of the study, performed transfections, Western blots, immunofluorescence, adherence assays and drafted the manuscript. VM constructed the plasmids, performed the Northern analysis, RT-PCR and wrote parts of the manuscript. JFA participated in the design of the experiments, offered suggestions during the experiments, and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
================
This work was supported by Public Health Service grants AI43940 and AI45429 from the National Institutes of Health. Members of the laboratory are also acknowledged for their suggestions and discussion of our work.
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
***Tritrichomonas foetus*has lower levels of adherence to HeLa epithelial cells and immortalized human MS-74 VECs and has no detectable *ap65*adhesin gene**. (A) Bar graph showing the mean level of adherence by *T. foetus*(shaded bars) derived from four independent adherence experiments, each of which was performed with quadruplicate samples. Adherence levels of *T. vaginalis*(open bars) were normalized to 100% for comparative purposes. (B) Northern analysis to detect *ap65*mRNA amounts was performed using 10 μg of total RNA per lane separated on 1.2% agarose-formaldehyde gels and transferred to Hybond-P membranes. The blot was probed with DIG-labelled *ap65*, which hybridized to an \~2-kb *ap65*transcript in *T. vaginalis*, as before \[16\]. No similar signal was detected in *T. foetus*and the drug resistant MR100 *T. vaginalis*used as a negative control \[15\]. (C) Ethidium bromide (EtBr)-stained RNA gel showing the rRNA bands. Equal loading of RNA in all the lanes is evident by similar amounts of the rRNA bands, which served as internal controls for the Northern blot in part B.
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Transfection and episomal expression of *ap65*in *T. foetus***. (A) Agarose gel stained with EtBr shows results of a PCR reaction to amplify the 795-bp *neo*gene from transfected *T. foetus*total genomic DNA. The *neo*gene PCR product is detected only in *T. foetus*transfected with pBS-*ap65-neo*(lane 3) and pBS-*neo*(lane 4), but not the wild type *T. foetus*(Tf; lane 2) and control *T. vaginalis*(Tv; lane 1). (B) Agarose gel stained with EtBr shows RT-PCR products for the *ap65*transcript in wt *T. foetus*(lane 1), *T. vaginalis*(lane 2), *T. foetus*transfected with pBS-*ap65-neo*(lane 3), and *T. foetus*transfected with the pBS-*neo*plasmid as a negative control. RT-PCR was performed using primers to amplify a 580-bp region of the *ap65*gene and a 650-bp region of the a-tubulin gene used as an internal control.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Monoclonal antibody (mAb) detects AP65 expressed in pBS-*ap65-neo*(Parts A and B) and pBS-*ap65-HA-neo*(Part C) transfected *T. foetus*parasites**. (A) Total proteins from 10^7^parasites were separated on 10% SDS-PAGE and blotted onto Hybond-P membranes for probing with mAb DM116 specific for AP65, as before (Garcia *et al*., 2003). The AP65 protein was readily detected in total protein blots of transfected *T. foetus*(lane 2) and *T. vaginalis*(lane 4), but not in wt *T. foetus*(lane 1) and in *T. foetus*transfected with the control pBS-*neo*plasmid (lane 3). The migration of detected AP65 was \~65-kDa as expected based on molecular weight standards. (B) A ligand assay showing AP65 binds to immortalized human MS-74 VECs, as before (Garcia *et al*., 2003). Parasites bound to VECs were solubilized and electrophoresed as above for blotting onto Hybond-P membranes. The DM116 mAb was used to detect AP65 in transfected *T. foetus*(lane 2) and *T. vaginalis*(lane 4). No protein was detected in wt *T. foetus*(lane 1) and *T. foetus*transfected with the control plasmid (lane 3).
:::
:::
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Immunoblots with specific mAb 12G4 detecting AP65 immunoprecipitated from extracts of total trichomonad protein preparations (A) and of proteins after the ligand assay (B) derived from purified hydrogenosomes (lanes 1 through 3) and membrane fractions (lanes 4 through 6)**. AP65 was immunoprecipitated with mAb from protein extracts from hydrogenosomes and membrane fractions of *T. foetus*(*Tf*; lanes 1 and 4), *T. vaginalis*(*Tv*; lanes 2 and 5), and transfected *T. foetus*(Tf-pBS-*ap65-neo*; lanes 3 and 6). Immunoprecipitated AP65 was prepared as described in the Experimental design section and used in a ligand assay to monitor the amount of AP65 adhesin bound to MS-74 VECs as shown above in Figure 3.
:::
:::
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Immunofluorescence and corresponding brightfield microscopy showing pBS-*ap65-neo*transfected *T. foetus*(Tf-pBS-*ap65-neo*) expressing AP65 on the surface of non-permeabilized trichomonads (A2) and in hydrogenosomes in permeabilized parasites (B2)**. The distinct patterns were obtained with the mAbs 12G4 for surface fluorescence and F11 for hydrogenosome fluorescence shown recently to detect the protein in respective compartments (Garcia *et al*., 2003). *T. vaginalis*fluorescence patterns (Tv; A4 and B4) were positive controls, and results are identical to those obtained recently (Garcia *et al*., 2003). No evidence of fluorescence was seen for wild type *T. foetus*(Tf) and for *T. foetus*transfected with control plasmid (Tf-pBS-*neo*). Control hybridoma supernatant lacking anti-AP65 mAb was unreactive with trichomonads.
:::
:::
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**Immunoblot detection of AP65-HA in *T. foetus*transfected with pBS-*ap65-HA-neo*, and immunofluorescence microscopy detecting AP65-HA protein on the surface (non-permeabilized) and in hydrogenosomes (permeabilized)**. (A) Duplicate immunoblots prepared from total proteins as above were probed with 12G4 mAb to AP65 (lanes 1 through 3) and anti-HA mAb (lanes 4 through 6). AP65 was readily detected by 12G4 in blots of *T. vaginalis*(lane 1) and *T. foetus*transfected with pBS-*ap65-HA-neo*(lane 2) but not *T. foetus*with control plasmid (lane 3). The anti-HA mAb detected the AP65-HA protein only in *T. foetus*transfected with pBS-*ap65-HA-neo*(lane 5). No protein was detectable by anti-HA mAb in blots of control *T. vaginalis*(lane 4) and *T. foetus*with the pBS-*neo*plasmid (lane 6). As an additional negative control for all immunoblots, no protein was detected in the absence of primary antibody or using control hybridoma supernatant. (B) Fluorescence is visualized using anti-HA mAb only in the *T. foetus*organisms transfected with the pBS-*ap65-HA-neo*plasmid but not the pBS-*neo*plasmid. Not unexpectedly, no fluorescence was seen using anti-HA mAb with *T. foetus*and with the use of control hybridoma supernatant lacking mAb to AP65. Finally, no fluorescence with anti-HA mAb was obtained using *T. vaginalis*organisms under identical experimental conditions.
:::
:::
::: {#F7 .fig}
Figure 7
::: {.caption}
######
***T. foetus*transfected with pBS-*ap65-neo*(Tf-pBS-*ap65-neo*) displays enhanced levels of adherence to immortalized human MS-74 VECs compared to *T. foetus*(Tf) parasites**. The percent level of adherence was adjusted with that seen for *T. vaginalis*(Tv). *T. foetus*with control plasmid gave levels of adherence similar to *T. foetus*(data not shown). The enhanced adherence obtained with transfected *T. foetus*(Tf-pBS-*ap65-neo*) was inhibited by anti-AP65 IgG (hatched middle bar). The solid gray bar (right) represents inhibition by normal rabbit serum (NRS) control. The results are the average from four different experiments, and each experiment was carried out using quadruplicate samples. The statistical significance of the results is indicated by the asterisk above the bar graph.
:::
:::
|
PubMed Central
|
2024-06-05T03:55:55.307686
|
2005-3-4
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079839/",
"journal": "BMC Mol Biol. 2005 Mar 4; 6:5",
"authors": [
{
"first": "Ashwini S",
"last": "Kucknoor"
},
{
"first": "Vasanthakrishna",
"last": "Mundodi"
},
{
"first": "JF",
"last": "Alderete"
}
]
}
|
PMC1079840
|
Background
==========
The calcitonin gene-related peptide (CGRP) belongs to a family of related peptides that includes calcitonin (CT), adrenomedullin (AM) and amylin (AMY) \[[@B1],[@B2]\]. To date, CGRP is one of the most potent endogenous vasodilatory peptides discovered. CGRP mediates sensory neurotransmission and inhibits insulin action on carbohydrate metabolism \[[@B2]\]. CGRP has been shown to modulate immune function by inhibiting the proliferation of T cells and synthesis of T cell-derived cytokines IL-2 and IFN-γ \[[@B3]-[@B6]\]. In the lung, CGRP mediates multiple effects some of which have potential implications in airway homeostasis \[[@B7]\]. CGRP has also been shown to have cardioprotective effects in rats and humans \[[@B8],[@B9]\]. In skeletal muscle, CGRP potentiates muscle contraction \[[@B10]\], increases the numbers of acetylcholine receptors (AchR) \[[@B11]-[@B13]\] and their rate of desensitisation \[[@B14]\]. In addition, CGRP locally increases the rate of blood flow following muscle contraction \[[@B15]-[@B17]\].
The effects of CGRP are mediated by CGRP receptors that are generated by a complex of proteins \[[@B18]\]. CGRP receptors are formed by the interaction of two separate proteins. The first protein component is the calcitonin receptor like (CL) receptor. The CL receptor is a seven transmembrane-domain receptor but is inactive when expressed in cells alone \[[@B19]\]. The second protein component required for CGRP receptor function is receptor activity modifying protein-1 (RAMP-1). RAMP-1 acts as a molecular chaperone and is required for the transportation of the CL receptor to the cell surface in addition to pharmacologic specificity \[[@B18]\].
RAMPs are a recently identified group of single transmembrane-domain accessory proteins. To date, three members of the RAMP family have been identified (RAMP-1, RAMP-2 and RAMP-3) \[[@B18]\]. All share 30% sequence identity, differ in their tissue distributions and are comprised of approximately 160 amino acids that make up a large extracellular N-terminal domain, a single membrane-spanning domain and a short cytoplasmic domain \[[@B18],[@B20]\]. Recently, Christopoulos et al. \[[@B21]\] demonstrated that RAMPS interact with a number of Class II G protein-coupled receptors (GPCRs) in addition to the CL receptor. These include the vasoactive intestinal polypeptide/pituitary adenylate cyclase activating peptide receptor (VPAC1R), the glucagon and parathyroid hormone receptors (PTH1 and PTH2). VPAC1R/RAMP-2 heterodimers display a significant enhancement of agonist-mediated phosphoinositide hydrolysis compared with VPAC1R alone. This suggests that RAMPs may play a more general role in modulating cell signalling through other GPCRs than previously thought.
Despite the crucial role RAMP-1 plays in the generation of CGRP receptors, little is known about the molecular mechanism(s) regulating RAMP-1 gene expression. It is likely that such mechanism(s) play an important role in modulating the responsiveness of specific tissues to CGRP. To initiate studies on RAMP-1 gene regulation, we cloned and characterised the mouse RAMP-1 gene promoter. Analysis in three different RAMP-1 positive mouse tissues revealed multiple start sites of transcription. RT-PCR analysis of RAMP-1 mRNA in the C2C12 myoblast cell line demonstrated that endogenous RAMP-1 gene expression was greatest in confluent cultures compared to non-confluent or fused cells. Through the use of RAMP-1 promoter/luciferase constructs transfected into C2C12 myoblast cells and NIH 3T3 cells, we studied the effects RAMP-1 promoter deletions had on RAMP-1 transcriptional activity. We further demonstrated that the RAMP-1 promoter activity was tissue-specific and not expressed in the RAMP-1 negative cell line HEK293. Finally, we identified a repressor element in the RAMP-1 promoter which when deleted activates RAMP-1 promoter activity in HEK293 cells. This suggests the RAMP-1 gene is negatively regulated in non-RAMP-1 expressing cells.
Results
=======
Determination of RAMP-1 transcriptional initiation sites and PCR cloning the mouse RAMP-1 promoter
--------------------------------------------------------------------------------------------------
To date, the transcriptional initiation sites for any mouse RAMP transcript have not yet been determined. We therefore performed 5\' rapid amplification of cDNA ends (RACE) analysis to map RAMP-1 transcriptional initiation sites. For these experiments, RNA from a number of known RAMP-1 positive mouse tissues (heart, brain and skeletal muscle) was employed. Sequence analysis of 8--10 5\' RACE clones from each tissue demonstrated multiple start sites of transcription (Table [1](#T1){ref-type="table"}). All transcriptional start sites in mouse heart, brain and skeletal muscle occurred within a 50 bp region directly 5\' to RAMP-1\'s start site of translation. As demonstrated by the DNA sequence analysis above, this region is highly GC rich (78%). These results suggest that RAMP-1 transcriptional initiation sites do not significantly vary between the different RAMP-1 positive tissues tested.
Having identified RAMP-1\'s major transcriptional initiation sites, we next wished to clone and characterise the mouse RAMP-1 gene promoter. Since extensive sequence data both 5\' and 3\' to RAMP-1\'s transcriptional start sites is available on the mouse genome data base (Accsession no. NT\_039173.2) we employed RAMP-1 specific primers and PCR on mouse genomic DNA to generate a DNA fragment containing 4.7 kb of RAMP-1 5\' flanking region. Promoter sequence motifs in the proximal 2.5 kb of 5\' flanking region corresponding to consensus binding sites for a variety of transcription factors are present (Fig. [1](#F1){ref-type="fig"}). These include binding sites for the transcription factors NFκB, PBX1, AML1, OCT1, MITF, TEF1 and Sp1. Of note is a 147 bp region (-2188 to -2336) containing multiple GAAA and GGAA repeat sequences and a highly GC rich (78%) region of 144 bp region immediately 5\' of the start site of translation (ATG). This GC rich region contains a potential binding site for the transcription factor Sp1.
RAMP-1 gene expression in C2C12 myoblasts and myotubes
------------------------------------------------------
Northern blot analysis has demonstrated that the RAMP-1 gene is expressed in mouse skeletal muscle \[[@B20]\]. Since a variety of muscle cell lines are widely available and are easily transfected with DNA constructs, we decided to employ such cell lines for our studies on the transcriptional regulation of the mouse RAMP-1 gene. Before undertaking transfection experiments in a muscle cell line, we first studied endogenous RAMP-1 gene expression in the myoblast cell line C2C12. Upon incubation with reduced serum levels and high cell density, C2C12 myoblasts fuse and form myotubes. Pittner et al. \[[@B22]\] previously demonstrated that incubation of this cell line with CGRP induces an increase in intracellular levels of cAMP making it highly likely this cell line expresses RAMP-1 mRNA. We therefore assayed RAMP-1 gene expression in this cell line by semi-quantitative RT-PCR analysis on total RNA from non-confluent, confluent and fused C2C12 myoblasts. Southern blot analysis of RT-PCR products from these cells demonstrated the presence of RAMP-1 mRNA under all culture conditions (Fig. [2](#F2){ref-type="fig"}). However, RAMP-1 mRNA levels appeared to be highest in confluent versus non-confluent and fused C2C12 myoblasts. Densitomitry demonstrated that RAMP-1 mRNA expression was at least 6 times higher in confluent vs. non-confluent myoblasts.
RAMP-1 promoter deletional analysis in transiently transfected C2C12 myoblasts
------------------------------------------------------------------------------
To delineate potential transcriptional regulatory elements in the 4.7 kb RAMP-1 promoter, a series of 5\' RAMP-1 promoter/luciferase constructs were generated. Luciferase analysis on transfected confluent C2C12 myoblasts revealed that a RAMP-1 promoter/luciferase construct with approximately 4.7 kb of 5\' flanking region (relative to RAMP-1\'s start site of translation) had the greatest levels of luciferase activity (30 fold over the background plasmid pGL3 basic). Deletion of approximately 700 bp of 5\' flanking region (to -4102) caused up to a 55% drop in expression levels. Deletions to -343 of RAMP-1 5\' region did not cause additional decreases in promoter activity. However, deletion to -164 reduced RAMP-1 promoter activity to 7-fold above background levels (Fig. [3](#F3){ref-type="fig"}). Similar results were obtained when RAMP-1 promoter/lucifersase constructs were transfected into RAMP-1 positive mouse NIH 3T3 cells.
To determine if the RAMP-1 promoter demonstrates tissue specificity in transfected cell lines, we compared the transcriptional activity of the RAMP-1 promoter in transfected C2C12 myoblasts and a cell line known to not express RAMP-1 mRNA (HEK293). Luciferase analysis demonstrated that constructs containing 4.7 kb to 782 bp of RAMP-1 5\' flanking region were transcriptionally inactive in HEK293 cells. A construct containing 470 bp of RAMP-1 5\' flanking region demonstrated modest transcriptional activity while a construct containing 343 bp of 5\' flanking region demonstrated significant promoter activity. The results of these experiments suggest that the RAMP-1 promoter region between -782 and -343 contains a regulatory element that represses RAMP-1 promoter activity in a non-RAMP-1 expressing cell line.
Silencer activity of the RAMP-1 repressor element
-------------------------------------------------
We wished to determine if the RAMP-1 repressor element could repress RAMP-1 promoter activity in a position and orientation independent manner (silencer function). We therefore employed PCR to generate a fragment (-343 to -782 of the RAMP-1 promoter) containing the RAMP-1 repressor element. The fragment was then cloned in two orientations 3\' to the luciferase gene in the -343 RAMP-1 promoter/luciferase construct. The expression of these constructs was compared to activity from the -343, -470 and -4724 RAMP-1 promoter/luciferase constructs in transfected HEK293 cells. Luciferase assays demonstrated that the repressor element reduced expression from the -343 RAMP-1/luciferase construct by approximately 50% (see Figure [4](#F4){ref-type="fig"}). An unpaired Student\'s t-test revealed that these differences were statistically significant (p \< .002 and p \< .004 for 5\'-3\' and 3\'-5\' silencer constructs, respectively). However, these constructs demonstrated significantly higher levels of activity than the -470 RAMP-1/luciferase construct (p \< .006 and p \< .003 for 5\'-3\' and 3\'-5\' silencer constructs, respectively). These results suggest that the RAMP-1 repressor element located between -343 and -782 bp of RAMP-1 5\' flanking region has modest but statistically significant silencer activity and functions best in its native position.
Discussion
==========
Despite the important role RAMP-1 plays in CGRP cell signalling little is known about the transcriptional regulation of RAMP-1 gene. To date, no RAMP-1 gene promoter has been identified/cloned and RAMP-1 transcriptional initiation sites have not been determined in any RAMP-1 positive tissues. In the present study, we cloned the mouse RAMP-1 gene promoter and mapped multiple RAMP-1 transcriptional initiation sites in a variety of RAMP-1 positive tissues. Our data demonstrates that RAMP-1 transcriptional initiation occurs at similar sites and all within a 50 bp region in different RAMP-1 mRNA positive tissues.
To initiate studies on the transcriptional regulation of the mouse RAMP-1 gene during muscle differentiation, we examined RAMP-1 gene expression in C2C12 mouse myoblasts by semi-quantitative RT-PCR. Our results suggest that RAMP-1 gene expression is present in non-confluent, confluent and fused C2C12 cells. However RAMP-1 mRNA expression appears to be increased in confluent and fused C2C12 cells. Chakravarty et al. \[[@B23]\] demonstrated that a strong correlation exists between the level of RAMP-1 mRNA expression and CGRP binding. This suggests that the transcriptional activity of the RAMP-1 gene may control the responsiveness of tissues to CGRP. It will be important to determine if increased RAMP-1 mRNA expression in confluent and fused C2C12 cells results in significantly higher levels of CGRP receptor expression and increases in intracellular cAMP after treatment with CGRP. The role increased RAMP-1 mRNA gene expression plays in the process of C2C12 induced muscle differentiation is not yet clear. Nobel et al. \[[@B24]\] showed that CGRP causes increased creatine kinase activity in myoblast cultures. Creatine kinase activity is a well established marker of fused myoblasts (myotubes). It is reasonable to hypothesise that increased RAMP-1 gene expression in confluent C2C12s makes these cells more responsive to CGRP and thus may facilitate myoblast fusion.
Deletional mapping of the RAMP-1 gene promoter revealed that a 4.7 kb promoter fragment generates the highest levels of RAMP-1 promoter activity. However, a RAMP-1 promoter construct with only 343 bp of 5\' region contains nearly 40% percent activity of the entire promoter. The delineation of such a small functional RAMP-1 promoter should facilitate the identification of DNA regulatory elements and the transcription factors they interact that activate RAMP-1 promoter activity in C2C12 cells.
Transfection of RAMP-1 promoter/luciferase constructs into RAMP-1 negative HEK293 cells demonstrated that a transcriptional control element present between -782 and -343 in the RAMP-1 promoter represses RAMP-1 promoter activity in the RAMP-1 negative-expressing cell line HEK293 cells. While this element appears to have modest silencing activity, our results suggest that full repressor activity of this element requires it to be in its native position within the RAMP-1 promoter. We hypothesise that negative-regulatory transcription factors, present in non-RAMP-1 expressing cells, bind to this element and repress RAMP-1 gene expression. Computer analysis of the 480 bp RAMP-1 repressor element has revealed potential binding sites for a variety of positive acting transcription factors such as MyoD, c-Myb, NF1 and Elk-1. Importantly, binding sites for two transcriptional repressors have also been identified: mammalian transcriptional repressor RBPJkappa/CBF (RBPJK) and bZIP domain, transcriptional repressor (E4BP4). Studies on the roles these transcription factors play in the repression of RAMP-1 gene expression are now underway.
Conclusion
==========
Promoter deletional analysis demonstrated that both positive and negative acting transcription factors regulate the tissue specificity of the RAMP-1 gene. DNase 1 protection and gel-shift experiments with extracts from C2C12 and HEK293 cells are underway to identify specific DNA sequences and the potential transcription factors they bind that mediate the activity of the RAMP-1 gene in RAMP-1 expressing and non- expressing cells. Once specific sequences/factors binding to the RAMP-1 repressor element are isolated, the effects that mutating these sequences has on RAMP-1 promoter activity in transgenic mice will be tested. These experiments will enable us to test the role such sequences/factors play in regulating the tissue-specific expression and activity levels of the RAMP-1 gene *in vivo*. Such transcriptional mechanism(s) are likely to play an important role in modulating the responsiveness of tissues to CGRP and may prove useful targets for future drug development aimed at regulating CGRP receptor gene expression.
Methods
=======
5\' RACE analysis
-----------------
5\' RACE on total RNA from mouse heart, skeletal muscle and brain was performed employing a First Choice RLM-RACE kit (Ambion) according to manufacturers instructions. Briefly, 10 μg of total RNA from each tissue was treated with Calf Intestine Alkaline Phosphatase (CIP) for one hour at 37°C followed by extraction with phenol:chloroform and ethanol precipitation. The RNA was re-suspended in 11 μl of H~2~0. 5 μl of this RNA was then treated with tobacco acid pyrophosphate (TAP) for one-hour at 37°C. A 5\' RACE adapter (provided in kit) was ligated to 5 μl of the TAP treated RNA. 2 μl of ligated RNA was subjected to reverse transcription (RT). RT products (1 μl) were subjected to two rounds of nested PCR according to manufacturers instructions. The first round of PCR employed an outer primer binding to the 5\' RACE adapter (5\'GCTGATGGCGATGAATGAACACTG 3\') and an outer RAMP-1 gene specific primer (5\' CGGGACCCTGACTATGGGAC 3\'). The second round of PCR employed an inner primer binding to the 5\' RACE adapter (5\'GCTGATGGCGATGAATGAACACTG 3\') and an inner RAMP-1 gene specific primer (5\' TCTTCATGGTCACTGCCTGC 3\'). Following the second round of PCR, PCR products were purified on an agarose gel and ligated into the PCR cloning vector pGEM-T Easy (Promega). A total of 8--10 clones representing each tissue were purified and subjected to DNA sequencing.
Generation of RAMP-1 promoter/luciferase expression constructs
--------------------------------------------------------------
To generate RAMP-1 promoter/luciferase deletion constructs, PCR utilising an antisense primer corresponding to sequences -26 to -43 (see Fig. [1](#F1){ref-type="fig"}) and sense primers corresponding to -145 to -164; -324 to -343; -470 to -450; -760 to -782; -935 to -948; -1220 to -1243; -1574 to -1593; -1827 to -1846; -2032 to -2093; -2342 to -2365; -2533 to -2511; -4102 to -4079 and -4724 to -4702 was carried out as described in Jagger et al. \[[@B25]\]. The PCR fragments were cloned into the vector pGEM-T Easy (Promega) and sequenced. The plasmids were then cut with Not 1 and the released RAMP-1 promoter fragment filled in with Klenow, purified and cloned into the luciferase expression vector pGL3 basic (Promega).
To generate a construct containing the RAMP-1 repressor element cloned at the 3\' end of the luciferase gene, the RAMP-1 promoter/luciferase construct containing 343 bp of 5\' flanking region was cut with Bam H1. A region between -343 and -782 of the mouse RAMP-1 promoter was generated by PCR as above and cloned into the Bam H1 cut -343 RAMP-1 promoter/luciferase construct in two orientations.
Tissue culture and RAMP-1/luciferase transfections
--------------------------------------------------
C2C12, HEK293 and NIH 3T3 cells were maintained in DMEM supplemented with 10% foetal calf serum, 100 μg/ml penicillin and 100 U/ml of streptomycin. To generate fused C2C12 myotubes, cells were allowed to become confluent and growth media was replaced with media containing only 1% foetal calf serum. For transient transfections in all the above cell lines, cells were transfected with 1.0--1.5 μg of specific RAMP-1/luciferase construct, 0.2 μg of a co-transfection control plasmid pRLTK (Promega) and Fugene 6 (Roche) according to manufacturers\' instructions. Luciferase assays were performed employing a Dual Luciferase Reporter Assay System (Promega) and a Turner Design TD-20/20 luminometer.
Semi-quantitative RT-PCR on RAMP-1 mRNA in differentiating C2C12 myoblasts
--------------------------------------------------------------------------
Total RNA from non-confluent, confluent and fused C2C12 myoblasts was isolated as above and reverse transcriptase (RT) reactions performed as previously described \[[@B25]\]. PCR amplification employing mouse RAMP-1 sense (5\' CTATGGGGAGCTCACTTACTGC 3\') and antisense (5\'GCAGTCTTCCTTGGAGTTCAGAAT 3\') oligos; mouse actin sense (5\'GTTAACCAACTGGGACGACATGG 3\') and anti-sense (5\'GATCTTGATCTTCATGGTGC 3\') mouse actin primers was performed as previously described \[[@B26]\]. A total of 30, 35 and 40 cycles were employed. The PCR products were run on a 1% agarose gel and Southern blotted to nylon membranes. The blots were probed with an end-labelled ^32^P actin probe (5\' CACACTGTGCCCATCTACGA3\') or RAMP-1 probe (5\'GCCCAATCCGGAAGTGGACA 3\') followed by autoradiography and densitomitry.
Abbreviations
=============
CGRP, calcitonin gene related peptide; CTR, calcitonin receptor; RAMP, receptor activity modifying protein; GPCR, G protein coupled receptor; CL, calcitonin receptor like
Authors\' contributions
=======================
RM performed the 5\' RACE analysis and semi-quantitative RT-PCR. All other experiments were conducted by MP.
Acknowledgements
================
This work was supported by a grant from the Wellcome Trust.
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
DNA sequence of mouse RAMP-1 5\' flanking region. Potential transcription factor binding sites have been underlined. Underlined ATG sequence is start site of translation = 0. Transcriptional initiation sites mapped by 5\' RACE analysis are indicated by \*asterisks.
:::
:::
::: {#F2 .fig}
Figure 2
::: {.caption}
######
Semi-quantitative RT-PCR on RAMP-1 gene expression in C2C12 cells. Total RNA was isolated from C2C12 non-confluent cells (**nc**), confluent (**c**) and myotubes (**mt**). The RNA was subjected to reverse transcription followed by PCR with mouse RAMP-1 and actin specific primers. The PCR reactions were stopped after 30, 35, 40 cycles and size fractionated on a 1% agarose gel. The DNA was then Southern blotted overnight and hybridised with a ^32^P end-labelled RAMP-1 and actin specific probes followed by autoradiography. This experiment is representative of three experiments each giving similar results.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
RAMP-1 promoter/luciferase activity in RAMP-1 positive and negative cell lines. RAMP-1 promoter-deletion constructs were generated by PCR and ligated into the luciferase expression vector pGL3-basic. The amount of 5\' flanking region relative to the start site of translation (ATG = 0) is indicated for each construct. Constructs were transiently transfected into RAMP-1 positive expressing cells (C2C12 and NIH3T3) and RAMP-1 negative cells (HEK293). Forty-eight hours after transfection, cells were harvested and assayed for luciferase activity. The co-transfection control plasmid pRL-TK was employed for all transfections. Relative luciferase activities were determined by comparing luciferase activity of cells transfected with RAMP-1/luciferase deletion constructs to cells transfected with the promoterless pGL3-basic plasmid. Background levels of luciferase = 1. The data are an average of three independent sets of transfections each giving similar results.
:::
:::
::: {#F4 .fig}
Figure 4
::: {.caption}
######
Silencer activity of RAMP-1 repressor element in transfected HEK293 cells. A RAMP-1 repressor element fragment (-783 to -343) was generated by PCR and ligated directly 3\' to the luciferase gene in the RAMP-1 expression plasmid -343 RAMP-1/luciferase. The fragment was oriented in two different positions (5\' to 3\' and 3\' to 5\'). Relative luciferase activity was compared between these constructs, pGL3-basic, a plasmid containing 470 and 4.7 kb of RAMP-1 5\' flanking region. Three independent sets of transfections were performed each giving similar results
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
RAMP-1 5\' RACE analysis. Total RNA was extracted from mouse brain, skeletal muscle and heart. RNA was treated with alkaline phosphatase, tobacco acid pyrophosphatase and ligated to a small adapter. Following reverse transcription, multiple rounds of PCR were performed and the PCR products cloned and sequenced. Transcriptional initiation sites are indicated relative to the start site of translation (ATG = 0).
:::
Mouse RAMP-1 transcriptional start sites
------------------------------------------ ------- -------- -------
Clone Brain Muscle Heart
1 -41 -50 -41
2 -41 -50 -41
3 -41 -40 -41
4 -9 -40 -9
5 -9 -50 -35
6 -11 -35 -35
7 -50 -9 -9
8 -41 -11 -50
9 -50 \- -41
10 -35 \- \-
:::
|
PubMed Central
|
2024-06-05T03:55:55.310514
|
2005-3-24
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079840/",
"journal": "BMC Mol Biol. 2005 Mar 24; 6:7",
"authors": [
{
"first": "Marc D",
"last": "Pondel"
},
{
"first": "Richard",
"last": "Mould"
}
]
}
|
PMC1079841
|
Background
==========
Non-syndromic hearing loss (NSHL) accounts for approximately 80% of cases of hereditary deafness and variety of genes are involved in NSHL \[[@B1]\]. Mutations in the *GJB2*gene, encoding the connexin 26 gap-junction protein, account for a significant proportion of NSHL \[[@B2]\]. To date, the substantial contribution of several nuclear genes \[[@B3],[@B4]\] and the pathogenic mitochondrial mutations in NSHL was established in some populations \[[@B5],[@B6]\]. Different population distributions of more than 70 *GJB2*mutations have been described and prevalence of some *GJB2*mutations is known to depend on population ethnic origin \[[@B2],[@B7]\]. The high frequencies of the 35delG mutation in the Caucasians, the 167delT in Ashkenazi Jews, and the 235delC mutation among east Asian populations have been shown to be the results of founder effects \[[@B8]-[@B12]\]. The underlying mechanisms accounting for the high prevalence of certain founder *GJB2*alleles are unknown. The data concerning the molecular basis of deafness in the populations of Russia are scarce and mostly limited to the European part of Russia and to the screening of the sole mutation 35delG \[[@B13]-[@B16]\]. Such data are yet unknown for Siberia (that is an expansive Asian part of Russia) populated by different ethnic groups.
The investigation presented here was performed on individuals with HI and subjects with normal hearing who are living in the Republic of Altai, situated in South Siberia and bordering Mongolia, China and Kazakhstan. Because of location, the Altai have been a particular region of great evolutionary importance. The archaeological, anthropological, and recent genetic data, mainly based on Y-chromosome and mitochondrial DNA studies \[[@B17],[@B18]\] suggest that for thousand of years the Altai-Sayan Mountains with surrounding territories have been a zone of the first contacts between paleo-Caucasoids and Mongoloids; repeatedly this region was a peculiar \"hybrid\" area closely connected with Central Asia, and a source of subsequent human migrations.
Indigenous inhabitants of the Altai Republic are the Altaians originated from several ancient Turkic-speaking tribes \[[@B19]\]. The Altaians mainly belong to the Central Asian type of the North Asian race. The Altai Republic population accounts for about 200,000 including Altaians, Russians, Kazakhs and other nationalities. The total number of Altaians is about 60,000. The rural population constitutes 3/4 of the Altai Republic people whereas approximately 50,000 persons live in the city of Gorno-Altaisk. The Altaians currently living in small settlements in the distant regions of the Altai Republic have mainly retained their native language, ethnic identity, and traditional marriage structure (patrilineal clan exogamy) \[[@B20],[@B21]\]. On the contrary, the population of the town Gorno-Altaisk is characterized by higher rates of interethnic marriages among Altaians, Russians and other ethnicities. As a whole, the Altaian population was shown to be genetic heterogeneous resulted from its territorial subdivision and the specific Altaian clan compositions in different localities \[[@B20]-[@B22]\]. Recently, the first epidemiological data in Altai Republic revealed the spectrum of hereditary pathology in the Altai Republic population with different prevalence rates among the urban and the rural populations, and among main ethnic groups of population (Russians, Altaians, Kazakhs) \[[@B23]\].
We report here the results of a study based on patients presenting mainly prelingual deafness, with the mutation analysis of the *GJB2*gene, the screening of the *GJB6*-D13S1830 deletion, and the study of five mitochondrial mutations involved in NSHI. Moreover, the *GJB2*screening was performed in 130 unrelated control subjects of Altaian ethnicity.
Methods
=======
Patients
--------
This study was approved by the Altai Republic Ministry of Health. Data on individuals with hearing impairment were obtained from the Republican medical institutions, Association of Deaf People, Specialized School for Deaf Children of the Republic of Altai (the town Gorno-Altaisk), and local settlements on the territory of the Republic. We ascertained 76 patients with HI (37 males and 39 females) of different ethnic affiliation (Altaians, n = 40; Russians, n = 17; Kazakhs, n = 3; mixed or other ethnicity, n = 16) representing a total of 51 unrelated families. Their ages ranged from 3 to 80 years old (mean 30.2). Based on family histories, 19 patients were defined as sporadic cases, while 57 patients were issued from 32 unrelated families, presumably showing an autosomal recessive (20 families), autosomal dominant or maternal (8 families), and ambiguous (4 families) mode of inheritance. In addition to the patients, 58 DNA samples of family members with normal hearing were included. Informed consent was obtained from all adult participants and from parents of the children studied.
Bilateral (mainly symmetrical) and prelingual/early onset HL was defined in most of the patients. Among the 76 patients 44 could be otoscopically examined, and pure tone audiometry was performed in the special medical service (of the town Gorno-Altaisk). As the other patients were mostly living in the small settlements distant from the town Gorno-Altaisk, the data was collected from the local unspecialized medical services and by direct interview with the patients and their relatives. Among the clinically documented patients (n = 44), HL was severe to profound in 32 patients, moderate in 10, and mild in 2 patients. Sensorineural type of HI was precisely defined for 38 out of 76 patients, and the remaining patients presented mixed (conductive-sensorineural, n = 6) or uncertain type of HL. It should be noted that the chronic otitis media, widespread in Altaian population, was frequently registered in the medical histories of these patients. We have analyzed the cohort of patients with HI without the possibility of excluding cases of environmental exposures (infections, prenatal or postnatal ototoxicity etc) or unknown etiology.
Controls
--------
130 unrelated subjects of Altaian origin with no familial history of hearing problems were used as controls. Most of DNA samples were obtained from the Southern Altaian population of the Ust\'-Kan administrative region of the Altai Republic and the others from Northern Altaians living in the Turochak administrative region. All DNA samples studied were anonymized.
Mutation analysis
-----------------
DNA was extracted from peripheral blood using standard protocols. *GJB2*analysis was performed for the non-coding exon E1 and coding exon E2 and their flanking sequences by PCR-DHPLC as previously described \[[@B24]\]. Any abnormal profile observed by DHPLC was consequently analyzed by sequencing (ABI 310 sequencer, Applied Biosystems). The patients without any mutations in the coding exon E2 were analyzed in the non-coding exon E1 and tested for the mutation Δ(*GJB6*-D13S1830) that includes the deletion of most of the *GJB6*gene \[[@B24]\]. The screening of five known mitochondrial mutations (m.A1555G, m.7445A\>G, m.7472insC, m.7510T\>C, m.7511T\>C) was performed in the *GJB2*negative patients that were compatible with a maternal inheritance of HI. The A1555G mutation was analyzed as previously described \[[@B5]\] whereas the four others were analyzed by direct sequencing (position 7321--7620 of the mtDNA).
Statistical analysis
--------------------
Differences between groups were analyzed by chi square statistics. *P*-values were considered to be significant when \<0.05.
Results
=======
*GJB2*gene
----------
*GJB2*mutations were detected in 23.7% of the patients (18 out of 76) (Table [1](#T1){ref-type="table"}). We identified five different mutations, all of them lying in the coding exon: three known as recessive mutations (35delG, 235delC and 312del14), one as a dominant mutation (R75Q), the fifth mutation is a newly identified missense mutation, W172C. All the patients with *GJB2*mutations were familial cases presenting prelingual profound HL except one patient, genotyped 35delG/235delC, who had moderate-severe HL. Mutations were present as homozygous or compound heterozygous state, with the exception of the R75Q that was identified as single in two members of family 11. Among all the families studied, family 11 deserves particular interest because dominant and pseudo-dominant deafness is observed with a complex segregation of different mutations, as shown on Fig. [1](#F1){ref-type="fig"} (see in Discussion).
Table [2](#T2){ref-type="table"} shows the distribution of *GJB2*sequence variations in the two predominant ethnic groups (Altaians and Russians) of patients as well as in Altaian controls. Of the 60 unrelated Altaian patient alleles, only 8.3% were carrying *GJB2*mutations with the 235delC representing 6.67% of the alleles and the W172C, 1.67%. In contrast, 46.9% (15 out of 32) of the unrelated Russian patient chromosomes carried *GJB2*mutations with the 35delG accounting for the most frequent mutation (34.4%). The other mutations 312del14, R75Q and 235delC were detected in 6.3%, 3.1% and 3.1% of the Russian alleles, respectively.
The polymorphism V27I and E114G were observed in Altaian patients as well as in Altaian controls (Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}). None of the Russian patients had both variants. Because the E114G variant was never identified as single in our study but also in others \[[@B25],[@B26]\] we assumed that the E114G and V27I variants are in *cis*-configuration. The estimated allele frequencies of V27I and \[V27I; E114G\] are respectively 16.7% and 11.7% among 60 unrelated Altaian patient chromosomes, versus 13.1% and 6.5% in the Altaian controls (260 chromosomes). Furthermore, three other known sequence variations were identified in Altaian controls: the 235delC mutation (2.3%), the F191L variation with unknown pathogenic consequences (0.4%), and the non-pathogenic V153I variant (0.4%) as shown on Table [2](#T2){ref-type="table"}. A new sequence variation, IVS1+27G\>C, was identified in one Altaian patient with severe-profound HL who had genotype IVS1+27G\>C / \[V27I; E114G\] (Table [1](#T1){ref-type="table"}) and in his non-affected mother (genotyped IVS1+27G\>C/+). The IVS1+27G\>C variation was not previously described, but splice site prediction programs would suggest its non-pathogenic consequences \[[@B27]\]. Finally, the 765C\>T polymorphism (referred to as SNP1 \[[@B9]\]) located in the 3\'UTR of the *GJB2*gene was analyzed in patients. All alleles carrying the 35delG mutation were associated with the 765T variant; that is in accordance with recent data on the French patients \[[@B24]\].
Δ(*GJB6*-D13S1830) and mitochondrial mutations
----------------------------------------------
Screening of the Δ(*GJB6*-D13S1830) and mitochondrial mutations m.1555A\>G, m.7445A\>G, m.7472insC, m.7510T\>C, m.7511T\>C was negative in patients who had no *GJB2*mutations.
Discussion
==========
Specific *GJB2*mutation spectrums and different *GJB2*contributions for deafness were observed for the two main ethnic groups studied here.
Prevalence of *GJB2*mutations in Russian patients
-------------------------------------------------
*GJB2*mutations are responsible for deafness in 54% (7 out of 13) of the independent Russian families tested and the 35delG mutation was carried by 34.4% of the Russian alleles. These findings correlate with earlier reported data concerning Russian subjects with HI \[[@B13]-[@B16]\] and confirm the main contribution of the 35delG mutation in deafness among Caucasian populations \[[@B28],[@B29]\].
*GJB2*mutations with dominant and pseudo-dominant transmission
--------------------------------------------------------------
We studied the segregation of different *GJB2*mutations in a three-generation family (family 11; Fig. [1](#F1){ref-type="fig"}). All affected members presented prelingual profound HL and were using sign language to communicate. The dominant mutation R75Q (224G\>A) was found in five members, as the sole mutation or in *trans*of another mutation (235delC or 312del14). R75Q was previously identified as a non-syndromic dominant mutation \[[@B7]\] and as a syndromic mutation \[[@B30]\] in a four-generation Turkish family with autosomal dominant inherited HI and congenital diffuse palmoplantar keratoderma. The age of onset and progression of HL were variable among the affected members of this family \[[@B30]\]. It should be noted that a previously described dominant mutation located at the same codon 75 (R75W) was found to have a variable penetrance regarding skin disease symptoms (even to their absence) \[[@B31],[@B32]\]. Preliminary dermatological examination performed in all individuals of family 11 carrying the R75Q mutation did not reveal any skin disease in patients I-2, II-2 and II-4 though mild manifestation of probable keratosis was detected in patients III-2 and III-3. However, additional investigations need to be performed for elucidation of non-syndromic and/or syndromic status of the R75Q mutation as other factors can modulate the expression of a particular phenotype. The new missense mutation W172C (516G\>C) was identified in *trans*of the 235delC allele in the Altaian patient I-1. This patient had both parents and a brother with normal hearing but an affected sister (not shown). A G\>A substitution at position 516, generating a stop codon at position 172 (W172X) had been previously described \[[@B7]\]; moreover, the tryptophan at position 172 is conserved in many species (human, mouse, rat, cow, sheep). These data tend to imply that this *GJB2*sequence variation is a severe recessive mutation although *de novo*appearance of the W172C mutation cannot be excluded in this case.
*GJB2*contribution in Altaian patients and controls
---------------------------------------------------
This study has revealed an Asian-specific diversity and the prevalence of *GJB2*sequence variations in the Altaians. High rates of the V27I and \[V27I; E114G\] alleles were detected in Altaian patients and Altaian controls with no significant differences between both groups (p \> 0.05). These results, once more, confirmed that both sequence variations correspond to non-pathogenic polymorphisms and (that their high rates) represent a distinctive feature of particular Asian populations \[[@B10]-[@B12],[@B25],[@B33]-[@B36]\]. *GJB2*gene contribution in deafness among Altaians (8.3%) is mainly defined by the recessive mutation 235delC (6.67%). This mutation is specific to some east Asian populations (Japan, China, Korea, Taiwan) with an allelic frequency up to 20% among deaf patients and a carrier frequency of 0% to 2.8% in normal populations and it represents the most prevalent *GBJ2*pathogenic mutation as it accounts for up to 80% of the pathogenic alleles in patients \[[@B10]-[@B12],[@B33]-[@B37]\].
The 235delC mutation, found in 4/60 unrelated chromosomes in Altaian patients, was also detected in 6/130 Altaian controls resulting in a carrier frequency of 4.6%, higher than in any other Asian populations. This high frequency could be attributed to the gene drift/founder effects in a certain subpopulation. The Altaian control group studied here consists mainly of the inhabitants of one administrative region of the Altai Republic. Earlier, it was shown that the marriage migrations among this Altaian subpopulation were, in their majority, restricted to the territory of this region \[[@B21]\].
Comparatively low *GJB2*contribution among the Altaian patients (8.3%) could be explained by the fact that a few patients presented acquired hearing loss. Moreover, a specific marriage structure (patrilineal clan exogamy) practiced by the Altaians for a long time to avoid close inbreeding, could influence apparent scarcity of affected *GJB2*homozygotes by restriction of random marriages of Altaians belonged to certain clans. Finally, some other deafness-causing genes, not considered in this study, could be involved in deafness among Altaians. The additional study of the *GJB2*gene diversity in other Altaian groups will be informative for verification of the 235delC prevalence pattern among the Altaians.
Recent studies had established that the 235delC mutation among east Asian populations was derived from a common ancestral founder \[[@B10]-[@B12]\]. Yan et al. \[[@B12]\] roughly estimated the age of the 235delC mutation to be about 11500 years old and speculated that the 235delC mutation might have arisen in the Lake Baikal area and then spread to Mongolia, China, and Japan through subsequent migrations. Taking into account that the territories of South Siberia, including the Altai region and the Lake Baikal area were melting-pots and dispersion centers for world-wide human expansions in the past, we suggest that the Altai region could be one of the founder sources for the 235delC mutation widespread in Asia. To elucidate this hypothesis, the 235delC prevalence will be tested among indigenous people living in other regions of Altai and surrounding territories and SNP analyses will be performed to verify the common origin of the 235delC alleles.
Conclusion
==========
In this study we found an Asian-specific *GJB2*diversity among the Altaians and different *GJB2*contribution for deafness in the Altaian and Russian patients. Additional investigations remain necessary to elucidate all genetic origins of HL in Altaian patients. The high carrier frequency of the 235delC mutation in normal hearing Altaians (4.6%) is probably defined by gene drift/founder effect in a particular Altaian group. The question whether the Altai region could be one of the founder sources for the 235delC mutation widespread in Asia is open. Finally, we presented new evidence that the CX26 R75Q mutation, earlier described as syndromic mutation (autosomal dominant inherited hearing impairment with congenital diffuse palmoplantar keratoderma), could be associated with a high variable expression of the skin symptoms.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
VT and OP collected the family data and recruited patients and controls for the study; LP provided a part of control blood samples; OP and NPR carried out the molecular studies; AFR supervised the whole study in the laboratory of MC. AFR and OP wrote the manuscript. All authors read and approved the final version of the manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2350/6/12/prepub>
Acknowledgements
================
The authors wish to thank the families who participated in this work. They also thank audiologists Fionov G.S. and Poposheva G.T. who assisted with this study. To achieve this work, OP has been supported as a \"Directeur de Recherche Associé\", CNRS, DADR/01-IM657. This work was supported by the \'Centre Hospitalier et Universitaire\' of Montpellier (France).
Figures and Tables
==================
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Pedigree of family 11 presented with the *GJB2*genotypes.**Analyses were performed for the numbered subjects. \'\'+\'\' denotes that no *GJB2*mutations were detected; \'\'nt\'\' -- not tested. Ethnic affiliation of the parents: I-1 was Altaian; I-2, I-3, father of II-1 and father of III-1 were Russians; II-3 was of mixed ethnicity (Russian/Tatar).
:::
:::
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
*GJB2*sequence variations identified in patients
:::
Genotype Affected subjects (n = 76) Ethnic affiliation
------------------------------- ------------------------------ ------------------------------------
Mutations
35delG / 35delG 7 (4 independent families) Russians
235delC / 235delC 1 Altaian
312del14 / 312del14 1 Russian/Tatar
35delG / 235delC 2 (unrelated) 1 -- Russian/Altaian, 1 -- Russian
35delG / 312del14 1 Russian
R75Q^a^/ 235delC 1 Altaian/Russian
R75Q / 312del14 2 (two sibs) mixed ethnicity
R75Q/ + 2 (father and daughter) Russians
235delC / W172C 1 Altaian
Total 18 (23.7%)
Non-pathogenic variants
V27I / + 14 (11 independent families) 13 Altaians, 1 Kazakh
(V27I; E114G) ^b^ 7 (unrelated) 6 Altaians, 1 Altaian/Russian ^c^
IVS1+27G\>C / \[V27I; E114G\] 1 Altaian
Total 22 (28.9%)
^a^dominant mutation; ^b^*cis*-configuration of V27I and E114G was confirmed by family studies in 5/7 patients, ^c^the \[V27I; E114G\] allele of this proband is known to be of Altaian origin (maternal branch).
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Frequency of the *GJB2*sequence variations among patients and Altaian controls
:::
-----------------------------------------------------------------------------------------------------------------------------------
Name Nucleotide change Number of unrelated chromosomes tested Frequency (95% CI)
----------------- ------------------- ----------------------------------------------------------- ---------------- ----------------
235delC \- 4\ 6\ 1\
0.067\ 0.023\ 0.031\
(0.019--0.144) (0.009--0.045) (0--0.118)
W172C 516G\>C 1\ 0\ 0\
0.017\ 0\ 0\
(0--0.065) (0--0.004) (0--0.010)
35delG \- 0\ 0\ 11\
0\ 0\ 0.344\
(0--0.017) (0--0.004) (0.192--0.515)
312del14 \- 0\ 0\ 2\
0\ 0\ 0.063\
(0--0.017) (0--0.004) (0.007--0.172)
R75Q 224G\>A 0\ 0\ 1\
0\ 0\ 0.031\
(0--0.017) (0--0.004) (0--0.118)
IVS1+27G\>C 1\ \- 0\
0.017\*\ 0\*\*\
(0--0.066) (0--0.059)
V27I 79G\>A 10\ 34\ 0\
0.167\ 0.131\ 0\
(0.085--0.271) (0.093--0.175) (0--0.010)
\[V27I; E114G\] 79G\>A+341A\>G 7\ 17\ 0\
0.117\ 0.065\ 0\
(0.049--0.210) (0.039--0.099) (0--0.010)
F191L 571T\>C 0\ 1\ 0\
0\ 0.004\ 0\
(0--0.017) (0--0.016) (0--0.010)
V153I 457G \>A 0\ 1\ 0\
0\ 0.004\ 0\
(0--0.017) (0--0.016) (0--0.010)
-----------------------------------------------------------------------------------------------------------------------------------
^a^the total number of unrelated Altaian chromosomes (n = 60) is the sum of twenty-six independent Altaian patients (52 chromosomes) and eight independent patients of mixed (Altaian/other) ethnicity (8 chromosomes); ^b^the total number of unrelated Russian chromosomes (n = 32) is estimated from thirteen independent Russian patients (26 chromosomes) plus six chromosomes corresponding to six independent patients of mixed (Russian/other) ethnicity; \* 58 chromosomes were tested for the *GJB2*non-coding exon; \*\* 16 chromosomes were tested for the *GJB2*non-coding exon.
:::
|
PubMed Central
|
2024-06-05T03:55:55.312691
|
2005-3-24
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079841/",
"journal": "BMC Med Genet. 2005 Mar 24; 6:12",
"authors": [
{
"first": "Olga",
"last": "Posukh"
},
{
"first": "Nathalie",
"last": "Pallares-Ruiz"
},
{
"first": "Vera",
"last": "Tadinova"
},
{
"first": "Ludmila",
"last": "Osipova"
},
{
"first": "Mireille",
"last": "Claustres"
},
{
"first": "Anne-Françoise",
"last": "Roux"
}
]
}
|
PMC1079842
|
Background
==========
Temporal lobe epilepsy (TLE) preceded by febrile seizures (FS) was historically considered to be an acquired disorder. However, recent studies have confirmed the existence of distinct genetic syndromes encompassing both FS and TLE \[[@B1],[@B2]\]. Additionally, a positive history of FS in families with temporal lobe epilepsy is more common than expected by chance; this suggests that FS may be an initial manifestation of some temporal lobe epileptic syndromes \[[@B3]-[@B6]\]. The genetic etiology of TLE preceded by FS remains unknown. Two loci for this condition have been mapped, however, the causative genes have not been identified \[[@B1],[@B2]\].
γ-aminobutyric acid (GABA) is the main inhibitory neurotransmitter system in the brain. Mutations in the GABA-A receptor subunit genes (GABR~A~), encoding products that form a heteropentameric chloride channel, have been identified as rare causes of idiopathic generalized epilepsies showing a Mendelian inheritance pattern \[[@B7]\]. However, mutations in the GABR~A~genes have thus far not appeared to be a major cause of TLE or other focal epilepsies \[[@B8]\].
The second, or B-subtype of GABA receptors (GABR~B~) function by acting through G-protein coupled signaling systems. Alteration in this GABA system during epileptogenesis is suggested from experimental data \[[@B9],[@B10]\]. Furthermore, a previously described variant (G1465A) in the *GABBR1*locus, encoding the GABR~B~β1 subunit has been reported in association with TLE, including TLE preceded by FS \[[@B11]\]. Genetic association studies are fraught with many difficulties and independent replication is important to establish the validity of any reported association \[[@B12]\]. We report genetic analysis of *GABBR1*gene by thorough screening of *GABBR1*exons for disease-causing mutations, and attempted replication of association at the G1465A polymorphism in a cohort of our patients exhibiting TLE preceded by FS.
Methods
=======
Subjects
--------
This study included 120 unrelated patients carrying a diagnosis of TLE preceded by FS. Informed consent, approved by the Institutional Review Board was obtained from every subject enrolled. A TLE diagnosis was made based on a history obtained from the patient and witnesses, with particular attention to descriptions of aura and complex partial seizure semiology, suggesting temporal lobe origin. Diagnoses were supported by electroencephalograph (EEG) recordings and MRI neuroimaging; some cases involved EEG-video recordings. MRI examination included 3D gradient echo imaging (Spoiled Gradient Recalled Acquisition) performed in the coronal plane with the slice thickness 1 mm, FLAIR and T2-weighted coronal sections; all coronal sections were perpendicular to the long axis of the hippocampus. Additional MRI sequences included T2-weighted axial and T1-weighted sagittal brain images. Only subjects with a reliable clinical classification were included in these studies.
Data collection involved age and duration at first FS, presence of any focal ictal or postictal features, total number of FS, and age at FS remission. Information was obtained from parents and/or older siblings. FS were classified as complex if they lasted longer than 15 minutes, had focal ictal or postictal features, or occurred more than once during the same day. Detailed family histories were obtained from all patients, including relatives with a history of epilepsy, FS, or both. Information was obtained from all first-degree relatives (mother, father, brothers, sisters, and children); information regarding second-degree relatives involved questions related to grandparents, grandchildren, nieces, nephews, aunts and uncles.
Genetic analysis
----------------
DNA was isolated from peripheral blood from all subjects and relevant affected and unaffected family members enrolled in this study using standard methods. The current study focused solely on affected probands. The *GABBR1*G1465A polymorphism was analyzed as previously described \[[@B13],[@B14]\]. The PCR product was digested overnight with *EagI*restriction enzyme following the manufacturer\'s protocol (New England Biolabs; Beverly, MA). Products were analyzed by electrophoresis on 2% agarose gels and subsequent visualization of resolved fragments.
PCR primers flanking all exons were designed from published sequence and public genomic assemblies covering the locus (primer sequences are available upon request) \[[@B15]\]. Exon screening was performed using single strand confirmation polymorphism (SSCP) analysis; abnormal mobility shifts were sequenced.
All polymorphisms identified, including G1465A, were analyzed in two distinct ethnically- (Caucasians) and gender-matched control samples. The first comparison group consisted of DNA samples from 118 unrelated, healthy subjects obtained from the Vanderbilt Center for Human Genetics Research DNA Resource Core. The individuals from this group were absent any family history of epilepsy. The second control group corresponded to the Caucasian panel of 100 DNA samples from healthy individuals purchased from the Coriell Cell Repositories \[[@B16]\].
Statistical analysis
--------------------
Allele frequencies comparisons between patient and control groups was determined using a chi-square test. Mann-Whitney U test was used to compare demographic and clinical characteristics between patients with epilepsy and control subjects.
Results
=======
The 120 case subjects enrolled in this study included 49 men and 71 women (average age 34.8 ± 14.9 years) with a positive history of TLE, in which FS preceded the onset of epilepsy. Average age of TLE onset was 22.81 ± 10.08 years, with an average syndromic duration of 12.65 ± 8.12 years. Positive family history of epilepsy was reported in 66 individuals, while 54 were apparently sporadic cases. These two subgroups did not significantly differ in their demographic composition; however, familial cases had a significantly earlier age of epilepsy onset compared to sporadic cases (13.64 ± 8.23 and 27.44 ± 11.78, p \< 0.01). All familial cases had at least one TLE affected relative in either previous or successive generations, consistent with vertical transmission of this trait. Medically refractory epilepsy was present in 23 patients (19%).
Genotype analysis of the G1465A variant detected a single A allele in one subject with TLE, who exhibited poor seizure control in spite of optimal clinical management. This patient had a positive family history of TLE and FS on one parent\'s side of the family. Genotype analysis of both parents indicated the A allele was present in the unaffected parent from the other side of the family, and thus did not segregate with disease. Two alleles of the same variant were detected in the control group in one control subject who was homozygous for this polymorphism. Allele frequencies for case and control groups corresponded to 0.45% or 0.87%, respectively; p = 0.988). Genotype accuracy for G1465A was confirmed by combined sequence and SSCP analyses that showed distinct and unambiguous mobility shift patterns for both heterozygous and homozygous states.
SSCP analysis of all coding exons did not reveal any apparent disease-causing mutations in *GABBR1*. We did, however, identify one previously described nonsynonymous variant (C59T) in exon 1, resulting in an alanine to valine substitution in one patient. This known single nucleotide polymorphism (SNP) is present in 1--3% of normal subjects, and it has been reported in a homozygous state in one affected individual \[[@B13]\]. Given the known normative frequency of the C59T SNP (rs\#1805056), control samples were not screened. Several SNPs were detected in exon 11, which also contains the G1465A, variant, previously reported to be in exon 7 \[[@B11]\]. We detected three previously reported synonymous (silent) sequence changes -- T1473C (rs\#N/A), T1476C (rs\#N/A), and T1485C (rs\#29225) \[[@B14]\]. They appeared to be present on the same haplotype, based on segregation analysis of carrier relatives with this allele and constant SSCP mobility shift patterns. The frequency of this three-SNPs haplotype was determined to be 10.83% (26/240, including one homozygous individual); there was no difference between sporadic and familial cases. Analysis of controls (236 chromosomes) initially detected an allele infrequency of 4.23%, suggesting a significant trend (p \< 0.05). Subsequent analysis of the second control sample identified a frequency very similar to the TLE patient sample (18/200, 9%, p = 0.63). Finally, we identified several novel intronic SNPs not predicted to alter splicing, and their frequency did not differ significantly between the TLE and control samples (data not shown).
Discussion
==========
The 1465A allele at *GABBR1*was proposed as a potential risk factor for TLE, particularly in patients with medically refractory seizures \[[@B11]\]. We were unable to replicate this result, and no association of this allele with TLE was detected in our cohort of patients. Furthermore, systematic screening of all *GABBR1*exons did not identify any obvious disease-related variants or novel SNPs or haplotypes significantly associated with the development of TLE.
The conflicting results between our study and that reported by Gambardella et al, \[[@B11]\] are unlikely due to clinical differences in the subject population or insufficient power. The sample sizes were similar (120 in our study and 141 in the original report \[[@B11]\]) in both studies. Our cohort had a history of antecedent FS, due to our ongoing effort to characterize this TLE subtype. The samples analyzed by Gambardella et al.\[[@B11]\] were mixed in this regard, and included patients with and without FS. Only 14.8% of their patients had a positive history of febrile convulsions, but the presence or absence of an FS history did not correlate with carrier status of the 1465A allele (16.7% with vs. 14.5% without 1465A) \[[@B11]\]. Our patient sample was more heavily loaded towards those with positive family history (55%) when compared with the study by Gambardella and colleagues \[[@B11]\] whose patient samples comprised of 36.2% of patients with a family history of epilepsy.
The control 1465A allele frequency in the current study was somewhat higher than in the previous study (0.87% vs. 0.26%, respectively), however, the difference between these two populations is not significant. This is consistent with other studies that found \<1% frequency for the 1457A variant with an exception of one report, which identified 10% frequency of this allele in a cohort of 50 patients \[[@B11],[@B13],[@B14]\]. Our results support the conclusion that G1465A is a rare polymorphism. We also verified the presence of G1465A polymorphism by SSCP and sequencing and we had a complete agreement between different methods, arguing against incorrect genotyping as an explanation of our negative finding.
A systemic screening of all exons at the *GABBR1*locus identified alleles at three known SNPs in exon 11 that also contains the 1465A variant, thus defining a unique haplotype \[[@B14]\]. These sequence changes are silent without any amino acid change in the protein. The comparison of the haplotypic frequency with our first control group suggested a potential increase, although a plausible functionality is currently unclear. The nonreplication in the second control group leaves the initial trend in doubt, and compels a conservative interpretation.
Spurious association that cannot be subsequently replicated can be caused by several factors \[[@B17]\]. Other than an inadequate sample size, selection bias is probably the most likely explanation of discrepancies between association studies. Unrecognized population stratification has been detected even in studies with otherwise proper design \[[@B18]\]. The problem of population stratification despite a careful selection is also demonstrated by the differences between the frequencies of the 1465A variant between our two control groups. It has been suggested that two control groups should be used in association studies, one population-based and one family based \[[@B19]\]. We did not have a family-based control group available but our conflicting data between two independent population-based control groups suggest that this approach is also useful in confirmation of true genetic association.
Conclusion
==========
Our data suggest that variants in *GABBR1*are not significantly involved in the pathogenesis of TLE preceded by FS. The proportion of TLE patients with FS in the original study, which reported this positive association did not differ between allele A negative and positive cases, and thus our failure to reproduce this result is likely applicable to all non-lesional TLE epilepsies.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
SM carried out the molecular genetic studies, participated in the results interpretation, and helped to draft the manuscript. BA-K participated in the design of the study and its coordination, and helped to draft the manuscript. JSS participated in the design of the study and helped to draft the manuscript. JLH participated in the design of the study and helped to draft the manuscript. PH conceived of the study, participated in its design and coordination, performed statistical analysis, and drafted the manuscript. All authors read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2350/6/13/prepub>
Acknowledgements
================
We thank the members of the families described here, whose help and participation made this work possible. This work was supported by NIH grant K08NS42743 to PH. Patients\' recruitment and DNA collection was done in part through the core facilities of the Vanderbilt Center for Human Genetics Research and the VUMC General Clinical Research Center, supported in part by GCRC Grant RR00095.
|
PubMed Central
|
2024-06-05T03:55:55.315212
|
2005-3-30
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079842/",
"journal": "BMC Med Genet. 2005 Mar 30; 6:13",
"authors": [
{
"first": "Shaochun",
"last": "Ma"
},
{
"first": "Bassel",
"last": "Abou-Khalil"
},
{
"first": "James S",
"last": "Sutcliffe"
},
{
"first": "Jonathan L",
"last": "Haines"
},
{
"first": "Peter",
"last": "Hedera"
}
]
}
|
PMC1079843
|
Background
==========
Bipolar affective disorder (BPD) is a chronic, severe mood disorder characterized by recurrent episodes of mania and depression. It often interferes with the patient\'s ability to cope with his daily routine and has a high mortality rate from suicide. BPD is estimated to have a lifetime prevalence of 0.1% to 1% in Taiwanese Han \[[@B1]\]. The high prevalence, together with the high frequency of hospitalization, psychosocial impairment, suicide, and substance abuse has made the disease a major public health concern. Although previous studies in epidemiology strongly suggested genetic factors in the etiology of affective disorder \[[@B2]\], establishing the mode of transmission and searching for predisposing genes using linkage analyses have not been successful. Association studies have been useful in mapping genes for complex diseases and are now being applied to many psychiatric traits \[[@B3]\]. However, the success of this approach requires the markers of the candidate gene being used are either the causative variant(s) or in linkage disequilibrium with the causative variant(s) \[[@B4]\]. Results from these studies suggest the involvement of several genes, each of which has a minor effect on the pathogenesis of BPD \[[@B5]\].
Serotonin (also known as 5-hydroxytryptamine) is a major neurotransmitter in the central nervous system (CNS) and is involved in various physiological events, such as mood control, sleep, thermoregulation, learning, and memory \[[@B6]\]. Its role in psychiatric disorders is well documented \[[@B7]\]. Disruption of serotonergic function has been implicated in the pathogenesis of many psychiatric disorders, including BPD \[[@B8],[@B9]\], making genes involved in serotonin transmission and its metabolic pathway good candidates for involvement in BPD pathogenesis.
Tryptophan hydroxylase (TPH) catalyzes the biopterin-dependent monooxygenation of tryptophan to 5-hydroxytryptophan, which is subsequently decarboxylated to form the neurotransmitter, serotonin \[[@B10]\]. Two tryptophan hydroxylase isoforms have been identified \[[@B11]-[@B13]\] and the expression of TPH1 and TPH2 were found to be mainly in the pineal gland \[[@B14]\] as well as in the peripheral tissues (duodenum, kidney or lung; \[[@B15]\]) and in the brain stem \[[@B12],[@B15]\], respectively. The *TPH1*gene has been the subject of intensive study regarding its possible involvement in many psychiatric and behavioral traits \[[@B16]\]; And an increasing number of investigations have studied the *TPH2*gene effect in association with several psychiatric and behavioral traits including antidepressant response \[[@B17]\] and major depression \[[@B18]\].
The *TPH1*gene, which has been mapped to human chromosome 11p15.3-p14 \[[@B13]\], spans a region of 29 kilobases and consists of 11 exons \[[@B19]\]. Numerous mRNA species are transcribed from a single promoter by alternative splicing of three exons (1A, 1B, and 1C) in the 5\'- untranslated region (UTR) \[[@B19],[@B20]\] and two exons (11A and 11B) in the 3\'UTR \[[@B19],[@B20]\]. The structure of the gene and the sequence of the coding region are very similar to those for the genes for human tyrosine hydroxylase and phenylalanine hydroxylase, which belong to the super-family of aromatic amine acid hydroxylases. Polymorphisms within intron 7 of the *TPH1*gene have been implicated to be associated with several psychiatric conditions such as BPD \[[@B21]\], suicidal behavior and alcoholism \[[@B22]\], and aggression-related traits \[[@B23]\]. Furthermore, the *TPH1*intron 7 polymorphism was associated with the 5-HIAA level in both impulsive Finnish alcoholics \[[@B24]\] and healthy men \[[@B25]\] thus suggesting that *TPH1*genotypes may participate in the regulation of serotonin turnover rate in the central nervous system. However, such associations were not found in other studies involving suicidal behavior \[[@B26]\], BPD and mood disorders \[[@B27],[@B28]\], and attention-deficit hyperactivity disorder \[[@B29]\]. These studies indicate that the role of TPH1 in predisposing different psychiatric traits varies and could be modified by other factors like ethnicity or gene-to-gene interactions \[[@B30]\]. Preliminary study in our laboratory has found an allelic association (P = 0.03) between the *TPH1*intron 7 A218C polymorphism and BPD in 42 Taiwanese bipolar patients and 70 matched healthy controls \[[@B31]\]. Results from these studies suggest that *TPH1*may play a role in regulating serotonin metabolism. More recently, sertraline, a selective serotonin reuptake inhibitor and an effective antidepressant, has been shown to up-regulate *TPH1*expression and serotonin synthesis \[[@B32]\], suggesting one of the mechanism for its pharmacological action in antidepressive therapy.
To provide information of the *TPH1*gene for future genetic analysis of disorders involved the disturbances in the serotonergic system in Taiwan, we have systematically screened all 11 exons and the promoter region of the human *TPH1*gene for common sequence polymorphisms and have identified 5 single nucleotide polymorphisms (SNPs). In addition, we carried out a case-control study of 108 BPD type I patients and 103 normal controls looking for an association between BPD and these 5 SNPs plus 5 others taken from the literatures \[[@B33],[@B34]\] and a public SNP database \[[@B35]\], tested both as individual SNPs and as the haplotypes.
Methods
=======
Subjects
--------
Since 1998, probands were recruited from bipolar outpatients in the Chung Shan Medical University Hospital and the Taichung Rehabilitation Hospital in Taiwan. Controls were subjects from local volunteer blood donors who have no family or personal history of major affective disorder and other psychiatric disorders were matched to cases on the basis of ethnic or geographic origin, sex, and age. The gender ratios are identical in both case and control groups (male to female is 53% to 47 %) and the age of populations are 38 ± 13.7 and 26.4 ± 7.9 in case and control groups, respectively. The age of onset is 28.4 ± 12.7 in all patients and is 30.3 ± 14.5 and 26.4 ± 10.2 in male and female patients, respectively. Clinical interviews were conducted by experienced psychiatrists (leading by Dr. Lai) for all subjects after the study procedure had been fully explained and information on general demographic data, such as age, sex, and ethnicity, was obtained. Patients assessment was purely by direct clinical interview by the treating clinician according the procedure described in the DSM-IV diagnoses of lifetime Major depressive disorder and bipolar I. Additional information required to reach diagnosis was also collected from all clinical and hospital records where available but the comorbidity with other psychiatric/neurological disorders or medical problems were not considered as an inclusion/exclusion criteria. This study was approved by the University Ethics Committee and written informed consent was obtained from all participants. In total, 108 BPD and 103 controls (all Taiwanese Han) were recruited for this study.
DNA preparation
---------------
Approximately 10 ml of peripheral blood was collected from the recruits using EDTA anticoagulant venous blood tubes, and DNA prepared using a BIO-RAD InstaGene™ Whole Blood Kit (Bio-Rad Laboratory, Hercules, CA) according to the manufacturer\'s protocol.
Primer design for searching for, and typing of, TPH1 gene polymorphisms
-----------------------------------------------------------------------
All primers used were designed using the Primer3 program \[[@B36]\] or OLIGO 5 Primer Analysis Software (Molecular Biology Insights, Inc., cascade. CO., USA). The primers used to screen the *TPH1*gene (Table [1](#T1){ref-type="table"}) were designed using the published sequence (GenBank accession number: AC005728). The primers used in exon-wide SNP scanning were designed from the intronic sequence roughly 50 bp upstream and downstream of each exon to amplify the entire exon sequences. In addition, overlapping fragments covering from 1,151 bp upstream of the 5\' promoter region to 1,572 bp downstream of the 3\'UTR were also amplified to screen for polymorphisms within these regions. The forward and reverse primers for base excision sequence scanning (BESS) product amplification were labeled, respectively, with fluorescent 6-FAM and HEX to facilitate variant detection.
SNP identification in the coding and regulatory regions of the TPH1 gene
------------------------------------------------------------------------
In order to identify all SNPs in the coding and regulatory regions of the *TPH1*gene, we designed primers to generate PCR products for use in SNP identification using the BESS Base Reader Kit (Epicentre Technologies, Madison, WI) according to the manufacturer\'s protocol. Each identified SNP was sequenced to confirm the sequence variant. The screening panel included 50 unrelated subjects with or without BPD.
The BESS T & G Base Reader Kit, which identifies all types of point mutation, deletion, insertion, repeat expansion, and frameshift mutation at sites involving thymine or guanine, was used to systematically search for sequence variants in pooled samples. DNA sequence variants are detected by cleavage of the amplification products at modified nucleotides, generating a defined series of fragments which can be easily separated on a standard sequencing gel and detected using a fluorescent dye detection system. Briefly, PCR products were generated using FAM-labeled forward primers and HEX-labeled reverse primers. PCR amplification was performed in a 25 μl volume containing 1 unit of Taq polymerase, 1× PCR buffer, 0.2 μM of each labeled primer, an appropriate concentration of MgCl~2~, 200 μM BESS T/G Scan dNTP Mix, and 50 ng of genomic DNA. Thermal cycling conditions were a pre-denaturation of 3 min at 94°C; 35 cycles of 30 sec at 94°C, 30 sec at the appropriate annealing temperature indicated in Table [1](#T1){ref-type="table"}, and 30 sec of extension at 72°C; and a final extension at 72°C for 5 min.
For the excision reaction, 5 μl of the amplification reaction was mixed with 1 μl of the BESS T/G-Scan Excision Enzyme Mix and 1 μl of 10X BESS T/G-Scan Excision Enzyme Buffer, the mixture incubated for 30 minutes at 37°C, and the reaction stopped by adding 5 μl of Stop/Loading Buffer. One microliter of the excision reaction products was mixed with gel loading solution containing 12 μl of formamide and 8 μl of GENESCAN™-500 size standards (Applied Biosystems, Forster City, CA, USA), then the mixture was denatured for 5 min at 95°C, loaded onto a capillary polymer of the ABI 310 Genetic Analyzer, and run for 30 min for size separation. Analysis was performed using the GENESCAN 672 program (Applied Biosystems).
To reduce the cost and speed up the process, a pooling methodology was used. A preliminary test indicated that the sensitivity of BESS T/G-Scan analysis allowed the detection of alleles with a frequency in the population greater than 4.5 % (data not shown), i.e., The assay can recognize 1 heterozygous individual in a DNA pool from this individual and 9 homozygotes (equivalent to a minor allele with a frequency of 5%). DNA from 10 individuals was therefore pooled (10 ng/μl of DNA from each individual) and five such pools, representing 100 chromosomes, were prepared from 50 randomly selected subjects for polymorphism identification.
Selection of SNP markers for the TPH1 gene
------------------------------------------
In addition to the SNPs identified in the present study, five SNP markers within and flanking the human *TPH1*gene were also selected to test the association with BPD. Based on the literature \[[@B34]\] and a public SNP database \[[@B35]\], SNP markers roughly 10 kb apart and with a relatively high minor allele frequency were selected. The *TPH1*intron 7 A218C polymorphism, which has been suggested to be associated with Taiwanese BPD in our previously work, also has been included in the present study (i.e. in7SNP1; A20004C). The genomic localizations of the 10 SNPs examined relative to the transcription start site are given in Table [1](#T1){ref-type="table"}.
SNP genotyping
--------------
Except for the intron 7 A218C polymorphism, which was genotyped using a modified [A]{.underline}mplification [R]{.underline}efractory [M]{.underline}utation [S]{.underline}ystem (ARMS), all other SNPs of the *TPH1*gene were genotyped using a multiple SNP genotyping system which involves multiplex PCR and multiple single base extensions (MSBE). Briefly, two multiplex PCR reactions were performed. One was in a 20 μl volume containing 70 ng of genomic DNA, 2.4 μl of primer mix (0.1 μM of the primer pairs for 5\'flankingSNP3, in1SNP1, in2SNP1, and in6SNP1, and 0.2 μM of the primer-pair for in3SNP1), 400 μM dNTPs, 1× PCR buffer, 1.5 mM MgCl~2~, and 1 U of Taq polymerase. The conditions used were an initial denaturation step of 5 min at 95°C, 30 cycles of amplification (30 s at 95°C, 60 s at 51°C, 90 s at 72°C), and a final extension step of 10 min at 72°C. The other multiplex PCR was performed under the same conditions, but using 0.8 μl of primer mix containing 0.1 μM of the primer pairs for 5\'flankingSNP1, 5\'flankingSNP2, 3\'UTRSNP1, and 3\'UTRSNP2. A 4 μl aliquot of the PCR products was treated with 5 U of shrimp alkaline phosphatase (SAP) and 0.1 U of exonuclease I in a total volume of 10 μl to remove primers and unincorporated dNTPs; the reaction was carried out at 37°C for 1 hour, then terminated by incubation at 72°C for 15 minutes to inactivate the enzymes. The multiplex SBE reaction was carried out using SNP-specific primers and fluorescent-labeled terminators (the ABI PRISM SNaPshot Multiplex Kit). The short-extension reaction was performed on a thermal cycling machine for 25 cycles of 10 s at 96°C, 5 s at 50°C, and 30 s at 60°C. After short-extension, excess ddNTPs were removed from the SBE products by addition of 0.5 U of SAP to the reaction mixture and incubation at 37°C for 1 hour. The purified SBE products were electrophoresed on a ABI PRISM 310 Genetic Analyzer and analyzed using GeneScan software (ABI PRISM).
Statistical analysis
--------------------
The chi-squared test for allelic and genotypic distributions between patients and controls was performed using the CROSSTAB program implemented by SPSS. The Hardy-Weinberg equilibrium was analyzed using the HWE program, version 2.33 \[[@B37]\]. Pairwise LD coefficients D\' \[[@B38]\] among the 10 SNPs were estimated and statistical significances were determined by using the SNP Alyze^®^program (SNP and Disease Association Analysis software; Dynacom Co., Ltd. Kanagawa, Japan). In addition, the PHASE 2.0 program \[[@B39],[@B40]\] was used to construct haplotypes and perform a case-control permutation test, then the Fisher\'s exact test was applied to test differences in haplotype frequencies between cases and controls. All Fisher\'s exact tests (two tails) and estimation of the odds ratio of BPD associated with a particular haplotype were performed using the PROC FREQ program implemented by SAS package (SAS Institute Inc., Cary, NC, USA). Haplotype-tag SNPs were selected using SNPtagger software \[[@B41]\]. To consider the multiple comparisons, a Bonferroni correction was applied in this study thus the p value for reaching significance is 0.005 for 10 SNPs. In addition, the potential confounders such as personality disorders, substance abuses or organic disorders were not considered in the statistic analyses.
Results
=======
Detection of TPH1 gene polymorphisms in the Taiwanese Han population
--------------------------------------------------------------------
To detect SNPs in the human *TPH1*gene, we used the enzymatic cleavage approach to screen all 11 exons, parts of neighboring introns, and the promoter of the gene in 30 unrelated healthy and 20 BPD probands from a previously recruited population \[[@B31]\]. No sequence variation in the coding region of the *TPH1*gene was identified. However, screening for nucleotide variants up to -1,151 bases in the promoter region and down to 1,572 bases in the 3\'UTR identified five SNPs. Two, A-1067G (5\'flankingSNP2) and G-347T (5\'flankingSNP3), were found in the promoter region. Of the remaining 3 SNPs, T3804A (in1SNP1) was located in intron 1/exon 1c, while C27224T (3\'UTRSNP1) and A27237G (3\'UTRSNP2) were located in the 3\'UTR of exon 11.
SNP genotyping and single locus association analysis
----------------------------------------------------
For further association analysis, all 5 SNPs identified in this study, plus one polymorphism in the promoter region (5\'flankingSNP1;\[[@B33]\]) and four selected from the dbSNP database (in2SNP1, in3SNP1, in6SNP1, and in7SNP1), were genotyped with MSBE or ARMS (Table [1](#T1){ref-type="table"}). All genotypes were in Hardy-Weinberg equilibrium (HWE), except for the G-347 T marker in the promoter region that had departure from the HWE in both controls (p = 0.024) and patients (p = 0.024). Genotype coverage was generally good and about 90% of the subjects were successfully typed.
All 10 SNPs were genotyped and have been analyzed by Fisher\'s exact test between the case and control groups. No association can be obtained between any of the *TPH1*gene polymorphisms and BPD. The genotypic and allelic distributions of each marker are listed in Table [2](#T2){ref-type="table"}.
Pairwise linkage disequilibrium (LD) measurement and haplotype analyses
-----------------------------------------------------------------------
The size of the human *TPH1*gene is about 29 kb and the LD between all pairs of 10 SNP markers in all subjects was estimated using the SNP Alyze^®^program (Table [3](#T3){ref-type="table"}). The data above and below the diagonal line in Table [3](#T3){ref-type="table"} show, respectively, the results of the pairwise LD analysis, as indicated by the D\' \[[@B38]\]; a higher value indicating a higher degree of LD), and the p value for the chi-squared test. As shown in Table [3](#T3){ref-type="table"}, the D\' values ranges from 0.032 to 1 and the majority of the D\' value is greater than 0.5. These data suggested that sequence variants within the *TPH1*gene are in strong disequilibrium (the upper diagonal of the Table [3](#T3){ref-type="table"}). Furthermore, as indicated by the lower diagonal of Table [3](#T3){ref-type="table"}, statistically significant LD was detected for almost all marker pairs among these SNPs. Few exceptions were observed for marker pairs T3840A / A-1067G, A27237 / G-347T, and A27237 / GA12517C in which the D\' values are 0.076, 0.032, and 0.039, respectively.
Haplotypes constructed using the PHASE 2.0 program \[[@B39]\] revealed a total of 70 haplotypes in combined controls and patients (data not shown). Among these haplotypes, the majority is in very low frequency but the four major haplotypes with frequency greater than 2% are found in 59.8% of the total population. Estimation of haplotype distribution and the Fisher\'s exact test values for testing the significance of differences in individual haplotype frequencies between case and control groups for the haplotypes with frequency \> 2% in at least one group were listed in Table [4](#T4){ref-type="table"}. Additional test was performed to assess overall haplotype frequency profile differences, rather than frequency differences for a specific haplotype, and p value estimated from 1000 permutation tests was also obtained (Table [4](#T4){ref-type="table"}). Although a trend of significance was detected for the GGGAGACCCA haplotype between the case and control groups (p = 0.028) and few haplotypes showed slightly increased risks in which the value of odds ratios are larger than 2, the significances were abolished after Bonferroni correction for multiple tests thus suggested the effect is minor (Table [4](#T4){ref-type="table"}). Furthermore, the overall comparison for all haplotypes between two groups was not significant (p = 0.082) and the p value from 1000 time permutation test was also not significant (p = 0.602).
Haplotype tag selection
-----------------------
Selecting haplotype-tag SNPs (ht-SNPs) can systematically and efficiently refine the process of haplotype construction and reduce the marker number for genotyping. In this study, three htSNPs (5\'flankingSNP1, in2SNP1, in6SNP1) were selected to represent all haplotypes with frequencies larger than 2% using SNPtagger software \[[@B41]\] (Table [5](#T5){ref-type="table"}). Results from the htSNP analysis suggest that the information obtained by examining three SNPs (instead of 10) should be sufficient for genetic analysis, since these can be used to construct and represent all common haplotypes with frequency \>2%. Five haplotypes were constructed using these 3 htSNPs and statistical analyses detected no differences in haplotype distributions between cases and controls (Table [5](#T5){ref-type="table"}).
Discussion
==========
The involvement of the *TPH1*gene in the pathogenesis of affective disorder is supported by several lines of evidence. Previous studies have reported a significant association between *TPH1*intron 7 A218C polymorphism and BPD in the French population \[[@B21]\]. In addition, two current meta-analyses confirmed the significant association of suicide-related behavior with *TPH1*A218 polymorphism \[[@B42],[@B43]\] and suggested that the A allele has a dose-dependent effect on the risk of suicidal behavior \[[@B43]\]. Since these results suggested that *TPH1*could be a strong candidate for involvement in BPD, we have systematically searched all the exons and promoter region of the gene for SNPs. Five polymorphisms were identified, one of which, 3\'UTRSNP2, is a novel polymorphism not previously reported. The only sequence variant found in all *TPH1*exons was in the exon 1c/intron1 region (T3804A, dbSNP ID: rs623580), but the polymorphism is within the 5\'-UTR and therefore does not result in an amino acid substitution. This finding is consistent with a previous report in which no coding sequence variant of *TPH1*was detected in Americans (Indians or Caucasians), Italians, and Finns \[[@B44]\]. These results demonstrate the highly conserved nature of the human *TPH1*gene. The exon 1c/intron 1 T3804A polymorphism has been previously reported and has an estimated minor allele frequency of 0.001 in the Swiss population \[[@B34],[@B45]\].
The five identified SNPs, together with five additional SNPs taken from the literature and a public database, were tested for an association with BPD in Taiwanese patients. In general, the alleles in the *TPH1*gene were common, with minor allele frequencies ranging from 21% to 49% and from 22% to 49% in cases and controls, respectively. The genotypic frequencies of all markers, except the G-347T (5\'flankingSNP3) marker in the promoter region, showed Hardy-Weinberg equilibrium (HWE) in both populations. The departure from HWE of 5\'flankingSNP3 was observed in both controls (p = 0.024) and patients (p = 0.024). A separate study in our lab indicated that the location of the G-347T polymorphism is on the transcriptional repressor GATA1 binding site \[[@B46]\] and alteration of alleles indeed change promoter activity in a luciferase reporter gene system \[[@B47]\]. The departure from HWE could be explained by the negative selection of the homozygous TT individuals. And the presence of very low frequency of TT genotype in the population may be the reason for the impossibility of detecting any association. Despite the strong functional role of the *TPH1*G-347T polymorphism, no association was detected between any of the polymorphisms and BPD in Taiwanese.
The extent and distribution of LD in humans has been a hot topic, especially for gene mapping of complex diseases. In this study, significant LD, as measured by the D\' and P values using the SNP Alyze^®^program, could be detected between T-1721G and C27224T polymorphisms of the *TPH1*gene, which are separated by roughly 29 kb (Table [3](#T3){ref-type="table"}). Because LD-induced association between multiple loci that harbor disease-predisposing alleles can be identified by haplotype-based analyses \[[@B48]\], haplotypes were constructed and their frequencies were compared between cases and controls. Haplotype distributions among 10 *TPH1*SNPs were estimated and twelve haplotypes with frequency larger than 2 % in at least one group were listed (Table [4](#T4){ref-type="table"}). The three most common haplotypes were identical in both groups and were found in 58 % and 56.7 % of the patients and controls, respectively. Although significant difference in haplotype distribution was seen in one comparison, the association was weak and was lost after the Bonferroni correction for multiple tests. Additional permutation tests for haplotype distributions in case and control groups were performed and result indicated no differences in haplotype profile between two groups (p value = 0.602). The data suggest that both BPD patients and controls are actually from the same population thus it implies the *TPH1*gene may not be related to BPD etiologies.
Positive association between *TPH1*intron 7 polymorphism and BPD was identified in our previous study \[[@B31]\], but the replication with extended samples has failed to confirm this association in both single-locus and haplotype analyses. One possible reason for this discrepancy could be at the sampling bias as the age of controls in the extended population is much younger than the cases. Since mood disorders may occur late in life, the sampling bias represents a major limitation of this study. Another possible explanation may be that the *TPH1*gene has only a minor effect on BPD etiology, this effect being missed when heterogeneous samples are used. Alternatively, the positive association we obtained initially could be a false positive outcome from a very small sample size being used. Recently, studies of Tph1 (the original Tph) knockout mice were found to express normal amounts of serotonin in brain, but not in the periphery \[[@B11]\], and resulted in a cardiac dysfunction phenotype \[[@B49]\]. Follow up studies found that the second tryptophan hydroxylase (TPH2, also known as nTPH) gene is predominantly expressed in the brain stem, while the classical TPH1 is expressed in the pineal gland \[[@B14]\] and peripheral tissues (duodenum, kidney or lung;\[[@B15]\]. The amount of TPH2 mRNA expression in individual raphe cells was estimated to be approximately 2.5-fold greater than the level of TPH1 expression in pinealocytes \[[@B14]\]. These findings have changed the consideration of linking polymorphism of *TPH1*gene with various psychiatric diseases \[[@B12]\] and perhaps, could explain the lack of association between the TPH1 gene and BPD obtained in this study. Whether these two paralog proteins are regulated independently or if they have distinct functions in the brain are still under investigation; studies to establish connections between *TPH2*gene and various psychiatric diseases including BPD are on going and should provide more insights regarding the TPH2 function in the brain.
Studies on sequence variations in the human genome have revealed that the human genome can be parsed objectively into haplotype blocks, with limited diversity within each block \[[@B50],[@B51]\]. Johnson et al. \[[@B52]\] determined the extended haplotype at any given locus in a population to identify the SNPs in a gene or a LD region, information that is essential for association studies. The so-called \"haplotype tag SNPs (htSNPs)\" capture the majority of haplotype diversity within a region and thus represent the minimal number of markers that need to be typed to define the common haplotypes (higher than 5% frequency in the population). In this study, we used a publicly program to define htSNPs that represent all haplotypes of the *TPH1*locus with a frequency \> 2% \[[@B41]\]. Three htSNPs, corresponding to all common haplotypes were generated using SNPtagger software \[53\] (Table [5](#T5){ref-type="table"}). Although further analysis using haplotypes constructed with htSNPs revealed no differences in frequency distributions between cases and controls (Table [5](#T5){ref-type="table"}), the htSNPs selected for human *TPH1*gene significantly reduce the number of markers required for genotype analysis (in this case from 10 to 3 SNPs), which may be useful in other studies on the Taiwanese Han population.
Conclusion
==========
Bipolar disorder is a complex genetic disorder with a spectrum of phenotype is associated with bipolar susceptibility genes. Since the subject assessment in this study was only by direct clinical interview according to the DSM-IV but not supported by any other standard tools like the SCID, it is worth noting the potential weakness of the study. The simple assessment procedure without dimensional measures of other psychopathological features also limits the possibility to detect association with particular attribute related to the disease. In summary, results from both single-locus and haplotype analyses did not support the role of *TPH1*gene in BPD etiology. Furthermore, our data indicated significant LD within this 29 Kb interval of the *TPH1*gene in Taiwanese population. Since haplotype frequencies and LD often differ between racial/ethnic groups, the three htSNPs identified in this study should be beneficial in future application for genetic study in Taiwanese population.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
TJL was in charge of clinical examination and provided clinical samples. CYW designed polymorphism screening methods and identified all *TPH1*polymorphisms in Taiwanese. HWT conducted genotyping experiments and provided statistical analysis. YJL contributed to genotyping data analysis. HSS provided overall study design, analysis, and drafted the manuscript. All authors read and approved the final manuscript.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1471-2350/6/14/prepub>
Acknowledgements
================
This study was supported by grants from the National Science Council, Taiwan, ROC (NSC 89-2320-B-006-062-A89-089 and NSC 90-2320-B-006-073).
Figures and Tables
==================
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Summary of the 10 human *TPH1*SNPs genotyped in this study^a^
:::
----------------------------------------------------------------------------------------------------------------------------------
SNP name (Genomic localization)^b^ Position in the gene SNP ID^c^ PCR primers and short-extension probe^e^
------------------------------------ ---------------------- -------------- -------------------------------------------------------
5\'flankingSNP1 (T-1721G) 5\' flanking region SNP000574351 F: 5\'-ctgttcttttggtgtcctc-3\'\
R:5\'-gctcctggcacttaacata-3\'\
P: 5\'-taatttctttcatgagtattttatagtt
5\'flankingSNP2 (A-1067G) 5\' flanking region SNP000574353 F: 5\'-ctgttcttttggtgtcctc-3\'\
R: 5\'-gctcctggcacttaacata-3\'\
P: 5\'-ttttttgctgagtatggatgtactttaaagctcagga
5\'flankingSNP3 (G-347T) 5\' flanking region SNP000574354 F: 5\'-cgataataggcgttatcttg-3\'\
R: 5\'-ctcaatctctgcgtgtatct-3\'\
P: 5\'-tcaggactgggctattaaatagcccagaagcacagaga
in1SNP1 (T3804A) Intron 1 (exon 1c) rs623580 F: 5\'-taattatcctccctccaagt-3\'\
R: 5\'-cttacccattcaattaccac-3\'\
P: 5\'-agagtatgggcgacgttgtccta
in2SNP1 (G7465A) Intron 2 rs684302 F: 5\'-tgctcttatatgtcttttcaagt-3\'\
R: 5\'-gagagatggagcaaaacac-3\'\
P: 5\'-ttaaataaaatacctgtatgtcttcttccatca
in3SNP1 (A12517C) Intron 3 rs211105 F: 5\'-tcaggaaaacagaagggta-3\'\
R: 5\'-ggtaaattgccctatttctaa-3\'\
P: 5\'-aggtggcaaagacaaatgatttctaagatcttttccatcggc
in6SNP1 (C18626G) Intron 6 rs2237907 F: 5\'-gggaagaaattatgtaagtgg-3\'\
R: 5\'-gaaatgttccatatctgtgc-3\'\
P: 5\'-ttgtaatgcacacaaaactgaaagctgatctcttagggtctggagc
in7SNP1 (A20004C)^d^ Intron 7 rs1800532 CF: 5\'-acccacctacactttcctc-3\'\
CR: 5\'-taattgacaacctattaggttc-3\'\
AR: 5\'-agcacatgtgaagcatttag-3\'\
AF: 5\'-cctatgctcagaatagcagctct-3\'
3\'UTRSNP1 (C27224T) 3\' UTR rs2108977 F: 5\'-cacttgaatatcacagtccatc-3\'\
R: 5\'-gcttacagtagatttccttgc-3\'\
P: 5\'-tacatttgatggtaaatagatgctagctaatct
3\'UTRSNP2 (A27237G) 3\' UTR New F: 5\'-cacttgaatatcacagtccatc-3\'\
R: 5\'-gcttacagtagatttccttgc-3\'\
P: 5\'-aactataaatcagataatcaata
----------------------------------------------------------------------------------------------------------------------------------
a\. in7SNP1 was genotyped using ARMS, while the other SNPs were genotyped using MSBE.
b\. Genomic localizations of SNPs are given in bp relative to the transcription start site (position 0).
c\. The ID for 5\'flankingSNP1, 5\'flankingSNP2, and 5\'flankingSNP3 are from HGVbase <http://hgvbase.cgb.ki.se/>. The ID for in1SNP1, in2SNP1, in3SNP1, in6SNP1, in7SNP1, 3\'UTRSNP1, and 3\'UTRSNP2 are from dbSNP <http://www.ncbi.nlm.nih.gov/SNP/>.
d\. The A20004C is known as intron7 A218C polymorphism.
e\. All PCR reactions were performed at the same annealing temperature (53°C). F: forward primer, R: reverse primer, P: short extension probe.
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Genotype and allele frequencies of the *TPH1*gene polymorphisms.
:::
Marker Group^a^ Genotype^b^(frequency) Allele (frequency)
--------- ---------- ------------------------ -------------------- --------- ------- ------- ---------- --------- ------- -------
T-1721G N(94) 54(.57) 37(.39) 3(.03) 2.79 0.248 145(.77) 43(.23) 0.36 0.548
P(92) 53(.58) 31(.34) 8(.09) 137(.75) 47(.26)
A-1067G N(90) 53(.59) 34(.38) 3(.03) 1.62 0.446 140(.78) 40(.22) 0.74 0.389
P(92) 51(.55) 34(.37) 7(.08) 136(.74) 48(.26)
G-347T N(102) 58(.57) 43(.42) 1(.01) 0.048 0.976 159(.78) 45(.22) 0.037 0.847
P(94) 52(.55) 41(.44) 1(.01) 145(.77) 43(.23)
T3804A N(101) 56(.55) 38(.38) 7(.07) 2.17 0.304 152(.74) 52(.26) 0.015 0.946
P(96) 49(.51) 44(.46) 3(.03) 142(.74) 50(.26)
G7465A N(100) 31(.31) 43(.43) 26(.26) 0.46 0.793 105(.53) 95(.48) 0.439 0.508
P(74) 25(.34) 33(.45) 16(.22) 83(.56) 65(.44)
A12517C N(101) 58(.57) 39(.39) 4(.04) 0.232 0.890 155(.77) 47(.23) 0.207 0.649
P(96) 58(.60) 35(.37) 3(.03) 151(.79) 41(.21)
C18626G N(100) 28(.28) 47(.47) 25(.25) 1.24 0.538 103(.52) 97(.49) 0.392 0.531
P(71) 25(.35) 28(.39) 18(.25) 78(.55) 64(.45)
A20004C N(102) 40(.39) 45(.44) 17(.17) 1.726 0.422 125(.61) 79(.39) 1.795 0.18
P(89) 29(.32) 39(.44) 21(.24) 97(.54) 81(.46)
C27224T N(95) 31(.33) 49(.52) 15(.16) 3.58 0.167 111(.58) 79(.42) 2.13 0.144
P(80) 23(.29) 35(.44) 22(.28) 81(.51) 79(.49)
A27237G N(95) 49(.52) 40(.42) 6(.06) 0.57 0.751 138(.73) 52(.27) 0.677 0.411
P(78) 37(.47) 34(.44) 7(.09) 107(.69) 49(.34)
a\. N indicates the matched normal controls, P indicates the bipolar patients.
b\. 1 represents the major allele and 2 the minor allele.
:::
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Pairwise linkage disequilibrium indicated by the D\' value (above the diagonal)^*a*^and statistical significance indicated by the p value (below the diagonal)^*b*^for the ten *TPH1*SNPs.
:::
T-1721G A-1067G G-347T T3840A G7465A A12517C C18626G A20004C C27224T A27237G
--------- ------------------------- ------------------------- ------------------------- --------- --------- ------------------------- --------- --------- --------- ---------
T-1721G 1.000 0.911 0.129 0.834 0.505 0.601 0.726 0.329 0.177
A-1067G \<0.001 0.913 0.076 0.863 0.537 0.655 0.722 0.391 0.274
G-347T \<0.001 \<0.001 0.121 0.840 0.450 0.645 0.655 0.447 0.032
T3840A 0.020 **[0.166]{.underline}** 0.033 0.969 0.575 0.870 0.652 0.512 0.574
G7465A \<0.001 \<0.001 \<0.001 \<0.001 0.685 0.711 0.826 0.741 0.929
A12517C \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 0.684 0.320 0.663 0.039
C18626G \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 0.765 0.684 0.780
A20004C \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 0.624 0.625
C27224T \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 0.956
A27237G **[0.203]{.underline}** **[0.096]{.underline}** **[0.840]{.underline}** \<0.001 \<0.001 **[0.809]{.underline}** \<0.001 \<0.001 \<0.001
^a^Standardized linkage disequlibrium coefficient (absolute D\' value) of Lewontin was shown.
^b^Significant levels based on the chi-squared distribution to test whether the disequilibrium value differs from zero (one degree of freedom). Bold and underline indicate no significant LD at the *p*\< 0.05 level.
:::
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Haplotype frequency estimation and test statistics between case and control.
:::
**Haplotype^a^** **Patient** **Normal** **Fisher\'s Exact Test (*P*value)^b^** **OR^b,c,d^** **95% CI^d^**
---------------------------------- ------------- ------------ ---------------------------------------- --------------- ---------------
**TAGTAAGACA** 0.340 0.359 0.755 0.919 0.61--1.38
**TAGAGACCTG** 0.145 0.140 1.000 1.035 0.59--1.80
**GGTTGCCCTA** 0.095 0.068 0.366 1.439 0.70--2.96
**TAGTAACACA** 0.010 0.039 0.105 0.250 0.05--1.19
**TAGTAAGATA** 0.025 0.010 0.278 2.615 0.50--13.64
**TAGTGCGACA** 0.025 0.015 0.498 1.735 0.41--7.36
**GGGAGACCCA** 0.025 0 0.028 \-- \--
**TAGTAACCTG** 0.020 0 0.058 \-- \--
**TAGTGACCTG** 0.020 0.010 0.443 2.081 0.38--11.49
**TAGAGACATG** 0.020 0.019 1.000 1.031 0.25--4.18
**GGTTGCCCTG** 0.020 0.010 0.443 2.082 0.38--11.49
**GGTAGACCCA** 0.015 0.024 0.724 0.612 0.14--2.59
**Others** 0.240 0.306 0.149 0.717 0.46--1.11
Total 0.082
***P*value of permutation test** **0.602**
a\. The order of the markers, shown 5\' → 3\', is (left to right) T-1721G, A-1067G, G-347T, T3804A, G7465A, A12517C, C18626G, A20004C, C27224T, A27237G.
b\. Chi-square, Fisher\'s Exact Test and Odds Ratio were calculated by using the SAS program.
c\. The Odds Ratio could not be calculated for the haplotype GGGAGACCCA and TAGTAACCTG, because of the zero value in the population.
d\. OR: odds Ratio; CI: confidence Interval.
:::
::: {#T5 .table-wrap}
Table 5
::: {.caption}
######
Haplotype frequency and test statistics between cases and controls using htSNPs.
:::
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Haplotypes from htSNPs^a^** **Represented haplotypes from 10 SNPs^a^** **Patient (frequency)** **Normal** **Fisher\'s Exact Test (*P*value)^b^** **OR^a,c^** **95% CI^c^**
------------------------------- -------------------------------------------- ------------------------- ------------ ---------------------------------------- ------------- ---------------
**TAG** **TAGTAAGACA**\ 0.365 0.369 1.000 0.983 0.66--1.47
**TAGTAAGATA**
**TGC** **TAGAGACCTG**\ 0.185 0.169 0.699 1.109 0.66--1.85
**TAGTGACCTG**\
**TAGTGACATG**
**GGC** **GGTTGCCCTA**\ 0.155 0.102 0.137 1.616 0.89--2.92
**GGGAGACCCA**\
**GGTTGCCCTG**\
**GGTAGACCCA**
**TAC** **TAGTAACACA**\ 0.030 0.039 0.787 0.765 0.26--2.25
**TAGTAACCTG**
**TGG** **TAGCGCGACA** 0.025 0.015 0.498 1.735 0.41--7.36
**Others** 0.240 0.306 0.149 0.717 0.46--1.11
**Total** 0.446
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
a\. SNPs in bold and dark color are htSNPs selected from common haplotypes using the SNPtagger program.
b\. Chi-square, Fisher\'s Exact Test and Odds Ratio were calculated by using the SAS program.
c\. OR: odds Ratio; CI: confidence Interval.
:::
|
PubMed Central
|
2024-06-05T03:55:55.316862
|
2005-3-31
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079843/",
"journal": "BMC Med Genet. 2005 Mar 31; 6:14",
"authors": [
{
"first": "Te-Jen",
"last": "Lai"
},
{
"first": "Chia-Yen",
"last": "Wu"
},
{
"first": "Hsu-Wen",
"last": "Tsai"
},
{
"first": "Yi-Mei J",
"last": "Lin"
},
{
"first": "H Sunny",
"last": "Sun"
}
]
}
|
PMC1079844
|
Background
==========
Observational studies have found blood total cholesterol concentrations below about 4.0 mmol/L (155 mg/dL) to be associated with higher rates of mortality and morbidity from certain non-vascular causes (in particular, cancer of lung, liver, pancreas and blood; chronic obstructive pulmonary disease; cirrhosis; suicide; and haemorrhagic stroke) \[[@B1],[@B2]\]. Excluding events within the first few years after the baseline measurement of cholesterol attenuates some, but not all, of these associations. It remains unclear, however, whether these inverse associations are causal (with low cholesterol actually causing certain diseases), or instead due to confounding or reverse causation (with certain habits or conditions independently causing both lower cholesterol and disease) \[[@B3],[@B4]\]. Randomised trials are not subject to such biases, but the trials of cholesterol-lowering interventions before the statins were unable to assess causation reliably, chiefly because they involved too few non-vascular outcomes (even in combination) and assessed only modest cholesterol reductions from high pretreatment concentrations \[[@B5],[@B6]\]. Some slight excesses of non-vascular deaths or cancers in particular trials provoked much comment over the years, contributing to the uncertainty that surrounded cholesterol-lowering \[[@B7]-[@B11]\].
More recently, in previous trials of statin treatment, LDL cholesterol was typically reduced by about 1.0 mmol/L (38 mg/dl) and major vascular events reduced by about 25% \[[@B12]-[@B19]\]. Although meta-analyses of those trials indicate that statin therapy reduces vascular mortality \[[@B20]-[@B23]\], the secondary prevention trials involved few deaths from non-vascular causes \[[@B12]-[@B15]\], and the primary prevention trials involved relatively few deaths from any cause \[[@B16]-[@B19]\]. Moreover, these trials provided relatively little information about the effects of lowering total cholesterol to the low concentrations (e.g. around 4.0 mmol/L) that had previously raised concerns. By contrast, the Heart Protection Study (HPS) involved large numbers of deaths from both vascular and non-vascular causes among people presenting with below-average cholesterol levels who were randomly allocated to have their cholesterol lowered substantially for several years \[[@B24]\]. Consequently, it allows outstanding concerns about the potential hazards of lowering cholesterol to be addressed much more reliably than has previously been possible. The present report provides detailed information about the effects of lowering cholesterol on cause-specific mortality, site-specific cancer incidence and other major morbidity in a range of different circumstances.
Methods
=======
Details of HPS have been reported previously \[[@B24]-[@B29]\] (see also <http://www.hpsinfo.org>) and are summarised below.
Recruitment and follow-up
-------------------------
Men and women aged about 40 to 80 years with non-fasting blood total cholesterol concentrations of at least 3.5 mmol/L (135 mg/dL) were eligible provided they had a medical history of: occlusive arterial disease; diabetes mellitus; or treated hypertension (if also male and aged at least 65 years). People were ineligible if their own doctor considered statin therapy to be clearly indicated or contraindicated, or if they had a past history of: stroke, myocardial infarction or angina hospitalisation within the previous 6 months; chronic liver disease or evidence of abnormal liver function; severe renal disease or evidence of substantially impaired renal function; inflammatory muscle disease or evidence of muscle problems; concurrent treatment with cyclosporin, fibrates or high-dose niacin; child-bearing potential; severe heart failure; life-threatening conditions other than vascular disease or diabetes (including any cancer except non-melanoma skin cancer); or any other condition that might limit long-term compliance.
Eligible and consenting patients entered a pre-randomisation \"run-in\" treatment phase, which involved 4 weeks of placebo followed by 4--6 weeks of 40 mg simvastatin daily. Compliant individuals who were still not considered by their own doctors to have a clear indication for, or contraindication to, statin therapy were then randomly allocated to receive 40 mg simvastatin daily or matching placebo tablets for about 5 years (and separately, using a 2 × 2 factorial design, to receive antioxidant vitamins or matching placebo capsules \[[@B29]\]). Randomised participants were to be seen at 4, 8 and 12 months, and then 6-monthly (with follow-up by telephone for individuals who did not attend or, alternatively, via their general practitioner). Compliance with study treatment was assessed by questioning participants and reviewing the calendar-packed tablets remaining. Blood samples were taken at each follow-up visit for central laboratory assay of alanine transaminase to monitor liver function, and of creatine kinase in any participant reporting unexplained muscle symptoms or concomitant use of a non-study statin. To assess the effects of the treatment allocation on the lipid profile, assays were performed on non-fasting blood collected from a sample of participants due for follow-up at about the same time each year, and from all participants attending follow-up at an average of 4.6 years. Differences in average blood lipid concentrations were based on comparisons between all those allocated simvastatin and all those allocated placebo, irrespective of whether or not they were still compliant.
Information was recorded at each follow-up about any suspected myocardial infarction, stroke, vascular procedure, cancer or other serious adverse experience, and about the main reasons for all other hospital admissions. UK national registries provided information about the sites of any registered cancers and the certified causes of death. Further details were sought from general practitioners (and, if considered necessary, hospital records) about all reports that might relate to major vascular events, cancers (e.g. investigations) or deaths. All such information was coded according to prespecified criteria by coordinating centre clinical staff, who were unaware of the participants\' study treatment allocation. Analyses were based on confirmed plus unrefuted reports of events. Cancers were classified according to their primary anatomical site rather than their histology (except that skin cancers were subclassified as melanoma or non-melanoma), with definite confirmation for 98% of the included cancers. Mortality follow-up was available for 99.7% of participants, with certified causes for 98.2% of deaths.
Statistical analyses
--------------------
The main comparisons involved logrank analyses of the first occurrence of particular events during the scheduled treatment period among all those allocated simvastatin versus all those allocated matching placebo tablets (i.e. \"intention-to-treat\"). These logrank analyses yielded both the event rate ratio (RR) and the test of statistical significance (two-sided probability value). The prespecified primary comparisons were of the effects of allocation to simvastatin on deaths from all causes and, separately, on deaths from all coronary causes and from all non-coronary causes. But since simvastatin appeared to reduce the risk of death not only from coronary causes but also from other vascular causes (as well as preventing non-fatal vascular events) \[[@B24]\], the present analyses are of all vascular deaths and of all non-vascular deaths.
Prespecified secondary comparisons included the effects of simvastatin allocation on specific vascular and non-vascular causes of death \[[@B25]\], with due allowance made in their interpretation for multiple hypothesis testing, for the effects observed on relevant non-fatal events, and for evidence from other studies. It was not anticipated that there would be adequate statistical power to assess the effects of study treatment on vascular or total mortality directly in different circumstances. Instead, the prespecified comparisons involved assessment of the effects on major coronary events (defined as non-fatal myocardial infarction or coronary death), and on the even larger numbers of major vascular events (i.e. major coronary event, stroke or arterial revascularisation), in the first two years and in the later years of scheduled treatment (in order to determine whether any protective effect increases with time), and in various prespecified subcategories determined at study entry, including: sex; age (\<65; ≥65\<70; ≥70 years); and pretreatment plasma concentrations of total cholesterol (\<5.0; ≥5.0\<6.0; ≥6.0 mmol/L) and of LDL cholesterol (\<3.0; ≥3.0\<3.5; ≥3.5 mmol/L). Pre-specified tertiary comparisons included the effects on vascular mortality separately during years 1--2 and years 3+ of follow-up, on cause-specific mortality in the prespecified subcategories of pretreatment total cholesterol, and on site-specific cancer incidence and cerebral haemorrhage. Tests for heterogeneity or, if more appropriate, trend were to be used to assess whether the proportional effects observed on particular outcomes in specific subcategories differed clearly from the overall effect (after due allowance for multiple comparisons).
Role of the funding sources
---------------------------
The investigators were responsible for the study design, data collection, data analysis, data interpretation, and writing of the report, independently of all funding sources.
Results
=======
Patient enrolment
-----------------
A total of 20,536 individuals were randomised between July 1994 and May 1997. Mean age at study entry was 64.0 years (SD 8.4), with 5806 aged at least 70 years. Prior to any statin treatment being started, participants who were subsequently randomised had mean non-fasting plasma concentrations of total cholesterol of 5.9 mmol/L (SD 1.0), directly-measured LDL cholesterol of 3.4 mmol/L (0.8), HDL cholesterol of 1.06 mmol/L (0.3) and triglycerides of 2.1 mmol/L (1.4). There were 4072 randomised participants whose pretreatment measurements of total cholesterol were below 5.0 mmol/L (193 mg/dL) and 6793 with pretreatment LDL cholesterol below 3.0 mmol/L (116 mg/dL). The large size of the study (and the use of minimised randomisation \[[@B30]\]) produced good balance between the treatment groups for the main prognostic features that were measured (see subcategory figures below), and should have done likewise for those that were not.
Compliance and effects on blood lipids
--------------------------------------
The mean duration of follow-up was 5.0 years for all randomised participants: 5.3 years for those who survived to the scheduled end of study treatment and about half that for those who did not (yielding 51,121 person years of follow-up for all those allocated simvastatin and 50,664 for all those allocated placebo). Compliance at each follow-up was defined as at least 80% of the scheduled simvastatin or placebo tablets having been taken since the previous follow-up. Among the participants allocated 40 mg simvastatin daily, average statin use during the scheduled treatment period was 85% (with 82% compliant with their allocated simvastatin, 3% on non-study statin alone and 2% on both). By contrast, among those allocated placebo, 4% at the end of the first year of follow-up, but 32% at the end of the fifth year, were taking non-study statin therapy, yielding an average of 17%. Non-study statin use in the placebo group was more common among those who already had diagnosed coronary disease at entry, were younger or had higher pretreatment total or LDL cholesterol concentrations \[[@B24]\]. In each subcategory studied, however, the average difference in total cholesterol was about 1.2 mmol/L (range: 1.1--1.3 mmol/L: table [1](#T1){ref-type="table"}) and in LDL cholesterol was about 1.0 mmol/L (range: 0.9--1.1 mmol/L). In particular, among the 4072 participants whose pretreatment measurements of total cholesterol were below 5.0 mmol/L (193 mg/dL), the average total cholesterol concentration during the trial was 3.5 mmol/L in the simvastatin group compared to 4.6 mmol/L in the placebo group.
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Average plasma total and LDL cholesterol concentrations during follow-up
:::
**Baseline characteristic** **Plasma total cholesterol (mmol/l)** **Plasma LDL cholesterol (mmol/l)**
-------------------------------- --------------------------------------- ------------------------------------- ---------- --------- --------- ----------
**Sex**
Male 4.1 5.3 -1.2 2.2 3.2 -1.0
Female 4.6 5.8 -1.2 2.5 3.4 -0.9
**Age (years)**
\<65 4.3 5.4 -1.1 2.4 3.2 -0.9
≥65 \<70 4.1 5.4 -1.3 2.2 3.3 -1.0
≥70 4.2 5.5 -1.3 2.2 3.3 -1.1
**Total cholesterol (mmol/L)**
\<5.0 3.5 4.6 -1.1 1.8 2.6 -0.9
≥5.0 \<6.0 4.0 5.2 -1.2 2.1 3.1 -1.0
≥6.0 4.8 6.0 -1.2 2.7 3.7 -1.0
**LDL cholesterol (mmol/L)**
\<3.0 3.7 4.9 -1.1 1.8 2.7 -0.9
≥3.0 \<3.5 4.1 5.3 -1.2 2.2 3.2 -1.0
≥3.5 4.7 5.9 -1.2 2.7 3.7 -1.0
**ALL PATIENTS** **4.2** **5.4** **-1.2** **2.3** **3.3** **-1.0**
\* Intention-to-treat comparisons with missing data imputed from initial pre-treatment screening values. The absolute difference in LDL-cholesterol that would be produced by full compliance with 40 mg simvastatin daily can be estimated as the ratio of the absolute difference to the estimated compliance. For example, -1.0/67% = -1.5 mmol/L.
:::
Effects on vascular mortality and morbidity
-------------------------------------------
Overall, allocation to simvastatin produced a highly significant 17% (SE 4; p \< 0.0001) proportional reduction in vascular mortality during the 5 years of the study (table [2](#T2){ref-type="table"}). This reflected a definite 18% (SE 5; p = 0.0005) reduction in deaths due to coronary causes, together with a non-significant 20% (SE 12; p = 0.1) reduction in fatal strokes and 12% (SE 13; p = 0.3) reduction in deaths from other vascular causes, with no significant difference between the effects observed on these different vascular causes (heterogeneity p = 0.9). Among the coronary causes there were also no statistically significant differences among the effects of statin allocation on deaths attributed to acute myocardial infarction, sudden death, heart failure secondary to coronary disease, or other coronary causes (heterogeneity p = 0.5). There was no apparent difference between the treatment groups in the small number of deaths attributed to cerebral haemorrhage, whereas there was a non-significant reduction in fatal strokes due to ischaemic (or unknown) causes. Other vascular causes comprised deaths from peripheral vascular disease (which included ruptured aortic aneurysm) and from a variety of other cardiac conditions (including a few attributed to heart failure), again with no significant heterogeneity in the effects of simvastatin allocation on these different causes of death (heterogeneity p = 0.7).
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Effect of simvastatin allocation on vascular and non-vascular causes of death
:::
**Cause of death** **Simvastatin-allocated (10,269)** **Placebo- allocated (10,267)** **Death rate ratio (& 95% CI)** **P-value**
------------------------------- ------------------------------------ --------------------------------- --------------------------------- --------------
**Coronary**
Acute MI 141 (1.4%) 191 (1.9%) 0.73 (0.59 -- 0.91)
Sudden death 147 (1.4%) 154 (1.5%) 0.95 (0.75 -- 1.19)
Heart failure\* 65 (0.6%) 78 (0.8%) 0.82 (0.59 -- 1.15)
Other coronary 234 (2.3%) 284 (2.8%) 0.82 (0.69 -- 0.97)
**Subtotal: Coronary** **587 (5.7%)** **707 (6.9%)** **0.82 (0.74 -- 0.92)** **0.0005**
**Stroke**
Haemorrhagic 23 (0.2%) 24 (0.2%) 0.95 (0.54 -- 1.68)
Ischaemic (or unknown) 73 (0.7%) 95 (0.9%) 0.76 (0.56 -- 1.03)
**Subtotal: Stroke** **96 (0.9%)** **119 (1.2%)** **0.80 (0.61 -- 1.05)** **0.1**
**Other vascular**
Peripheral vascular 58 (0.6%) 63 (0.6%) 0.91 (0.64 -- 1.30)
Other cardiac\* 40 (0.4%) 48 (0.5%) 0.83 (0.54 -- 1.25)
**Subtotal: Other vascular** **98 (1.0%)** **111 (1.1%)** **0.88 (0.67 -- 1.15)** **0.3**
**VASCULAR** **781 (7.6%)** **937 (9.1%)** **0.83 (0.75 -- 0.91)** **\<0.0001**
**Neoplastic**
Respiratory 127 (1.2%) 133 (1.3%) 0.94 (0.74 -- 1.20)
Gastrointestinal 112 (1.1%) 103 (1.0%) 1.08 (0.82 -- 1.41)
Genitourinary 47 (0.5%) 46 (0.4%) 1.01 (0.67 -- 1.52)
All others 73 (0.7%) 62 (0.6%) 1.17 (0.83 -- 1.63)
**Subtotal: Neoplastic** **359 (3.5%)** **345 (3.4%)** **1.03 (0.89 -- 1.19)** **0.7**
**Other non-vascular**
Respiratory 90 (0.9%) 114 (1.1%) 0.78 (0.59 -- 1.03)
Gastrointestinal 35 (0.3%) 41 (0.4%) 0.85 (0.54 -- 1.33)
Other medical^†^ 47 (0.5%) 49 (0.5%) 0.95 (0.64 -- 1.42)
Non-medical 16 (0.2%) 21 (0.2%) 0.75 (0.40 -- 1.44)
**NON-VASCULAR** **547 (5.3%)** **570 (5.6%)** **0.95 (0.85 -- 1.07)** **0.4**
**ALL DEATHS** **1328 (12.9%)** **1507 (14.7%)** **0.87 (0.81 -- 0.94)** **0.0003**
\* Heart failure deaths were subdivided into those considered to be due to coronary disease and those (5 versus 8) due to other vascular (or unknown) causes.
† Includes renal, infectious, metabolic, neurological and unknown deaths.
:::
The apparent reductions in different types of vascular death with allocation to simvastatin are reinforced by more definite effects on the larger numbers of non-fatal and fatal vascular events considered together. For example, the very definite 27% (SE 4; p \< 0.0001: figure [1](#F1){ref-type="fig"}) reduction in first non-fatal myocardial infarction or coronary death (i.e. \"major coronary events\") following randomisation reinforces the observed effect on coronary death \[[@B24]\]. Similarly, the definite 25% (SE 5; p \< 0.0001) reduction in first non-fatal or fatal stroke following randomisation (444 \[4.3%\] simvastatin versus 585 \[5.7%\] placebo) indicates that the trend toward fewer fatal strokes with simvastatin is likely to be real \[[@B28]\]. Despite concerns about possible adverse effects of statin therapy on heart failure \[[@B31]\], the non-significant trend toward fewer heart failure deaths due to any cause (70 \[0.7%\] vs 86 \[0.8%\]; RR 0.81 \[0.59--1.10\]; p = 0.2) is supported by a marginally significant reduction in first hospital admission for worsening heart failure or heart failure death (354 \[3.4%\] vs 405 \[3.9%\]; RR 0.86 \[0.75--1.00\]; p = 0.05).
::: {#F1 .fig}
Figure 1
::: {.caption}
######
**Effect of simvastatin allocation on major coronary events by year of follow-up**. Rate ratios (RR: black squares with area proportional to the amount of \"statistical information\" in each subdivision) comparing outcome among patients allocated simvastatin to that among those allocated placebo are plotted, along with their 95% confidence intervals (CI: horizontal lines). For relevant subtotals, the result and its 95% CI is represented by a diamond, with the overall proportional reduction and statistical significance given alongside. Squares or diamonds to the left of the solid vertical line indicate benefit with simvastatin (with nominal significance of at least two-sided P \< 0.05 when the horizontal line or diamond does not overlap the vertical line). The broken vertical lines indicate the overall rate ratios.
:::
:::
In this high-risk population with occlusive arterial disease or diabetes, about 1.5% of placebo-allocated patients died from vascular causes during each year of follow-up. A highly significant 24% (95% CI 10--35; p = 0.002) proportional reduction in vascular mortality emerged during the first two years after the initiation of simvastatin treatment (figure [2](#F2){ref-type="fig"}), which is reinforced by the prespecified analyses of major coronary events (figure [1](#F1){ref-type="fig"}) and of major vascular events (figure [3](#F3){ref-type="fig"}). Further reductions in vascular mortality were observed during the subsequent years of the scheduled treatment period, and again these are reinforced by the observed effects on vascular events. About one third of the placebo-allocated participants were taking a statin by the end of year 5, and this may account for the somewhat smaller reductions in vascular deaths (and vascular events) observed during the later years. Even so, the continuing reductions in vascular mortality (and morbidity) during each period resulted in increasing absolute benefits with more prolonged treatment and follow-up (figure [4](#F4){ref-type="fig"}). As a consequence, whereas 9.1% of the placebo-allocated patients died of vascular causes during an average of 5 years of follow-up, only 7.6% of those allocated simvastatin did so. Hence, lowering LDL cholesterol by an average of 1.0 mmol/L for 5 years was associated with the prevention of 14 (SE 5) vascular deaths per 1000 participants.
::: {#F2 .fig}
Figure 2
::: {.caption}
######
**Effect of simvastatin allocation on vascular, non-vascular and all-cause mortality by period of follow-up**. Symbols and conventions as in Figure 1.
:::
:::
::: {#F3 .fig}
Figure 3
::: {.caption}
######
**Effect of simvastatin allocation on major vascular events by year of follow-up**. Symbols and conventions as in Figure 1.
:::
:::
::: {#F4 .fig}
Figure 4
::: {.caption}
######
**Life-table plot of effects of simvastatin allocation on vascular and non-vascular death**. See figure 2 for numbers of participants dying during follow-up.
:::
:::
Figure [5](#F5){ref-type="fig"} indicates that the proportional reduction in the rate of vascular death was about one sixth in various different circumstances, and this pattern is also reinforced by the prespecified subgroup analyses of the much larger numbers of major vascular events (figure [6](#F6){ref-type="fig"}). For example, there was a 17% (SE 5; p = 0.0004) reduction in vascular mortality among the 15,454 men and a 19% (SE 11; p = 0.08) reduction among the 5082 women (heterogeneity p = 0.9 between effect in men vs women), which is reinforced by the highly significant reductions in major vascular events both among men and among women. Similarly, the proportional reductions in vascular mortality appeared to be about the same among younger and older participants (heterogeneity p = 0.9), as was also the case for major vascular events. But since the older participants were at higher absolute risk of vascular death, these similar proportional effects translated into larger absolute benefits at older ages during the 5 year treatment period. The reduction in vascular mortality also appeared largely independent of the pretreatment lipid concentrations, and this pattern is again reinforced by the parallel analyses of major vascular events.
::: {#F5 .fig}
Figure 5
::: {.caption}
######
**Effect of simvastatin allocation on vascular death in participants subdivided by presenting features**. Symbols and conventions as in Figure 1. P-values for chi-squared tests for heterogeneity across different subgroups are given on the right.
:::
:::
::: {#F6 .fig}
Figure 6
::: {.caption}
######
**Effect of simvastatin allocation on first major vascular event in participants subdivided by presenting features**. Symbols and conventions as in Figures 1 & 5.
:::
:::
Effects on non-vascular mortality and morbidity
-----------------------------------------------
Overall, there was no evidence that reducing total cholesterol by an average of 1.2 mmol/L for 5 years produced any adverse effect on the aggregate of all non-vascular deaths (RR 0.95 \[95% CI 0.85--1.07\]; p = 0.4: table [2](#T2){ref-type="table"}). More than half of these non-vascular deaths were due to cancer, and there were no significant differences between the treatment groups in the numbers of deaths from all cancers or from cancers at particular sites (or in the numbers of incident cancers: see below). Among participants allocated simvastatin there were non-significantly fewer deaths from all respiratory causes, including those due to chronic obstructive pulmonary disease (26 \[0.3%\] simvastatin vs 39 \[0.4%\] placebo; RR 0.66 \[0.41--1.08\]; p = 0.1). This apparent lack of any adverse effect on respiratory death is reinforced by the similar numbers of participants in the two treatment groups who either died from, or were admitted to hospital for, any respiratory illness (811 \[7.9%\] simvastatin vs 820 \[8.0%\] placebo; RR 0.98 \[0.89--1.08\]; p = 0.7) or for chronic obstructive pulmonary disease (88 \[0.9%\] vs 110 \[1.1%\]; RR 0.79 \[0.60--1.05\]; p = 0.1). Moreover, lung function assessed by spirometry in all those attending the final visit showed no differences between the treatment groups: forced expiratory volume in 1 second (FEV~1~) of 2.06 L simvastatin vs 2.05 L placebo (difference 0.01 L \[SE 0.01\]; p = 0.5); and forced vital capacity (FVC) of 2.82 L vs 2.82 L (difference 0.00 L \[SE 0.01\]; p = 0.9). This was the case even among participants with pretreatment total cholesterol measurements below 5.0 mmol/L: FEV~1~of 2.16 L vs 2.15 L (difference 0.00 L \[SE 0.03\]; p = 0.9); and FVC of 2.94 L vs 2.95 L (difference 0.00 \[SE 0.03\]; p \> 0.9).
There were similar numbers of deaths in the two treatment groups from gastrointestinal causes, which included a small number attributed to liver disease (5 simvastatin vs 3 placebo). Few patients reported developing cirrhosis (4 vs 4), but there was no apparent adverse effect on the much larger number with any non-fatal or fatal liver-related serious adverse event, either overall (197 \[1.9%\] vs 200 \[1.9%\]; RR 0.98 \[0.80--1.19\]; p = 0.8) or among those with pretreatment total cholesterol measurements below 5.0 mmol/L (48 \[2.4%\] vs 41 \[2.0%\]; RR 1.18 \[0.78--1.78\]; p = 0.4). There was also no suggestion of any adverse effect of cholesterol-lowering therapy on other medical causes of death (table [2](#T2){ref-type="table"}), which included small numbers due to infections (7 vs 17) and renal disease (10 vs 10). This apparent lack of effect on renal death is reinforced by the similar numbers in the two treatment groups who developed renal failure or died from renal causes (71 \[0.7%\] vs 63 \[0.6%\]; RR 1.12 \[0.80--1.57\]; p = 0.5). Relatively few non-medical deaths occurred, with the majority being due to accidents and injuries (12 vs 15) and the remainder due to complications of medical or surgical procedures (4 vs 5) or suicide (0 vs 1). There was no apparent excess in the numbers who reported attempted suicide (14 \[0.1%\] vs 11 \[0.1%\]; RR 1.26 \[0.58--2.76\]; p = 0.6), development of depression (39 \[0.4%\] vs 34 \[0.3%\]; RR 1.14 \[0.72--1.80\]; p = 0.6) or any psychiatric disorder (96 \[0.9%\] vs 90 \[0.9%\]; RR 1.06 \[0.79--1.41\]; p = 0.7).
Any adverse effects of lowering cholesterol might be expected to emerge only after some years of lower cholesterol levels, but there was no suggestion of an excess of non-vascular death even during the later years of the study (figures [2](#F2){ref-type="fig"} and [4](#F4){ref-type="fig"}). For example, there were similar rates of non-vascular death in each treatment group during years 3 & 4 (RR 0.97 \[0.81--1.17\]; p = 0.8) and during years 5+ (RR 0.93 \[0.77--1.13\]; p = 0.5). Nor did lowering total cholesterol by an average of about 1.2 mmol/L for 5 years produce an excess of non-vascular mortality in any of the different types of patient studied (figure [7](#F7){ref-type="fig"}). For example, inverse associations between cholesterol concentrations and mortality have been reported from some observational studies among the very elderly \[[@B32],[@B33]\]. But no adverse effect on non-vascular mortality was seen among the 5806 participants in HPS who were aged 70 years or older at entry (RR 0.95 \[0.80--1.13\]; p = 0.6) or, indeed, even among the 1263 participants who were aged 75--80 years at entry (60 \[9.8%\] vs 78 \[12.0%\]; RR 0.78 \[0.56--1.09\]; p = 0.1). Similarly, despite concerns from non-randomised observational studies about higher mortality in association with low cholesterol concentrations, there was no adverse effect on non-vascular mortality even among the 4072 participants with pretreatment total cholesterol measurements below 5.0 mmol/L (RR 0.99 \[0.77--1.26\]; p = 0.9), in whom simvastatin allocation reduced total cholesterol concentrations to an average of 3.5 mmol/L (Table [1](#T1){ref-type="table"}).
::: {#F7 .fig}
Figure 7
::: {.caption}
######
**Effect of simvastatin allocation on non-vascular death in participants subdivided by presenting features**. Symbols and conventions as in Figures 1 & 5.
:::
:::
Effects on all-cause mortality
------------------------------
The lack of any overall effect of cholesterol-lowering therapy on non-vascular mortality in different circumstances (figure [7](#F7){ref-type="fig"}) suggests that the proportional reduction in all-cause mortality in any particular circumstance is likely to depend chiefly on the fraction of deaths that are due to vascular causes and on the proportional reduction in vascular mortality that is produced by lowering LDL cholesterol. For example, in the present study, the 13% proportional reduction in all-cause mortality produced by lowering LDL cholesterol by an average of about 1 mmol/L with statin therapy reflects the combination of the highly significant 17% reduction in the two-thirds of deaths due to vascular causes and the lack of any significant difference in the remaining one third of deaths from non-vascular causes (table [2](#T2){ref-type="table"}). Figure [2](#F2){ref-type="fig"} illustrates this combined effect on all-cause mortality over time, and the similar proportional reductions in total mortality in different types of participant (figure [8](#F8){ref-type="fig"}) reflect the consistent beneficial effects of simvastatin allocation on vascular death, and consistent lack of effect on non-vascular deaths, across subgroups.
::: {#F8 .fig}
Figure 8
::: {.caption}
######
**Effect of simvastatin allocation on all-cause mortality in participants subdivided by presenting features**. Symbols and conventions as in Figures 1 & 5.
:::
:::
Effects on cancer incidence
---------------------------
Overall, there was no evidence that reducing total cholesterol from an average of 5.4 mmol/L to an average of 4.2 mmol/L (table [1](#T1){ref-type="table"}) produced any adverse effect on the incidence of first cancer (excluding non-melanoma skin) at any site following randomisation (814 \[7.9%\] simvastatin vs 803 \[7.8%\] placebo; RR 1.00 \[0.91--1.11\]; p = 0.9: table [3](#T3){ref-type="table"}). Moreover, no significant excess of these cancers emerged even during the later years of the study (figure [9](#F9){ref-type="fig"}). Nor was there any significant evidence that cholesterol-lowering therapy produced adverse effects on the cancer incidence rate in any particular type of participant studied (figure [10](#F10){ref-type="fig"}). For example, by contrast with a marginally significant adverse trend in one recently reported randomised trial of statin therapy \[[@B34]\], simvastatin allocation was not associated with a significant excess of cancer among the 5806 participants aged 70 years or older at presentation (RR 1.02 \[0.88--1.19\]; p = 0.8). Furthermore, whereas non-randomised observational studies have found higher cancer rates in association with low cholesterol levels \[[@B1],[@B2]\], no adverse effect was seen on the cancer incidence rate in this large randomised comparison among the 4072 participants with pretreatment total cholesterol measurements below 5.0 mmol/L (RR 0.87 \[0.71--1.07\]; p = 0.2).
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Effect of simvastatin allocation on site-specific cancer incidence
:::
**Cancer site** **Simvastatin-allocated (10,269)** **Placebo-allocated (10,267)** **Event rate ratio (& 95% CI)** **P-value**
---------------------------------- ------------------------------------ -------------------------------- --------------------------------- -------------
**Respiratory**
Lung/larynx 171 (1.7%) 157 (1.5%) 1.08 (0.87 -- 1.34)
Other 8 (0.1%) 10 (0.1%) 0.79 (0.32 -- 2.00)
**Subtotal: Respiratory** **179 (1.7%)** **167 (1.6%)** **1.06 (0.86 -- 1.31)** **0.6**
**Gastrointestinal**
Upper GI 74 (0.7%) 62 (0.6%) 1.18 (0.84 -- 1.65)
Colon/rectum 114 (1.1%) 131 (1.3%) 0.86 (0.67 -- 1.11)
Other 41 (0.4%) 33 (0.3%) 1.23 (0.78 -- 1.94)
**Subtotal: Gastrointestinal** **228 (2.2%)** **223 (2.2%)** **1.01 (0.84 -- 1.22)** **0.9**
**Connective tissue**
Female breast 38 (1.5%) 51 (2.0%) 0.74 (0.49 -- 1.12)
Melanoma 17 (0.2%) 10 (0.1%) 1.66 (0.78 -- 3.54)
Other 5 (0.0%) 7 (0.1%) 0.71 (0.23 -- 2.20)
**Subtotal: Connective tissue** **60 (0.6%)** **68 (0.7%)** **0.87 (0.62 -- 1.24)** **0.4**
**Genitourinary**
Renal 23 (0.2%) 22 (0.2%) 1.04 (0.58 -- 1.86)
Bladder 74 (0.7%) 90 (0.9%) 0.81 (0.60 -- 1.11)
Prostate 145 (1.9%) 145 (1.9%) 0.99 (0.79 -- 1.25)
Gynaecological 19 (0.7%) 18 (0.7%) 1.05 (0.55 -- 2.00)
Other 6 (0.1%) 6 (0.1%) 0.99 (0.32 -- 3.08)
**Subtotal: Genitourinary** **259 (2.5%)** **272 (2.6%)** **0.94 (0.80 -- 1.12)** **0.5**
**Haematological**
Leukaemia/lymphoma 42 (0.4%) 32 (0.3%) 1.30 (0.82 -- 2.05)
Other 23 (0.2%) 23 (0.2%) 0.99 (0.55 -- 1.76)
**Subtotal: Haematological** **64 (0.6%)** **52 (0.5%)** **1.22 (0.85 -- 1.75)** **0.3**
**Other & unspecified** **54 (0.5%)** **57 (0.6%)** **0.94 (0.65 -- 1.36)** **0.7**
**ALL CANCERS\*** **814 (7.9%)** **803 (7.8%)** **1.00 (0.91 -- 1.11)** **0.9**
\* Prespecified that analyses of cancer incidence were to exclude non-melanoma skin cancer (243 \[2.4%\] simvastatin-allocated versus 202 \[2.0%\] placebo-allocated; p = 0.06).
:::
::: {#F9 .fig}
Figure 9
::: {.caption}
######
**Effect of simvastatin allocation on incidence of first cancer by period of follow-up**. Symbols and conventions as in Figure 1. Excludes non-melanoma skin cancer.
:::
:::
::: {#F10 .fig}
Figure 10
::: {.caption}
######
**Effect of simvastatin allocation on cancer incidence in participants subdivided by presenting features**. Symbols and conventions as in Figures 1 & 5. Excludes non-melanoma skin cancer.
:::
:::
Any adverse effects of cholesterol lowering on cancer might be expected to be restricted to cancers at particular sites: for example, respiratory, gastrointestinal and haematological cancers have been associated with low cholesterol concentrations in observational studies \[[@B1],[@B2]\]. Substantial numbers of respiratory, gastrointestinal and genitourinary cancers were reported during HPS, so it provides a reasonably reliable (and unbiased) assessment of whether lowering cholesterol by more than 1 mmol/L for about 5 years affects the risks of such cancers. There was no suggestion of any excess in the incidence of cancers at these sites, either overall (table [3](#T3){ref-type="table"}) or separately among participants who presented at younger or older age or with lower or higher cholesterol concentrations (figure [11](#F11){ref-type="fig"}). Even in a study of this size, however, too few cancers occurred at some particular sites (e.g. haematological cancers) for reliable assessment, but an on-going meta-analysis of individual patient data from all major statin trials should be able to provide such information \[[@B35]\].
::: {#F11 .fig}
Figure 11
::: {.caption}
######
**Effect of simvastatin allocation on site-specific cancers subdivided by age and total cholesterol at study entry**. Symbols and conventions as in Figures 1 & 5. Excludes non-melanoma skin cancer.
:::
:::
Individuals with a history of non-melanoma skin cancer remained eligible for inclusion in HPS (although those with a history of other cancers did not), and non-melanoma skin cancer reported following randomisation was to be considered separately from other cancers. During the 5-year scheduled treatment period, there were more reports of non-melanoma skin cancer among the simvastatin-allocated participants (243 \[2.4%\] vs 202 \[2.0%\]: RR 1.20 \[0.99--1.44\]), only one of which was associated with death. This difference is not conventionally significant (p = 0.06), even before allowing for the multiple comparisons involved. Moreover, it did not appear to reflect a clear excess in either basal cell carcinomas (124 vs 99; RR 1.24 \[0.96--1.61\]; p = 0.1) or in squamous cell carcinomas (137 vs 122; RR 1.12 \[0.87--1.42\]; p = 0.4). Nor did this slight excess of non-melanoma skin cancer become more apparent with more prolonged treatment, as might be expected if it was causally related. Indeed, the excess was observed largely during years 1 & 2 (107 \[1.0%\] vs 76 \[0.7%\]), with little apparent difference during years 3 & 4 (91 \[0.9%\] vs 86 \[0.9%\]) or years 5+ (45 \[0.5%\] vs 40 \[0.4%\]).
Discussion
==========
The large size of the Heart Protection Study, its prospective randomised design and the inclusion of a broad range of participants allow it to assess reliably both the efficacy and the safety of cholesterol lowering in a variety of different circumstances. The present results demonstrate that lowering LDL cholesterol by about 1 mmol/L (38 mg/dL) for 5 years with a statin reduces the rate of death from vascular causes by about one-sixth, with no apparent adverse effect on non-vascular mortality or morbidity. This proportional reduction in vascular mortality was remarkably consistent among the different types of participant studied, including women as well as men, older as well as younger individuals, and those who entered the study with below average cholesterol concentrations. Furthermore, the lack of any significant hazard was also consistent among the different types of participant, and there was no suggestion of any adverse effects emerging with more prolonged follow-up.
Consistent beneficial effects of cholesterol lowering on vascular mortality and morbidity
-----------------------------------------------------------------------------------------
The prespecified analyses for assessing the benefits of statin allocation in different types of participant were to be of major vascular and coronary events. Because much larger numbers suffered at least one such event, analyses of those outcomes can help interpret the observed effects on the smaller numbers of deaths from vascular causes. For example, the proportional reduction in vascular mortality observed among the 5082 participating women was very similar to that among the men. But since fewer women took part and their absolute risk of vascular death was somewhat lower, this result did not reach conventional levels of statistical significance on its own (p = 0.08). Even so, the very definite and highly significant reductions in major vascular events observed among the participating women (figure [6](#F6){ref-type="fig"}) indicate that the reduction in vascular death, and hence in all-cause mortality, among the women is real \[[@B36],[@B37]\]. Similarly, the highly significant reductions in major vascular events in the elderly and those with below average cholesterol concentrations at baseline reinforce the consistent reductions in vascular mortality of about one-sixth in these groups. Given the two-thirds compliance during HPS, actual use of 40 mg simvastatin daily in this population would lower LDL cholesterol by about 1.5 mmol/L (57 mg/dL), and this would probably reduce the vascular death rate by about one quarter. Furthermore, the continued divergence in the vascular death rate during successive years of follow-up (reinforced by the similar effect on major vascular events) suggests that more prolonged statin therapy would produce even larger absolute reductions in vascular mortality.
Despite the proven benefits of statins in people with coronary heart disease, it has been suggested that statins may have adverse effects among people with overt heart failure \[[@B31]\]. Only small numbers of participants died from heart failure during HPS, but simvastatin allocation was associated with a marginally significant reduction in heart failure hospitalisations or deaths during the treatment period. Patients with occlusive arterial disease or diabetes who had coexistent heart failure were still eligible for the study provided they were not breathless at rest. Although the presence of heart failure at study entry was not routinely recorded, ACE inhibitors were chiefly used for heart failure or hypertension during the recruitment period (1994--7), which predated evidence from the HOPE trial of benefit in other circumstances \[[@B38]\]. Among participants using ACE inhibitors at baseline, many of whom were likely to have had heart failure, simvastatin allocation significantly reduced the risk of major vascular events (495 \[24.9%\] simvastatin vs 568 \[28.5%\] placebo; RR 0.84 \[0.75--0.95\]; p = 0.006), and there was a non-significant trend towards fewer vascular deaths (265 \[13.3%\] vs 285 \[14.3%\]; RR 0.93 \[0.78--1.09\]; p = 0.4). These results are not consistent with any substantial adverse effect of statin therapy on heart failure, and suggest that the beneficial effect on major vascular events is likely to outweigh any small adverse effect that might exist. Subsequent analyses of HPS based on baseline blood levels of brain natriuretic peptide (which is a sensitive and validated marker of heart failure \[[@B39]\]) should help further to assess the effects of statin treatment in patients with heart failure (as will on-going trials in such individuals \[[@B40]\]).
The significant 13% (SE 4) reduction in all-cause mortality observed in HPS reflects the combined impact in this high risk population of the significant 17% (4) reduction in vascular mortality produced by lowering LDL cholesterol by 1 mmol/L and the lack of any significant effect on deaths from non-vascular causes during 5 years of treatment. Since the proportion of vascular to non-vascular deaths will differ between populations, the observed reduction in all-cause mortality is not readily generalisable to other situations, whereas the reduction in vascular mortality may well be. Direct assessment of the effect of cholesterol lowering on total mortality could also obscure potentially important differences in the effects on cause-specific mortality in particular circumstances \[[@B41]\]. Consequently, the separate analyses of vascular and non-vascular mortality (and morbidity) presented here provide both more sensitive and more generalisable evidence, not only of any beneficial effects of cholesterol-lowering statin treatment but also of any hazards in particular circumstances.
No evidence of adverse effects of cholesterol-lowering on non-vascular mortality or morbidity
---------------------------------------------------------------------------------------------
HPS involved much larger numbers of non-vascular deaths, cancers and other serious non-vascular outcomes than any previous cholesterol-lowering study, as well as including large numbers of participants with relatively low cholesterol concentrations. It can, therefore, help address remaining uncertainty as to whether associations in non-randomised observational studies between lower total cholesterol concentrations (e.g. \<4 mmol/L) and higher rates of particular non-vascular conditions are causal or, instead, due to confounding or reverse causation \[[@B3],[@B4]\]. Reassuringly, among the 20,000 participants in HPS, lowering total cholesterol by an average of 1.2 mmol/L (46 mg/dL) for 5 years did not appear to have any adverse effect on non-vascular mortality or morbidity. Although there were too few deaths from some particular causes to assess the effects of cholesterol-lowering directly, the results for relevant non-fatal outcomes were of help in overcoming these limitations. For example, despite inverse associations in observational studies between cholesterol concentrations and respiratory mortality \[[@B1],[@B2]\], the randomised evidence from HPS does not indicate any adverse effect of lowering cholesterol on either respiratory mortality or on the much larger number of hospitalisations for fatal or non-fatal respiratory illness (or, indeed, on lung function). Similarly, the apparent lack of any adverse effect in HPS for deaths from cancer, haemorrhagic stroke, liver or renal disease is reinforced by the results for the larger numbers of related non-fatal outcomes.
Some non-randomised observational studies have found lower cholesterol concentrations to be associated with higher mortality in the elderly \[[@B32],[@B33]\], and there has been uncertainty about the effectiveness of cholesterol-lowering in older people \[[@B37]\]. In HPS, however, allocation to cholesterol-lowering statin therapy not only reduced vascular (and total) mortality among 5806 participants aged 70 or over at presentation, but there was also no apparent increase in non-vascular mortality. More recently, a randomised trial reported a marginally significant excess with statin allocation among the 444 patients aged 70--82 at presentation who developed cancer during 3 years of treatment \[[@B34]\]. But that result may well represent a chance finding, since no significant excess was seen with statin allocation among the 674 HPS participants aged at least 70 at presentation who developed cancer during 5 years follow-up, nor has any such excess been reported from the other large statin trials \[[@B23]\]. Chance may also explain the slight (but non-significant) excess of non-melanoma skin cancers observed during HPS, since that excess was largely confined to the early years of the study, was seen with both squamous and basal cell carcinomas, and is not supported by published data from other major statin trials excluding HPS (724 \[4.2%\] statin vs 699 \[4.1%\] placebo; odds ratio 1.04; 95% CI 0.93--1.15; p = 0.5) \[[@B12],[@B17],[@B22],[@B42]\].
If lowering cholesterol really did have adverse effects on non-vascular mortality or morbidity \[[@B43]\] then this might be more apparent either among participants who entered HPS with below average cholesterol levels (and so had their cholesterol concentration reduced to very low levels) or after more prolonged exposure to treatment during the later years of follow-up. But even among the 4072 participants whose pretreatment total cholesterol was below 5.0 mmol/L, lowering total cholesterol to an average of 3.5 mmol/L (133 mg/dL) for 5 years was not associated with any excess of non-vascular mortality or morbidity. Nor did any adverse effects begin to emerge with more prolonged treatment and follow-up during the trial. Nevertheless, cancer and other risks may take many years to become manifest, and extended follow-up for mortality and morbidity in HPS (as well as in some other statin trials \[[@B44]\]) will help assess any longer term effects.
Efficacy and safety of large cholesterol reductions
---------------------------------------------------
As might be expected from the approximately log-linear association in observational studies between vascular disease risk and cholesterol concentrations \[[@B2],[@B5]\], a 1 mmol/L LDL cholesterol reduction in HPS from about 4 mmol/L to about 3 mmol/L (i.e. about 155 to about 116 mg/dL) reduced the risks of vascular death by about one sixth and of other major vascular events by about one-quarter, and so too did reducing it from about 3 mmol/L to about 2 mmol/L (i.e. about 116 to about 77 mg/dL), without any evidence of adverse effects. Recently, a randomised study of more intensive statin treatment versus a standard regimen in 4162 patients found that lowering LDL cholesterol to 1.6 mmol/L \[62 mg/dL\] rather than to 2.5 mmol/L \[95 mg/dL\] for about 18 months produced a 16% (95% CI 5--26) reduction in major vascular events \[[@B45]\]. These findings indicate that any thresholds below which lowering LDL cholesterol does not safely reduce vascular disease risk are at much lower concentrations (e.g. below 2 mmol/L \[77 mg/dL\] of LDL cholesterol or 3.5 mmol/L \[135 mg/dL\] of total cholesterol) than are typically seen in Western populations. Several large-scale trials that are currently assessing more intensive statin regimens \[[@B46]-[@B48]\] will provide further information as to whether even more substantial cholesterol reductions are not only effective at lowering vascular disease risk but also safe.
Conclusion
==========
Based on large numbers of deaths and other relevant outcomes, the present results show that lowering LDL cholesterol by an average of 1 mmol/L produces substantial reductions in vascular (and, hence, all-cause) mortality in a wide range of individuals at increased risk of occlusive arterial disease (as well as reducing their risks of heart attacks, strokes and revascularisation procedures). These results also provide considerable reassurance that lowering total cholesterol concentrations by more than 1 mmol/L for an average of 5 years does not produce adverse effects on non-vascular mortality or cancer incidence, even among those who had their cholesterol concentrations reduced to very low levels. Indirectly, this observation provides some reassurance about the likely efficacy and safety of the more intensive cholesterol-lowering achievable with higher-dose or newer statins, and with combinations of standard statin doses and drugs acting through other pathways (such as resins or cholesterol absorption inhibitors). The present results provide further evidence of the benefits and safety of using statin therapy routinely in anyone (irrespective of their initial cholesterol concentration or other factors, such as age or gender) in whom a reduction in their vascular disease risk of about one third would be considered worthwhile.
Competing interests
===================
The Clinical Trial Service Unit has a staff policy of not accepting honoraria or other payments from the pharmaceutical industry, except for the reimbursement of costs to participate in scientific meetings. Coordinating centre members of the writing committee (J Armitage, R Collins, L Bowman, S Parish, R Peto) have, therefore, only had such costs reimbursed. P Sleight has received honoraria as well as such reimbursement of costs.
Authors\' contributions
=======================
MRC/BHF Heart Protection Study Collaborative Group
*Writing Committee*-Jane Armitage, Rory Collins, Louise Bowman, Sarah Parish, Peter Sleight and Richard Peto.
JA was involved in data collection, analysis and interpretation, and drafted the manuscript; RC conceived and designed the study and was involved in data collection, analysis and interpretation, and drafting of the manuscript; LB was involved in data collection and drafting of the manuscript; SP was involved in data collection, analysis, validation and interpretation; RP conceived and designed the study, and was involved in analysis and interpretation; PS was involved in the study design and interpretation. All authors provided comments on and approved the final manuscript.
*Steering Committee*-R Collins (principal investigator), T Meade (chairman), P Sleight (vice-chairman), J Armitage (clinical coordinator), S Parish and R Peto (statisticians), L Youngman (laboratory director), M Buxton, D de Bono (deceased), C George, J Fuller, A Keech, A Mansfield, B Pentecost, D Simpson, C Warlow; J McNamara and L O\'Toole (MRC observers).
*Data Monitoring Committee*-R Doll (chairman), L Wilhelmsen (vice-chairman), K M Fox, C Hill, P Sandercock.
*Collaborators*- *(doctors; nurses; receptionists): Aberdeen Royal:*N Benjamin, J Webster; J Jamieson; L Donald. *Bassetlaw Hospital:*R Blandford; L Carrington, H McMahon; D Cheetham. *Royal United, Bath:*J Reckless; L Brice, R Carpenter, J Christmas; C Flower. *Bedford:*I Cooper; S Frampton, E Pickerell; J Wells. *Belfast City:*M Scott; V Crowe, A Shaw; L Shannon. *Birmingham City:*S Jones; G Faulkner, A Lavery, H O\'Leary, R Watson; C Capewell, S Hughes. *Birmingham Heartlands:*S Bain, A Jones; G Holmes, C Jewkes; T Bellamy, P Harrison. *Queen Elizabeth, Birmingham:*N Buller; J Hooks, H Jones, E Smith, P Vint, R Watson; P Crook, J Williams. *Bishop Auckland General:*M Bateson; P Cawley, P Gill; L Hawkeswell, K Simpson. *Royal Bournemouth:*M Armitage; C Cope, J Tricksey, M Wilson; S Cottrell. *Princess of Wales, Bridgend:*C Jones; M Llewellyn, P Smith; T Woodsford. *Royal Sussex County, Brighton:*R Vincent; E Joyce, N Skipper; P Peters. *Bristol Royal Infirmary:*M Lemon (late), D Stansbie; A Hagos Kidan, M Halestap; A Gibbons, J Meredith. *Frenchay, Bristol:*C Dawkins, M Papouchado; L Baker, K Boulton, C Dawe; A Lewis, J Wisby. *Addenbrooke\'s, Cambridge:*M Brown; J Emeny, W Smith, D Thurston, D Trutwein; M Cornwell, D Lloyd. *Castleford & Normanton:*C White; M Hudson, M Khalifa, N MacKereth, J Woolford; G Martin. *St Peter\'s, Chertsey:*M Baxter; R Chambers, S Glenn, J Kerr; G Golesworthy, A Watts. *Corby Community:*G Baines; J Groom, L Price; I Barlow. *Leighton, Crewe:*S Mallya; J Maiden, M Nash; V Lowe. *Derbyshire Royal Infirmary:*M Millar-Craig, A Scott; S Cozens, J Hannah, M Hinwood, S. Hopcroft, M Margetts, H Waterhouse; J Millward. *Darlington Memorial:*J Murphy; M Charters, B Graham; M Banks, M Boon, C Cassidy, R Nobbs, *Dewsbury District:*T Kemp; P Turner; S Sheldrake. *Russells Hall, Dudley:*M Labib; R Pearson, J Sidaway; P Davies, M Hodgkiss. *Queen Margaret, Dunfermline:*D MacLeod; R Stuart; J Albrock, J Fisher, F Stuart. *Edinburgh Royal Infirmary:*C Swainson; S Glenn, J Johnston, S Sadler; M Curren, S Feirnie, L Stenhouse. *Western General, Edinburgh:*R Lindley, C Warlow; A Kenny, F Waddell; M Brownlie, I Guilar. *Derriford, Plymouth:*A Marshall, J Went; S Clarke, A Inman, J Simmonds; B Duook, G Mortimore, A Pascoe. *Glasgow Royal Infirmary:*S Cobbe; C Campbell, H Young; M Keeble. *James Paget, Great Yarmouth:*S Absalom; L Baillie, N Bracey; L Falco, D Stone. *Hartlepool General:*G Tildesley; B Carr, G Longstaff, A Turner, H Wilkinson; S Wilkinson. *Hillingdon:*R Hillson; D Brookes, B Capper, N. Mahabir, K Price; V Badrick. *Huddersfield Royal Infirmary:*H Griffiths; J Fitzgerald, S Lewis; P Campbell. *Kettering General:*G Baines; J Cullen, G Claypole, J Lomas; A Rogers. *Royal Lancaster Infirmary:*A Brown; J Cheshire; J Rowley. *Leeds General Infirmary:*S Ball, C Prentice, A Hall; P Atha, K Caffrey, W Currie, K Drury, C Hague, S Hall, P Maguire, C Rose, R Watson; A Buxton, A Wedgwood. *St James University, Leeds:*S Gilbey; W Currie, K Drury, S Hall, C Rose, J Wilson; M Vaughan. *Walton Centre, Liverpool:*P Humphrey; J Blocksage, R McSloy, K Ost, L Owen, S Saminaden, D Watling, J Wiseman; J Davies. *Ealing General, London:*A Kehely, J Kooner; I Corbett, J Peters, K Price, S Trainor; M Van Goethem. *Guy\'s & St Thomas\', London:*J Chambers; M Crawshaw, A Jones, J O\'Sullivan, S Powell, M Reoch, J Sanders; M-F Beament, B Fangrad, Y Williams. *North Middlesex, London:*S Banim, T Crake; B Ford, V Glynn; S Ismail. *Royal Brompton, London:*N Buller, A Coats; L Aitken, E Cruddas, K Serup-Hansen; D Nosworthy, N Reilly. *Whittington, London:*S Coppack, J Malone-Lee; P Clifton, A Holmes; L Camplin. *Luton & Dunstable:*D Peterson, C Travill; S Gent, A Hunter, C Stroud; K Griffiths. *Macclesfield District General:*E Davies; M Mason, A Robinson; S Belfield. *Maidstone:*J Chambers; L Bispham, J Massey, A Mercer, J Sheppard; S Burrage. *Manchester Diabetes Centre:*K Cruickshank; KL Chan, V Wharfe, J Woodward; F Alexander, Y Williams. *Manchester Royal Infirmary:*M Walker; P Campbell, J Day, S Edwards, B Kelly, P Nicholson; S Barrett, S Gleeson. *North Manchester General:*M Savage, J Swan; D McSorland, Gillian Thompson, C Waywell; C O\'Neill, L Wharton. *Royal Victoria Infirmary, Newcastle-upon-Tyne:*P Adams, R Lindley, N Cartilidge; M Mace, M Thompson; J Hulmes. *Radcliffe Infirmary/John Radcliffe, Oxford:*J Armitage, R Collins, P Sleight; S Beebe, M Campbell, J Godden, S Goodwin, A Lawson, H Lochhead, P Whitbread; S Knight, A Taylor, S Turner; *Royal Alexandra, Paisley:*I Findlay; C Campbell, J Hunter, H Young; E McNally. *Whiteabbey, Newtownabbey:*P Crowe; V Crowe, B Hunter, A Shaw; L Shannon. *North Tyneside Health Care Centre, North Shields:*R Curless, R Lindley, P McKenna, S Roberts; A Black, J Martin; M Burt. *Northampton General:*J O\'Donnell; T Burdett, S Marsh, J Woodward; R O\'Hare, C Owen. *Poole General:*A McLeod; M Richardson; C Reeves. *Halton General, Runcorn:*R Mallya; J Forshaw, J Hodson; H Lenden, G Osborn. *St Helier:*J Barron; A Ballard, B Docherty, M McDonnell, S Ritson, D Tyler; S Carter, C Rigney. *Conquest, St Leonard\'s-on-Sea:*R Wray; K Gaughan, J Sinclair; J Burleigh, J MacDonald. *Royal Hallamshire, Sheffield:*G Venables; C Doyle, M Fox, L Mundey; D Thompson, S Rowley. *King\'s Mill Centre, Sutton-in-Ashfield:*R Lloyd-Mostyn; D Bailey, I McKenzie; R Bamford. *Singleton, Swansea:*P Thomas; R Thomas; C Alexander, R Chohan, K Wood. *Princess Royal, Telford:*N Capps; D Donaldson, C Stiles, L Tonks; S Crank. *Manor, Walsall:*A Cunnington, P Giles; N Groves, E Walton; W Dance. *Watford General:*M Clements; C Feben, A Hunter, E Walker; L Atkins, R Kaiser, R Williats. *Sandwell District General, West Bromwich:*E Hughes; J Elson-Whitaker, S Sumara, C. Verow; G Banks, R Glover, K Hall. *Worcester Royal Infirmary:*A Munro, C Pycock, D Tibbutt; J Cadwell, M Greenwood; M Betts. *Worthing:*M Signy; E Joyce, C Wrapson; G McCourt, R Moore. *Wycombe General:*S Price, R Regan; M Aldersley; P Pendry. *Coordinating Centre administration and computing, Clinical Trial Service Unit, University of Oxford:*J Barton, C Bray, and K Jayne (administrative coordinators), V Booker, H Bojowsky, R Brooker, M Corbett, J Crowther, A Grantham, C Harwood, D Haywood, J Heineman, C Hope, C Indge, R Jones, S Jones, R Kanahan, K Kidney, M King, S Knight, H Lang, C Mardsen, C Mathews, G Meade (deceased), H Monaghan, K Murphy, A Naughton, A Owers, A Peto, S Pickworth, G Pocklington, A Radley, S Southren, K Szumczyk, R Tong, E Wincott; Clinical support and outcome adjudication: J Armitage and R Collins (study coordinators), A Keech and S MacMahon (piloting and planning), C Baigent, L Bowman, K Burbury, Z Chen, R Clarke, S Dunachie, V Frigi, M Landray, E Lau, C Sudlow, C Turnbull; Statistics and computing: S Parish and R Peto (statisticians), P Harding, M Lay, and K Wallendszus (computing coordinators), N Bruce, A Charles, A Cody, N Goodwin, R Greenlaw, B Hauer, P McCabe, A Palmer, C Peto, A Rowe, S Wilson, A Young, A Young; *Coordinating Centre laboratory, Clinical Trial Service Unit, University of Oxford:*L Youngman (director), K Kourellias, S Clark, and M Radley (laboratory coordinators), K Bhamra, L Buckingham, M Bradley, T Chavagnon, B Chukwarah, C Colominas, S Crowley, K Emmens, S Edwards, J Gordon, J Hill, A Lee, C Lennon, M McAteer, N Miller, S Norris, H Priestley, J Taylor, J Wintour, M Yeung; *Nurse Monitors and trainers:*S Beebe, M Campbell, J Fitzgerald, J Godden, A Lawson, S Lewis, H Lochhead, M McDonnell, M Nash, P Whitbread.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1741-7015/3/6/prepub>
Acknowledgements
================
The most important acknowledgment is to the participants in the study, and to the doctors, nurses and administrative staff in hospitals and general practices throughout the United Kingdom who assisted with its conduct. The study was funded by the UK Medical Research Council, the British Heart Foundation, Merck & Co (manufacturers of simvastatin) and Roche Vitamins Ltd (manufacturers of the vitamins).
|
PubMed Central
|
2024-06-05T03:55:55.322858
|
2005-3-16
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079844/",
"journal": "BMC Med. 2005 Mar 16; 3:6",
"authors": []
}
|
PMC1079845
|
Background
==========
Accurate diagnosis is essential for both the clinical care of patients and research about disease conditions. Clinicians increasingly use online access to evidence in the course of clinical care as well as for continuing education and research \[[@B1]\]. For most clinicians and researchers the current best evidence published in health care journals is usually first widely accessible through major biomedical databases such as MEDLINE and EMBASE. However, information retrieval in these databases can be problematic due to the scatter of relevant articles across a broad array of journals, the very dilute concentration of high quality, relevant studies in a very large database, and the inherent limitations of indexing in any large bibliographic database, amplified by clinicians\' lack of search skills \[[@B2]\]. EMBASE searches, for example, take place in the milieu of over 9 million citations from over 4,600 journals with between 6,000 and 8,000 citations added weekly \[[@B3]\].
Researchers have developed search strategies to assist clinicians with searching, the majority of which have been developed for MEDLINE when searching for therapy and review articles \[[@B4]-[@B10]\]. More recently, search strategies have been reported for the retrieval of diagnostic studies in MEDLINE \[[@B11]-[@B14]\]. This is an important development because clinicians must be able to efficiently retrieve the increasing amount of innovation and new knowledge concerning diagnosis and the burgeoning number of validated treatments for specific conditions that are contained in these large electronic databases. Using search strategies or filters can assist clinicians with this retrieval. In addition to searching MEDLINE, clinicians may wish to search other electronic databases such as EMBASE to cover their topic of interest more comprehensively. EMBASE is complementary to MEDLINE in that EMBASE provides greater coverage of the European and non-English language publications and provides broader coverage in such areas as psychiatry and toxicology \[[@B3]\].
In the early 1990s, our group at McMaster University developed search filters for use by clinicians and researchers on a small subset of 10 journals and for 4 types of journal articles (therapy, diagnosis, prognosis and causation \[etiology\]) \[[@B15],[@B16]\]. This research was updated and expanded using data from 161 journals indexed in MEDLINE from the publishing year 2000 \[[@B17]-[@B20]\]. These search strategies have been adapted for use in the Clinical Queries interface of MEDLINE <http://www.ncbi.nlm.nih.gov/entrez/query/static/clinical.html>. Clinicians can easily access and use these search strategies by going to the Clinical Queries page in PubMed. We now report the extension of this research for EMBASE, including the information retrieval properties of single terms and combinations of terms for maximizing the sensitivity and specificity of identifying methodologically sound primary (original) studies on the diagnosis of health disorders. These search strategies will assist clinicians and researchers when searching for relevant, high-quality articles in EMBASE.
Methods
=======
We compared the retrieval performance of methodological search terms and phrases in EMBASE with a manual review of each article for each issue of 55 journal titles for the year 2000. Overall, research staff hand-searched 170 journal titles. These journals were chosen based on recommendations of clinicians and librarians, Science Citation Index Impact Factors provided by the Institute for Scientific Information, and the ongoing assessment of their yield of studies and reviews of scientific merit and clinical relevance for the disciplines of internal medicine, general medical practice, mental health, and general nursing practice (list of journals provided by the authors upon request). Of these 170 hand-searched journals, 135 were indexed in EMBASE. Search strategies were developed using a 55-journal subset chosen based on those journals that had the highest number of methodologically sound studies, that is, studies that clinicians should be using when making patient care decisions. This selection enriches the sample of target articles, improving the precision of estimates of search term performance and simplifying data processing, but is unlikely to bias the estimates of the sensitivity and specificity of search terms.
We compiled an initial list of search terms, including index terms and textwords from clinical studies. Input was then sought from clinicians and librarians in the United States and Canada through interviews of known searchers, and requests at meetings and conferences. We compiled a list of 5,385 terms of which 4,843 were unique and 3,524 returned results (list of terms tested provided by the authors upon request). Examples of the search terms tested are \'criterion standard\', \'cut point\', \'sensitivity\', and \'ROC curve\', all as textwords; \'diagnosis\', the index term, and the index term \'diagnostic test\', exploded (that is, including all of this term\'s indexing subheadings).
As part of a larger study \[[@B21]\], research staff performance was rigorously calibrated before reviewing the journals and inter-rater agreement for identifying the purpose of articles was 81% beyond chance (kappa statistic, 95% confidence interval (CI) 0.79 to 0.84). Inter-rater agreement for which articles met all methodological criteria was 89% (CI 78% to 99%) beyond chance \[[@B21]\]. Six research assistants then hand-searched all articles in each issue of the 55 journals and applied methodological criteria to determine whether the article was methodologically sound for evaluation of a diagnostic test. The methodological criteria applied for studies of diagnosis were as follows: Inclusion of a spectrum of participants; objective diagnostic (\"gold\") standard or current clinical standard for diagnosis; participants received both the new test and some form of the diagnostic standard; interpretation of diagnostic standard without knowledge of test result and vice versa; and analysis consistent with study design.
The proposed search strategies were treated as \"diagnostic tests\" for sound studies and the manual review of the literature was treated as the \"gold standard\". We determined the sensitivity, specificity, precision and accuracy of each single term and combinations of terms in EMBASE using an automated process. Borrowing from the concepts of diagnostic test evaluation and library science, sensitivity for a given topic is defined as the proportion of high quality articles for that topic that are retrieved; specificity is the proportion of low quality articles not retrieved; precision is the proportion of retrieved articles that are of high quality; and accuracy is the proportion of all articles that are correctly classified \[[@B22]\].
Individual search terms with sensitivity \> 25% and specificity \> 75% for a given purpose category were incorporated into the development of search strategies that included 2 or more terms. All combinations of terms used the Boolean OR, for example, \"predict.tw. OR specificity.tw.\". The Boolean AND was not used because this strategy invariably compromised sensitivity. For the development of multiple-term search strategies to optimize either sensitivity or specificity, we tested all 2-term search strategies with sensitivity at least 75% and specificity at least 50%. For optimizing accuracy, 2-term search strategies with accuracy \> 75% were considered for multiple-term development. In the development of diagnosis search filters, 6,574 search strategies were tested.
In addition to developing search strategies using the Boolean approach described above, we also evaluated the potential for improving performance using logistic regression. Two approaches were taken. First, we took the top performing Boolean search strategies and ORed additional terms to these base strategies using stepwise logistic regression. The level of significance for entering and removing search terms from the model was 0.05. Adding terms to the model stopped when the increase in the area under the ROC curve was \< 1%. Second, we developed search strategies from scratch with stepwise logistic regression using these same cut-off values. Both logistic regression approaches were compared with the Boolean approach to search strategy development when developing strategies for treatment articles and prognostic articles for MEDLINE. Treatment and prognosis were chosen because they represented the best and the worst cases for MEDLINE search strategy performance. For both purpose categories, the logistic regression approaches to developing search strategies did not improve performance compared with search strategies developed using the Boolean approach described above. Thus, for subsequent purpose categories, including diagnosis and databases, including EMBASE, the Boolean approach was used for search strategy development.
We also tested search strategies published by other researchers for detecting diagnosis studies.
Results
=======
Indexing information was downloaded from EMBASE for 27,769 articles from the 55 hand-searched journals. Of these, 433 were classified as diagnosis, of which 97 (22.4%) were methodologically sound. Search strategies were developed using all 27,769 articles. Thus, the strategies were tested for their ability to retrieve articles about high quality diagnosis studies from all other articles, including both low quality diagnosis studies and all non-diagnosis studies.
Table [1](#T1){ref-type="table"} shows the best single term for high-sensitivity, high-specificity, and best balance of sensitivity and specificity. The single term, \"di.fs.\" (Ovid syntax for diagnosis as a floating subheading) produced the best sensitivity of 91.8% while keeping specificity at 76.4%. Specificity was maximized at 98.2% using the single term \"specificity.tw.\", but this was achieved at the expense of sensitivity, 62.9%. The single term \"diagnos:.mp.\" (Ovid syntax for the appearance of \"diagnos:\" in any one of the title, abstract or subject headings), produced the optimal balance between sensitivity (89.7%) and specificity (84.7%).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Single Term with the Best Sensitivity, Best Specificity, and Best Optimization of Sensitivity and Specificity for Detecting Studies of Diagnosis in EMBASE in 2000. Values are percentages (95% confidence intervals).
:::
---------------------------------------------------------------------------------------------------------------------------------------------------------------
**Search term OVID search\*** **Sensitivity (n = 97)** **Specificity (n = 27672)** **Precision†** **Accuracy (n = 27769)**
------------------------------------------------------ -------------------------- ----------------------------- -------------------- --------------------------
**Best sensitivity**(keeping specificity ≥ 50%)\ 91.8 (86.3 to 97.2) 76.4 (75.9 to 76.9) 1.4 (1.1 to 1.6) 76.5 (76.0 to 77.0)
di.fs.
**Best specificity**(keeping sensitivity ≥ 50%)\ 62.9 (53.5 to 72.5) 98.2 (98.1 to 98.4) 11.0 (8.4 to 13.6) 98.1 (97.9 to 98.3)
specificity.tw.
**Best Optimization of Sensitivity & Specificity‡**\ 89.7 (83.6 to 95.7) 84.7 (84.3 to 85.2) 2.0 (1.6 to 2.4) 84.8 (84.3 to 85.2)
diagnos:.mp.
---------------------------------------------------------------------------------------------------------------------------------------------------------------
\*Search strategies are reported using Ovid\'s search engine syntax for EMBASE. †Denominator varies by row. ‡Based on the lowest possible absolute difference between sensitivity and specificity. di = diagnosis; fs = floating subheading; tw = textword (word or phrase appears in title or abstract); : = truncation; mp = multiple posting -- term appears in title, abstract, or subject heading. Sensitivity = the proportion of high quality articles for that topic that are retrieved; specificity = the proportion of low quality articles not retrieved; precision = the proportion of retrieved articles that are of high quality; accuracy = the proportion of all articles that are correctly classified.
:::
Combinations of terms with the best results for sensitivity, specificity and optimization of sensitivity and specificity are shown in Table [2](#T2){ref-type="table"}. Combinations of terms improved on single search term performance for sensitivity. The 3-term search strategy, \"di.fs. OR predict:.tw. OR specificity.tw.\", achieved a sensitivity of 100% with a specificity at 70.4%. The single term \"specificity.tw.\" had the highest specificity, outperforming all 2- and 3-term combinations. A 3-term combination resulted in the optimization strategy achieving slightly above 89% for both sensitivity and specificity (Table [2](#T2){ref-type="table"}).
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Combination of Terms with the Best Sensitivity, Best Specificity, and Best Optimization of Sensitivity and Specificity for Detecting Studies of Diagnosis in EMBASE in 2000. Values are percentages (95% confidence intervals).
:::
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Search Strategyn OVID search\*** **Sensitivity (n = 97)** **Specificity (n = 27672)** **Precision†** **Accuracy (n = 27769)**
----------------------------------------------------------------------- -------------------------- ----------------------------- -------------------- --------------------------
**Best Sensitivity**(keeping specificity ≥ 50%)\ 100.0 (100.0 to 100.0) 70.4 (69.8 to 70.9) 1.2 (0.9 to 1.4) 70.5 (69.9 to 71.0)
di.fs.\
OR predict:.tw.\
OR specificity.tw.
**Small drop in sensitivity with a substantive gain in specificity**\ 96.9 (93.5 to 100.0) 78.2 (77.7 to 78.7) 1.5 (1.2 to 1.8) 78.3 (77.8 to 78.8)
diagnos:.mp.\
OR predict:.tw.\
OR specificity.tw.
**Best Specificity**(keeping sensitivity ≥ 50%)\ 62.9 (53.5 to 72.5) 98.2 (98.1 to 98.4) 11.0 (8.4 to 13.6) 98.1 (97.9 to 98.3)
specificity.tw.
**Small drop in specificity with a substantive gain in sensitivity**\ 73.2 (64.4 to 82.0) 97.4 (97.2 to 97.5) 8.8 (6.9 to 10.8) 97.3 (97.1 to 97.5)
specificity.tw.\
OR accurac:.tw.
**Best Optimization of Sensitivity & Specificity‡**\ 89.7 (83.6 to 95.7) 91.6 (91.3 to 91.9) 3.3 (2.9 to 4.4) 91.6 (91.3 to 91.9)
sensitiv:.tw.\
OR diagnostic accuracy.sh.\
OR diagnostic.tw.
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
\*Search strategies are reported using Ovid\'s search engine syntax for EMBASE. †Denominator varies by row. ‡Based on the lowest possible absolute difference between sensitivity and specificity. di = diagnosis; fs = floating subheading; : = truncation; tw = textword (word or phrase appears in title or abstract); mp = multiple posting -- term appears in title, abstract, or subject heading; sh = subject heading. Sensitivity = the proportion of high quality articles for that topic that are retrieved; specificity = the proportion of low quality articles not retrieved; precision = the proportion of retrieved articles that are of high quality; accuracy = the proportion of all articles that are correctly classified.
:::
Slight modifications to the above-noted most sensitive and most specific search strategies led to some attractive trade-offs in sensitivity and specificity (Table [2](#T2){ref-type="table"}). For instance, by replacing \"di.fs\" with \"diagnos:.mp.\" in the most sensitive search strategy (\"diagnos:.mp. OR predict:.tw. OR specificity.tw.\") specificity increased (70.4% to 78.2%) at the price of a small decrease in sensitivity (100% to 96.9%). Additionally, by ORing \"accurac:.tw.\" to \"specificity.tw.\", to the most specific search strategy, sensitivity increased by 10.2% (62.9% to 73.2%) with a small decrease in specificity (98.2% to 97.4%).
Our search strategies were simpler and compared well with two previously published strategies by Bachmann and colleagues for retrieving diagnostic test studies from EMBASE \[[@B23]\]. The most sensitive search reported by Bachmann and colleagues, an 8-term strategy, had a sensitivity of 96.9% in our database compared with 100% for our 3-term strategy (difference 3.1%, 95% CI -0.8% to 8.7%) (Table [3](#T3){ref-type="table"}). The most specific search reported by Bachmann and colleagues, a 2-term strategy, had a specificity of 90.9% in our database, compared with 98.2% for our 1-term strategy, but our strategy was less sensitive (62.9 vs. 79.4, difference 16.5%, CI 3.8% to 28.9%). Unlike Bachmann\'s study, our study evaluated the methodological rigor of diagnosis studies, and thus the performance of search strategies compared here is for detecting methodologically sound diagnostic studies.
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Comparison of previously published search strategies with search strategies developed our database. Values are percentages.
:::
---------------------------------------------------------------------------------------------------------------------------------------------------------
**Search Strategy OVID search\*** **Sensitivity** **Specificity** **Precision** **Accuracy**
--------------------------------------------------------------------- ---------------------- -------------------- ------------------ --------------------
**Bachmann\'s most sensitive search \[22\]**\ 100.0 Not reported Not reported
sensitiv:.tw.\
OR detect:.tw.\
OR accura:.tw.\
OR specific:.tw.\
OR reliab:.tw.\
OR positive.tw.\
OR negative.tw.\
OR diagnos:.tw.
**Bachmann\'s strategy tested in our database** 96.9 72.3 1.2 72.4
**Our most sensitive search**\ 100.0 70.4 1.2 70.5
di.fs.\
OR predict:.tw.\
OR specificity.tw.
**Difference (our strategy -- Bachmann) 95% CI for the difference** 3.1 -0.8 to 8.7† -1.9 -2.7 to -1.2‡ 0 -1.9 -2.7 to -1.2‡
**Bachmann\'s most specific search**\ 73.7 Not reported Not reported
sensitiv:.tw.\
OR detect:.tw.
**Bachmann\'s strategy tested in our database** 79.4 90.9 3.0 90.8
**Our most specific search**\ 62.9 98.2 11.0 98.1
specificity.tw.
**Difference (our strategy -- Bachmann) 95% CI for the difference** -16.5 -28.9 to -3.8‡ 7.3 7.0 to 7.7‡ 8.0 5.6 to 11.0‡ 7.3 6.9 to 7.7‡
---------------------------------------------------------------------------------------------------------------------------------------------------------
\*Search strategies are reported using Ovid\'s search engine syntax for EMBASE. †Differences are not statistically significant. ‡Differences are statistically significant. : = truncation; tw = textword (word or phrase appears in title or abstract); di = diagnosis; fs = floating subheading. Sensitivity = the proportion of high quality articles for that topic that are retrieved; specificity = the proportion of low quality articles not retrieved; precision = the proportion of retrieved articles that are of high quality; accuracy = the proportion of all articles that are correctly classified.
:::
Discussion
==========
Our study documents search strategies for use by clinicians and researchers that can help discriminate relevant, high-quality studies from lower quality studies of the diagnosis of health disorders and articles that are not about diagnosis. Those interested in all sound articles on diagnosis, for example researchers conducting systematic reviews of diagnostic tests, will be best served by the most sensitive search. If systematic reviewers wish to include diagnostic test articles that fail the methodological criteria we set, they will still be well served by starting with this strategy: in addition to retrieving all sound studies, the suboptimal specificity (70.4%) of our most sensitive search strategy means the many lower quality diagnostic test studies will also be retrieved. Reviewers may then use additional means to ensure that all pertinent studies are retrieved. Those with little time on their hands who are looking for a few good articles on diagnosis, most likely clinicians, will probably be best served by the most specific strategies. Clinicians could further broaden their search by using the strategies that optimize sensitivity and specificity while minimizing the difference between the two as these strategies provide the best separation of \"hits\" (target citations) from \"false drops\" (undesired citations) but do so without regard for whether sensitivity and specificity are affected.
In all cases precision was low. This is the inevitable result of a low proportion of relevant studies for a given purpose in a very large, multipurpose database. This means that clinicians and researchers will continue to need to invest their time in discarding irrelevant retrievals. While low precision in searching can be of concern, the low values here should not be over-interpreted: we did not limit the searches by clinical content terms, as would usually be the case in clinical searches. Precision might be enhanced by combining search strategies in these tables with additional methodological terms using the Boolean \'AND NOT\', thereby reducing the possibility of retrieving studies of lower methodological quality; however, this may decrease the sensitivity of the searches. Precision might also be increased by combining search strategies with content specific terms (e.g., \"diabetes\") or journal subsets using the Boolean \'AND\' thus reducing the volume of literature searched. The next phases of our project will focus on finding better search strategies through using more sophisticated strategies as outlined above.
Comparing the diagnostic search strategies developed for EMBASE with those that we developed for MEDLINE \[[@B19]\], we found that the single term \"specificity.tw.\" was the top performer for specificity in both databases and that this term outperformed 2- and 3-term strategies. Additionally, we found that textwords outperformed most index terms for sensitivity and specificity. The only index term that was a top performer was \"di.fs.\" or \"di.xs.\", which was the case for both databases. Although there are many differences between EMBASE and MEDLINE, some basic similarities are apparent, as just described.
Comparing our diagnostic search strategies developed for EMBASE with those previously published \[[@B23]\], our strategies had fewer terms and performed at least as well.
Conclusion
==========
Selected combinations of indexing terms and textwords can achieve high sensitivity or specificity in retrieving diagnosis studies cited in EMBASE. The reported search strategies will assist both clinicians and researchers when attempting the retrieve relevant, high-quality diagnostic articles.
Competing interests
===================
The author(s) declare that they have no competing interests.
Authors\' contributions
=======================
NLW and RBH prepared grant submissions in relation to this project. Both authors drafted, commented on and approved the final manuscript. Both authors also supplied intellectual content to the collection and analysis of the data. NLW participated in the data collection and both authors were involved in data analysis and staff supervision.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1741-7015/3/7/prepub>
Acknowledgements
================
This research was funded by the National Library of Medicine, USA. The Hedges Team includes Angela Eady, Brian Haynes, Susan Marks, Ann McKibbon, Doug Morgan, Cindy Walker-Dilks, Stephen Walter, Stephen Werre, Nancy Wilczynski, and Sharon Wong, all in the Department of Clinical Epidemiology and Biostatistics at McMaster University, Hamilton, Ontario, Canada.
|
PubMed Central
|
2024-06-05T03:55:55.328836
|
2005-3-29
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079845/",
"journal": "BMC Med. 2005 Mar 29; 3:7",
"authors": [
{
"first": "Nancy L",
"last": "Wilczynski"
},
{
"first": "R Brian",
"last": "Haynes"
}
]
}
|
PMC1079846
|
Background
==========
There is increasing emphasis on multi-disciplinary teams in modern clinical care \[[@B1]-[@B3]\]. Medical schools have attempted to embrace this through interdisciplinary learning \[[@B1],[@B4]-[@B6]\] and stressing relevant attitudinal, ethical and behavioural issues \[[@B7]-[@B9]\] in relation to teamwork.
Concern has been expressed about whether such teamwork skills can be effectively practised by junior doctors during the PRHO year while they cope with multiple professional, personal, physical and emotional demands \[[@B10]-[@B12]\]. The General Medical Council document, *The New Doctor*\[[@B13]\] stated that one of the professional learning outcomes of the pre-registration year is \'to work in a team and to take collective responsibility for patient care\'. Several authors \[[@B14]-[@B17]\] have recognised difficulties in delivery of these goals although most PRHOs appear to acquire the ability to work in a team by the time of qualification \[[@B18],[@B19]\]. More recent policy changes, such as The New Deal \[[@B20]\] might reduce the contribution of PRHOs to teamwork and decrease job satisfaction \[[@B21]\].
We wanted to explore how well our undergraduate curriculum, particularly the final year with 16 weeks shadowing in a student house officer role, prepared PRHOs for their new posts. In this study issues emerged about the stability of ward teams and the full integration of PRHOs in to teams. These findings may have relevance to the development of Foundation programmes.
Methods
=======
Two study cohorts of 33 PRHOs were selected by convenience (16 participants from year 2000/2001) and quota sampling (17 participants from year 2001/2002), stratified by gender, ethnicity and maturity, using the respective final year student population of the Medical School as the sampling frame (see Tables [1](#T1){ref-type="table"} &[2](#T2){ref-type="table"}). Written, informed consent was obtained from each participant. Semi-structured 1:1 interviews (see [Additional file 1](#S1){ref-type="supplementary-material"}) were carried out within the first three months of their PRHO posts. Students\' accounts were evaluated against the relevant professional competencies set out in *The New Doctor*\[[@B13]\] during a comprehensive evaluation of the newly developed final year in the medical school. This paper focuses on issues related to the role of the doctor within the health service, particularly \'the ability to work in a team\' and \'accept principles of collective responsibility\' \[[@B12]\]. All interviews were tape-recorded and transcribed in full. Two of the authors and an independent researcher read through the transcripts and identified a list of emerging themes. These lists of themes were then cross referenced and amended by the readers to reach agreement for the analytical framework of content analysis \[[@B22]\]. Throughout the data analysis the occurrence of key findings are noted by using simple counts. Combining qualitative and quantitative data allowed the authors to unpack, confirm and emphasise some similarities or differences between the different study cohorts. This approach assists in the generalisability of the findings \[[@B22]\] and offers evidence more convincingly (but not in a statistical significant sense) than relying on anecdotal accounts alone. This method can also contribute to the validation and credibility of qualitative research \[[@B22]-[@B24]\]. There was one key question in reference to team work but relevant information from all parts of the interview was used in the analysis (see Appendix 1).
::: {#T1 .table-wrap}
Table 1
::: {.caption}
######
Socio-demographic data from 16 PRHO cohort 2000/2001
:::
-------------------------------- -----------------------------------------------------------------------------------------------------------------------
**Gender** 11 Females, 5 Males
**Age** 24 years (mean)
**Ethnicity (self-described)** 5 Indian, 1 Gujarati, 1 Irish-Indian, 1 English-Chinese, 1 Sri-Lankan, 7 Caucasian
**Class** 13 middle class, 3 working class
**Family status** 15 single, 1 married
**Country of birth** 15 UK, 1 outside Europe
**Religion** 5 Christian, 3 Hindu, 5 none, 3 not stated
**Entry to Medical School** 11 after school, 2 after a gap year, 3 mature (3/16 obtained intercalated BA/BSc during their time at Medical School)
-------------------------------- -----------------------------------------------------------------------------------------------------------------------
:::
::: {#T2 .table-wrap}
Table 2
::: {.caption}
######
Socio-demographic data from 17 PRHO cohort 2001/2002
:::
-------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------
**Gender** 9 Females, 8 Males
**Age** 24 years (mean)
**Ethnicity (self-described)** 1 Black-African, 1 East Asian, 7 South Asian, 8 Caucasian
**Class** 14 middle class, 3 working class
**Family status** 16 single, 1 married
**Country of birth** 14 UK, 3 outside Europe
**Religion** 5 Christian, 3 Hindu, 1 Islam, 6 none, 2 not stated
**Entry to Medical School** 12 after school, 1 after a gap year, 1 after A level retake, 3 mature (10/17 obtained intercalated BA/BSc during their time at Medical School)
-------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------
:::
Results
=======
The 33 PRHOs worked in four acute hospital trusts and seven District General Hospitals (DGHs) in England, 16 in surgical and 17 in medical wards. They commented frequently on the importance of their clinical teams. Findings from PRHOs\' accounts are reported around 3 key themes: supportive environment, educational environment and organisational changes.
Overall the transition from being an \'outsider\' to a professional \'insider\' was a welcome change for all PRHOs, including involvement in the team. Comments on finding a role were common:
\'When I was a medical students in the 5^th^year I was very much aware that that I didn\'t have a role, I felt I was sort of very superfluous to the process\... \'Now I have, a definite, a clear cut role, and I know what I should be doing, and what I shouldn\'t be doing, and I have a continuity I have my patients and I have a team, which I have a good relationship with\'.
\'In the previous years we haven\'t been actually part of a team. We\'ve been assigned a team but you\'re not working with a team\...\.... you\'re not part of that patient\'s everyday care\'.
\'It\'s so different being a house officer than being a medical student -- you\'re a team, like, straightaway. The first day, that\'s it! That\'s a very big difference.\'
1) Supportive environment
-------------------------
Most PRHOs (27/33) described their clinical teams in positive terms (see Table [3](#T3){ref-type="table"}). Even so they also identified a number of specific characteristics, which had a profound influence on their first weeks working as qualified doctors.
::: {#T3 .table-wrap}
Table 3
::: {.caption}
######
Descriptions by 27/33 PRHOs about their medical and surgical ward teams
:::
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
\'The team is fine\'; \'the team is working brilliantly, we gelled really quickly\'; \'definitely working well\' (x2); \'working well together sometimes not so well\'; \'definitely I feel part of the team\' (x5); \'the team is great now, but not in the first two weeks\'; \'felt part of the team quickly, very quickly\'; \' feel part of the team, very much so\'(x2); \'it\'s a really nice team\'(x2); \'there is a good team spirit\'; \'we work as a team\' (x2); \'I really like my team, they are very supportive\'; our team functions very well, everybody is friendly, chats and makes jokes\'; \'there has been a welcome surprise, such as the team, it has been very supportive\'; \'and I got a really nice team\'; \'I work well with the team and felt accepted\'; \'definitely I am part of the team, you\'re a sort of team player really\'; \'oh yeah I get on really well with them\'.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
:::
### Constructive feedback from senior colleagues
PRHOs reporting support by senior medical staff (24/33) felt that they integrated with their clinical teams early and well, especially in emergency situations. Acute or life threatening events affected them emotionally, usually with little time for debriefing. Without senior support PRHOs often felt out of their depth. When they had dealt independently with a complex clinical situation, constructive feedback from senior staff added to their positive experiences. Conversely PRHOs who received no or little supervision or response from senior staff recorded these events as a negative impact on their confidence. The latter situations were more common in surgery, where PRHOs reported that senior staff was less accessible.
\'I would have quite liked an SHO \[Senior House Officer\] or somebody to tell me what I should be doing in that situation, or whether I should be phoning the GP or writing or speaking to him, or who actually should be talking to her \[patient\] and what actually telling them \[the couple\] \...ahm\...so I basically did it all myself.\'
\'\...getting called to see a man that had collapsed on his way back from the loo, and had no blood pressure and was basically dying. And I thought: Ahhh! But I just called the SHO and said: \"Can you come and help me?\" And I couldn\'t get access in him \[patient\] at all, and it was just a bit scary. When the SHO arrived we sorted him out and he was then fine after that, although he died a few days later\'.
### Sharing responsibilities and tasks
A sense of collective responsibility with other team members was positively expressed by most PRHOs (26/33). The predominant ethos was one of sharing the workload between medical and nursing colleagues in the team (see Table [4](#T4){ref-type="table"}).
::: {#T4 .table-wrap}
Table 4
::: {.caption}
######
Phrases used by 26/33 PRHOs about sharing work load and responsibilities
:::
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
\'jointly\' (x 2), \'share\' (x5) \'do the jobs\' (x8), \'help\' (x3), \'we do the same jobs\', \'split the jobs\', work generically as a team\', \'divide jobs up\', \'catch up with all the jobs\', \'we have jobs to do between the house officers and all the colleagues\', \'do all the work together\', \'work together closely\', \'plan our jobs\', \'carrying out the jobs together\', \'manage jobs together on the wards\', \'taking care of referrals\', \'do jobs as we go along\' \[during the ward round\], \'we finish of jobs\', we sort out our jobs\'.
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
\*some used more than one phrase
:::
PRHOs reported that sharing jobs was not always straightforward, particularly if individual team members did not work well together (7/33), or if teams were incomplete due to staff shortage and/or locum cover (8/33).
\'I mean it\'s \[work\] so erratic and our team is so disjointed! I don\'t really know half the patients and I don\'t\...I don\'t even know if half the jobs are getting done. I mean mine all get done but I don\'t know whether the other half\'s, the other House Officers are thinking I\'m doing theirs or they\'re doing mine, or the SHO\'s doing whatever\'.
### Interprofessional working
28/33 PRHOs reported that the relationship with nursing staff changed with qualification. Former inter-professional tensions seemed due to the perceived inconvenience that medical students caused to nurses. Prior to qualification students had relied on nurses in hospital and community for help with \'log book\' skills, such as catheterisation and injections, and they appreciated their patience and enthusiasm.
\'Before as a medical student you were a piece of furniture according to nurses\....but now, obviously you are the first-line if they want anything doing, I think in that respect there\'s a lot more regard\....and appreciation for what you do\'.
However, a number of inter-professional conflicts were identified such as adjustment to professional roles (x 15), discordant communication (x9), unclear what nurses are willing or (un)able to do (x10) and others.
### Relationship with PRHOs in the same team
8/33 commented positively on working with other PRHOs from their final year group in the same team, making sharing the workload and responsibility easier due to closer personal relationships. However, 2/33 identified this as impacting on their teamwork and patient care when PRHOs did not get on well together and 4/33 found competition to learn additional clinical skills.
\'There always has been camaraderie with other PRHOs, but at least I get to spend time around now as well, so ahm\...you know it is great, when you literally you say, will you do this, I will do that, and we meet up at the end \[of the day\]\....that is always really nice\'.
\'There\'s a mutual understanding there (amongst the PRHOs in the team\], and we\...amazingly, we are ever so professional on the job. I mean on the team, we\'re very good friends -- we go out for dinner, all sorts -- but when we\'re on the job, it\'s ever so professional\'.
2) Educational environment
--------------------------
PRHOs sought out educational opportunities to achieve a sense of progression after qualification within their clinical teams. Learning consisted of performing new practical skills, as well as observation, to acquire inter-personal and professional competencies. Continuity in their relationship with patients was perceived as important as team members, rather than temporary bystanders as students.
### Observing delicate situations
Most PRHOs (23/33) had witnessed delicate clinical or ethical situations, which required advanced communication skills between team members, patients and their families, e.g. breaking bad news, communicating with uncooperative patients. The interviewees had a variety of opportunities to observe how senior medical and nursing staff within their teams managed such situations sensitively and professionally.
\'\....it is quite interesting, watching your seniors how they deal with difficult patients: patients who are too demanding, or patients who, you know, are rude or angry, or breaking bad news, I try to sort of sit in on all that, so I can see how they \[seniors\] do it\....\'
\' \... when family gets involved, it gets awfully sad \[resuscitation status\]. I had a patient who died; I got to know the family relatively well -- two sons -- and they divulged that their mother had died, and\...all sorts of complex family dynamics going on there. They were quite clearly worried about their father. And then there\'s only so much input the house officer can do -- essentially, the seniors make the decisions\...\'
PRHOs were acutely aware that consultants were ultimately in charge of final decisions in ethical dilemmas. 6/33 PRHOs reported that they were given tasks which they felt were inappropriate, such as asking patients for consent to undergo an operation or investigation which they had never seen themselves.
### Learning new clinical skills from seniors
7/33 PRHOs were able to extend their practical skills, such as lumbar punctures or insertion of chest drains during their first house jobs. However, these opportunities were reduced where ward teams were incomplete and depended on the willingness and clinical competence of senior staff in their teams. Such limitations sometimes caused frustration. The message around unstable teams was often close to:
\'Our team\'s so disjointed at the moment. The SHO that we\'ve got is constant. But, she does \'on-calls\' for other teams as well so we don\'t see her that often. Our Registrars change every day because they\'re all locums. So on some days we don\'t have one and some days we do \...I don\'t feel\....I\'m not learning the things I thought I\'d learn.\'
\'There\'s only one regular member of the team and that\'s the staff grade, and so the consultant is a locum. He\'s been here a couple of months, or just before we started, basically he was here. The SHO is a locum SHO, who has now finished her rotation, and I\'ve just started so there was not much stability in the team to begin with.\'
3) Organisational changes
-------------------------
Organisational changes, such as new work patterns, and inadequate staffing levels caused frustration and anxiety affecting the PRHOs emotionally and their team work.
### Continuity of relationships with patients
Most PRHOs (25/33) described continuity of patient care in positive terms. They welcomed the fact that they could approach patients without apologies because they had a defined role and responsibility. 11/33 PRHOs reported fragmentation in their relationships with patients which they related to shift work and/or staff shortages.
\'\...the continuity is great, you know, walking on the ward, and sort of saying \"hi\" to people and you know they know you, and getting on with them and kind of trying to do your best for them.\'
\'\....there\'s no continuity of care: we start in the morning, we finish, and then someone else comes on -- and they \[consultant\] don\'t know the person. And then the next day that patient either moved to somewhere else or they\'ve gone home. So we\'re not following them \[patients\].\'
### New work patterns
One third of PRHOs in study cohort 2001/2002 confirmed that their hospital complied with new working hours while two thirds encountered difficulties.
The 2001/2002 group had a range of concerns: continuous long working hours, despite the implementation of new working patterns (12/17), loss of emergency experience at night (4/17), confusion from non-compliance or experimentation with new working hours (12/17), and unhappiness and ridicule by senior colleagues (5/17). These caused resentment and dissatisfaction amongst the majority of the cohort (12/17) more than the 2000/2001 group (5/16).
### Staffing levels
In addition to the new working arrangements, there was a change from 2000/01 to 2001/02 in the availability and continuity of permanent staff members, doctors and nurses. This was reported only by the second PRHO cohort. One third of this group commented that team work was disrupted and undermined by incomplete teams.
\'It\'s just working in a big hospital, and having an incomplete team was\...it was just unfortunate that that was the situation when I joined\...I didn\'t enjoy that and it upset me because, you know, it\'s taken me a long time to get through Medicine and I felt that that was bad.\'
Discussion
==========
In a few areas such as changes in rotas and hours the interviewees commented on the views of more senior doctors. These views may have influenced their responses. In order to encourage open responses interviewees were guaranteed anonymity. Interviews were conducted mainly by one interviewer, who was not part of the medical school hierarchy or trusts and had established a good rapport with the PRHOs cohorts. The interviews in this study ranged over a number of areas. They explored the preparation of PRHOs through their undergraduate curriculum and the sources of stress in the transition to the PRHO role. PRHOs discussed their experience of the working environment and their relationships with colleagues in a number of areas and team work was addressed specifically in a direct question. The themes explored in this study were identified clearly by the three authors and the independent researcher on analysis of the transcripts.
The PRHOs were enthusiastic about the way their assumption of their professional role allowed integration into multi-disciplinary working with a clear change in the working relationships with nursing colleagues. This was an area that changed from their medical school experience despite efforts to give them clearly defined patient management roles during 16 weeks of experience as student house officers in hospital.
It is possible that interprofessional education and even clearer integration of student house officers may help to minimise the changes on transition in future. Clear evidence about the efficacy of interprofessional education, however, has not yet been established.
The supportive and educational environments facilitated by ward teams received a welcome positive reaction from the majority of newly qualified PRHOs in both study cohorts. The importance of adequate senior supervision in the early stages of the PRHO experience has been emphasised in previous studies \[[@B25]\]. Opportunities to observe senior staff in clinical situations were especially useful.
The New Deal \[[@B20]\] and the New Doctor \[[@B13]\] were intended to improve working conditions for PRHOs. It has been suggested that introduction of \'new deal\' rotas may increase psychological morbidity and reduce job satisfaction for PRHOs \[[@B21]\], but that a well supervised working environment may counteract reductions in hours with high work intensity \[[@B26]\]. In this study difficulties with working patterns appeared to have increased between the two cohorts. Since this study was performed considerable further effort has gone in to the design of compliant and suitable rotas for doctors in training.
The problems were exacerbated by instability of the working teams which meant that supervision was not always adequate at these times when it needed to be most effective. Shortage of permanent staff in the ward teams was stressful and unsettling for one third of the PRHOs. These organisational difficulties affected five key areas of work: continuity of patient care, sharing responsibilities within the team, delayed integration in to the new teams, ad hoc learning opportunities to acquire new skills and ongoing lack of support from senior staff. These areas emerged without prompting from interviewers. It may be that PRHOs early in their post and low in confidence are worried by the possibility of inadequate support which might come from the unstable teams. However, the reports suggested that it was causing real concern for the interviewees who responded in this way.
It is important that these areas are addressed for the introduction of the Foundation programmes. It is possible that changes in working patterns may mean that those in F1 posts in the future will have to relate to larger teams with a broader range of skills with clear responsibilities but less opportunity for continuity of care. Instability in these larger teams could be even more of a problem. Both PRHOs and more senior doctors will have to learn such new ways of working. The loss of traditional clinical \"firms\" also needs to be addressed in undergraduate training. Many medical schools retain periods of apprenticeship, teaching traditionally and linking a small number of students to a small clinical team. This allows the students to relate to the staff and build their confidence. The combination and interaction of these teams and the increase in day and outpatient care means that new patterns have already been developed to optimise the students\' experience in the hospital environment. Changes in the student experience need to be developed in line with Foundation programmes to be sure that junior doctors are prepared for their roles on graduation.
The greatest satisfaction for PRHOs in this study came from involvement with a clear role and feeling part of the team, and this will be one of the core competencies assessed in Foundation programmes. Mechanisms of working within these teams and handover arrangements need to be clear. While the increased attention on evaluation should lead to clarity of roles with appropriate supervision, the greater emphasis on evaluation could also change the nature of the relationships.
Conclusion
==========
Overall this study emphasises the enthusiasm of PRHOs for well organised structures with adequate supervision in a supportive multidisciplinary environment. Over the period studied rotas and unstable staffing patterns affected this environment significantly.
Competing interests
===================
Dr. Mac Cochrane received funding from the Oak Foundation for both studies.
Authors\' contributions
=======================
HL developed study design, carried out most interviews, conducted the analyses and wrote the paper. MC supervised study, commented on study design, carried out few interviews, assisted in the data analysis. JR supervised the study, commented on study design, data analysis and contributed in the writing of paper.
Pre-publication history
=======================
The pre-publication history for this paper can be accessed here:
<http://www.biomedcentral.com/1472-6920/5/10/prepub>
Supplementary Material
======================
::: {.caption}
###### Additional File 1
Microsoft word file (teamwork additional.doc) containing details of the semi-structured interview.
:::
::: {.caption}
######
Click here for file
:::
Acknowledgements
================
We are grateful to all the students who participated in the study for their time and contributions during their first few weeks in post; Dr. Mary Seabrook, Dr. Sue Clarke and Dr. Paul Booton for their support and advice and Ms. Floss Chittenden for her unfailing support with the transcriptions of the interviews.
Formal permission to carry out both studies was obtained from the Education Committee of the Medical School where the study was undertaken. All relevant ethical conduct research procedures were adhered to and all interviewees gave written, informed consent.
|
PubMed Central
|
2024-06-05T03:55:55.331310
|
2005-3-9
|
{
"license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/",
"url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1079846/",
"journal": "BMC Med Educ. 2005 Mar 9; 5:10",
"authors": [
{
"first": "Heidi",
"last": "Lempp"
},
{
"first": "Mac",
"last": "Cochrane"
},
{
"first": "John",
"last": "Rees"
}
]
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.