text
string | predicted_class
string | confidence
float16 |
|---|---|---|
The cortices of mice at postnatal 2–5 days (P2–P5) were isolated and sheared into pieces, and digested by 0.25% Trypsin in Ca2+Mg2+-free PBS for 30 min at 37 °C. Trypsinized tissues were triturated and passed through 70 μm mesh. Dissociated cells were collected and washed by DMEM medium once before plating into a poly-l-lysine-coated dish (Φ100 mm) with 10 ml DMEM supplemented with 10% FBS. Culturing media were replaced completely with fresh DMEM plus 10% FBS the next day to remove the dead cells. Remained cells were cultured until reaching 100% confluence with media changed every 2–3 days. Primarily cultured astrocytes could be passaged once at 1 to 3 ratios, and the next generation of astrocytes could be collected when reaching the 100% confluence and frozen into liquid nitrogen. Over 95% of cells in the cultures were positive for GFAP and Acsbg1. When using for measurement of cell growth rate, astrocytes were seeded into a 24-well plate at 5×103 cells per well. Astrocytes in one of the 24 wells were trypsinized and collected every 3 days to count the cell number by hemocytometer. Averaged numbers from three independent experiments were obtained for each time point.
|
study
| 100.0 |
Apoptotic cells were examined with TUNEL assay (Yeasen) according to the manufacturer’s instructions. In brief, brain slices were digested for 10 min by proteinase K (20 μg/ml) at room temperature after IF staining. After washing twice with PBS, brain slices were incubated with equilibration buffer for 30 min at room temperature, and subsequently with Alexa Fluor 488-12-dUTP Labeling Mix for 60 min at 37 °C. After washing with PBS for three times, brain slices were stained with DAPI (2 μg/ml, Roche, Basel, Switzerland) before being mounted under coverslips.
|
other
| 55.06 |
Unless otherwise indicated, data were expressed as mean±s.d. from at least three independent experiments, and analyzed by paired t-test by comparing the data between littermates or the same batch of experiments. Statistical significance was considered when P is smaller than 0.05.
|
study
| 99.9 |
The effects of the 2007–08 financial crisis were strongly felt in Greece in the years that followed. In 2013, the economy entered the sixth year of recession, resulting in a substantial GDP decline. The main impact of the economic crisis was on the unemployment rate which rose eighteen percentage points from 9.6% (485,000 persons) in 2009 to 27.5% (1,330,000) in 2013 . The main share of jobless workers was from the private sector which amounted to 769,000 lost jobs in the years 2008–2012 compared to 89,000 in the public sector. At the same time, the sharp rise in the unemployment rate led to an equally sharp increase in poverty as the percentage of the population that was below the poverty threshold in 2012 increased to 38% . It is notable that in a survey from the Hellenic Statistical Authority (ELSTAT) the relative index of “people at-risk-of-poverty or social exclusion” in 2013 (after five years of austerity) was higher by 8.1 percentage points compared to 2009 (35.7% vs 27.6%) whereas the corresponding increment in Eurozone (EU-28) was only +1.3 pp (24.5% in 2013 vs 23.2% in 2009). In the same survey, the rate of “severe material deprivation” in 2013 was also shown to have climbed by 14.3 percentage points since 2009 (37.3% vs 23.0%).
|
other
| 70.44 |
The recession hit primarily the younger generation as was indicated by the share of young people in the unemployment index which amounted to almost 49%. The dramatic increase in youth unemployment and the 'scarring' effects of joblessness generated a large wave of human outflow from the country, mainly among educated and qualified people, scientists and other professionals in foreign countries. It is worth noting that according to the Athens Medical Association, there was a fivefold increase in the number of skilled Greek physicians who migrated abroad in 2012 compared to 2007 (1,166 vs 292 doctors respectively) [3–4]. The overall emigration showed an increasing trend and almost tripled from 2009 to 2013 (43,686 vs 117,094 people respectively) .
|
other
| 99.7 |
At the same time, an additional side effect of people’s inability to cover their insurance contributions because of unemployment and the undeclared work was the loss of their insurance coverage (and family dependents). It is worth noting that between 2008–2012, one out of three insured members in the two largest insurance organizations (IKA and OAEE) lost their health insurance eligibility . The large increment in the number of uninsured citizens has in turn resulted in limited or no access to medical care and pharmaceuticals exacerbating the inequalities in health care provision and increasing out-of-pocket expenses . It is notable that the share of household payments to public hospitals over the total household health expenditure rose substantially by 86% in the four-year period 2008–2012 (4.2% vs 7.8%) .
|
other
| 99.9 |
During the same period, there has also been a deterioration in the mental health of the population which has been attributed -directly or indirectly- to the economic crisis and high unemployment . The incidence of major depression increased by nearly five percentage points (from 3.3% to 8.2%), especially among young people . Other studies have recorded a 35% increase in the number of suicides (from 3.37 to 4.56 per 100,000 of the population between 2010 and 2012) , as well as in the number of people who had attempted suicide, with those who were experiencing financial difficulties to be in a particularly vulnerable group . The number of reported violent incidents also increased while the rate of homicide and theft cases almost doubled between 2007 and 2009 [12–13].
|
study
| 89.4 |
Together with the deterioration of mental health there was also evidence for the worsening of general health, particularly in vulnerable groups, as reported by Kentikelenis et al. . As for the perceived health status, the percentage of people who assessed their health at the level of "good" to "very good" decreased from 75.5% in 2009 to 74.1% in 2013, whereas the corresponding proportion for the assessment at the level of "bad" to "very bad" increased from 9.7% to 10.4% during the same period . A study from Zavras et al. verified relevant findings from previous studies which have shown that better levels of self-rated health are positively associated with income and education, the two variables that mostly characterize the socio-economic status of the population. At the same time the association was also found -as expected- to be negative with unemployment, the existence of chronic disease and age. In the same vein another study of Yfantopoulos et al. reported that the economic crisis has also led to an aggravated level of general and oral health in Greece. The study identified statistically significant oral health inequalities among the socio-economic groups with the aged, the less educated and those confronting financial difficulties to be associated with lower levels of oral health.
|
review
| 80.56 |
In the same period, an HIV epidemic outbreak was also recorded among heroin addicts owed mainly to budget cuts that led to the cancellation of many preventive programs (exchange of needles etc.) [12–13]. However, it should be noted that although drug use increased between the years 2007–2011, the per capita alcohol consumption in the same period decreased by 2.3% .
|
other
| 93.25 |
As far as mortality is concerned, the relevant data published by ELSTAT showed that the number of deaths increased from 108,316 in 2009 to 111,794 in 2013. It is notable that the 116,670 deaths that occurred in 2012 was the largest recorded number since 1949 whilst 2,000 of them (almost one third of the additional number of deaths) could be linked to austerity. However, it should be noted that the total number of dead and injured in road accidents has followed a continuously downward trend (20,097 in 2009 vs 16,054 in 2013) . The life expectancy in the same period continued an upward trend which had been recorded in the pre-crisis years. Specifically, according to ELSTAT , the relevant index increased from 77.7 years in 2009 to 78.3 years in 2013 for men and from 82.8 to 83.4 years for women, amounting to average annual increments of 0.27 and 0.20 years respectively.
|
study
| 99.94 |
In 2013, the GDP declined by almost one quarter compared to 2007 whilst in the same period, health care spending followed a parallel trajectory. It is noteworthy that in the years from 2005 to 2009, where GDP increased by 19.2%, the corresponding total health spending rose by 41% . The positive sign of the slope was reversed after the onset of the crisis. In the years between 2009–2012, the GDP declined by 16.4% which was accompanied by a 23.1% downturn in the total health care expenditure and 24.5% reduction in the public health spending (from 23.0 to 17.7 and from 16.1 to 12.0 billion € respectively) .
|
other
| 93.5 |
In terms of GDP percentage, total health care expenditure was increasing until it reached 9.9% of GDP in 2009 where it started declining to reach 9.1% of GDP in 2012. In the same period the GDP percentage of public health care expenditure was reduced from 7% in 2009 to 6.2% in 2012. Within this period and in order to fulfill loan conditions which require that public health care expenditures must not exceed 6% of GDP, The Ministry of Health had to go through a wide range of reforms aiming, among others, at the following goals: to achieve greater efficiency in all NHS services and, at the same time, to achieve a radical cut in expenses.
|
other
| 99.94 |
Specifically, the reform efforts were focused on two targets. The first was to merge neighboring public hospitals, as well as hospital clinics and departments, and the second to modernize the financing mechanism used by the unified, state-owned health insurer (EOPYY), by introducing a Greek version of Diagnostic Related Groups (DRGs ). Regarding the first reform, with the exception of 2–3 small hospitals which were integrated into larger units, hospital mergers were never implemented in Greece due to extensive political and social pressure by respective interest groups. As for the intended plan for EOPYY to finance public hospitals via the DRG system (known as KEN-DRGs in Greece), this is pending on the unlikely event that EOPYY can solve its inherent deficit problems and find sufficient resources to fulfill its intended purpose. Hence, public hospitals continue to be funded by The Ministry of Health through the state budget, which covers salaries and all current expenses. Apart from closed budgets, two horizontal measures were adopted by the Ministry to reduce public health care expenditures. One measure was an average 35% cutback in salaries since 2009, and the other an “until further notice” suspension of all public hiring. In light of these developments, the aim of this study was to investigate longitudinal efficiency of Greek public hospitals during the crisis period.
|
other
| 91.4 |
The Greek National Health System (NHS) includes 131 (general, special, university and university-affiliated) hospitals which admit more than 2.2 million patients per year. All NHS hospitals are geographically partitioned in seven administrative regional units. Each regional unit serves as the management link -at an intermediate level- between individual hospitals and the Ministry of Health. These hospitals constitute the backbone of the Greek NHS as they provide the bulk of secondary, tertiary as well as primary health care services to the majority of the population. Table 1 reports a summary of operational characteristics of Greek NHS hospitals. As it can be seen, there was a substantial increase in the demand for health care services from public hospitals as evidenced by the relevant increments in the total numbers of hospitalized cases (32.9%) and surgeries (9.2%) during the period 2009–2013. The increased demand for health care services in the public sector was due to a shift of patients from the private sector, mainly of lower income. In the 2009–2010 period, a decrease had been recorded in the total number of admissions in private hospitals by 25–30%, owed mainly to the inability of lower income groups to afford the relative high costs . In addition, increased demand for dental and obstetric services in public hospitals was also recorded, two areas that had been traditionally covered by private providers/hospitals up until then .
|
study
| 99.94 |
At the same time, there was a substantial gradual reduction in the average length of stay (ALoS) from 4.75 days in 2009 to 3.86 days in 2013 (i.e. -18.7%). The increased number of admissions combined with the reduced mean length of stay accounts for an overall reduction of 8% in the total number of inpatient days for the same five-year period. It is also interesting to note that there was an 8.5% reduction in the number of laboratory tests between 2009 and 2013 after an intermediate peak of 5.1% which occurred in 2011. It should also be taken into account that since the onset of the crisis, certain nongovernment organizations such as Médecins du Monde and Medecins Sans Frontieres, which had been providing health care services mainly to immigrants, started to cover additional groups of the population including the poor, the unemployed and the uninsured .
|
study
| 99.94 |
According to ELSTAT , the percentage of persons with self-declared unmet needs for medical examination or treatment (due to several reasons including financial barriers, long waiting times and traveling distances, lack of time etc.) increased from 4.1% in 2009 to 11.2% in 2013. Taking into account that all citizens could visit almost at no charge either general practitioners or outpatient clinics in hospitals, Kentikelenis et al. concluded that such reductions (during the first two years of the crisis) reflect most probably supply-side problems: hospital budgets were cut by 40%, many clinics were understaffed due to the suspension of public hiring, shortages in medical supplies were encountered rather frequently whilst patients in many cases had to bribe medical staff in order to be given priority especially in overloaded hospitals with large queues. It has also been reported that although physician visits for issues related to chronic diseases have been largely met, this was accomplished with an increase in out-of-pocket expenditures and cuts in family budgets . It is worth noting that the percentage of such illegal payments for health care services in Greece has been estimated to be more than 20% of the total private expenditure whilst the out-of-pocket payments in Greece was one of the highest among OECD countries .
|
study
| 99.9 |
As it can be seen in Table 1, there was a tremendous cut in hospital expenses in the period 2009–2013. Specifically, total expenses in 2013 were cut by almost 52% since 2009. A closer look at the evolution of expenses over time reveals that the sharpest decline occurred between 2011 and 2012. In particular, almost 35 out of the 52 percentage points of the total cut took place in the transition from 2011 to 2012.
|
other
| 88.1 |
Improving efficiency has become an increasingly important target for hospital managers. One of the most commonly used tools for efficiency measurement is Data Envelopment Analysis (DEA) . DEA is applied to a set of homogenous units -the so-called decision making units, DMUs- and seeks to maximize each unit’s efficiency as it is defined through the ratio of weighted sum of outputs over weighted sum of inputs. The concept behind the method is that it allows each unit to weigh production inputs and outputs in a way so as to achieve the maximum possible efficiency compared to the other units in the sample. Put differently, each unit is allowed, in essence, to consider its own production practice as best by weighting inputs and outputs in the most preferable way .
|
other
| 99.9 |
In mathematical terms, the afore-mentioned objective in conjunction with certain assumptions about the producible production points (i.e. efficient input-output combinations permitted by the technology) leads to the formulation of a fractional programming problem. The solution of an equivalent linear program identifies a set of units that are deemed as efficient and all other units are deemed as inefficient. Determination of fully efficient units (also known as best practice units) enables the construction of a piece-wise linear frontier, the so-called “best practice frontier” which isolates potentially efficient units (all points on the frontier) from inefficient ones (all productively attainable points surrounded –enveloped- by the frontier). All units residing on the frontier can be thought of as either the ones that produce a certain level of output using the lowest allowed amount of inputs or the ones than produce the highest attainable level of output using a certain amount of inputs. Thus, all these technically efficient units on the frontier are assigned an efficiency score of 1 (100%) whereas technically inefficient ones are assigned a positive score less than 1 (less than 100%). The percentage score of an inefficient unit can be derived in one of two alternative ways referred to as input-orientated and output-orientated efficiency scores. In input-orientation the score represents the maximum allowed equiproportionate (i.e. radial) reduction of its inputs that is still capable of producing the same level of output. Accordingly, in output-orientation the score reflects the maximum equiproportionate (radial) expansion of its outputs that can be produced using the same level of inputs. In geometric terms this can be thought of as the maximum allowed contraction or required expansion of the unit’s input/output position ray until the unit has reached the efficient frontier.
|
other
| 98.56 |
It is evident that DEA is a non-parametric method which is based solely on the observed input-output combinations of the units in the sample without any assumptions concerning the form of the production function. The term Data Envelopment Analysis (DEA) was introduced by Charnes et al. in their paper in 1978 which was based on the previous relevant research of Farell (1957) . The method has been used in numerous sectors, including healthcare in which its first applications date back to the1980s [26–27]. An advantage of the method is its ability to handle multiple inputs as well as multiple outputs, without the requirement of a common denominator of reference.
|
other
| 79.44 |
Formally, given a set of n DMUs (DMUr, r = 1,2, …p, …n) each one consuming m inputs (x1r, x2r, …, xmr) to produce s outputs (y1r, y2r, …, ysr), the input oriented efficiency θp* of unit p, is given by the solution of the following linear programming problem: θp*=max∑j=1sujpyjp−wp, p ∈ {1,2,…,n}.
|
other
| 99.9 |
Subject to: ∑i=1mvipxip=1 . ∑j=1sujpyjr−wp−∑i=1mvipxir≤0 ∀ r=1,2,…n. ujp≥ε, vip≥ε ∀ i,j. wp∈ℝ Where ε is a non-Archimedean infinitesimal value for forestalling weights ujp, vip to be zeroed. The formulation stated above refers to the BCC model which assumes variable returns to scale. Omitting wp in the above formulation the linear program is converted to what is known as CCR model which assumes constant returns to scale. When the BCC model is applied, the sign of wp that comes out from the solution of the linear program identifies the nature of returns to scale.
|
other
| 85.8 |
A drawback of DEA is that efficiency measures are defined relative to the best practice frontier of the sample under examination and consequently DMUs deemed as efficient are efficient only in relation to others in the particular sample . Therefore, it is not meaningful in general to compare the scores between two different samples as all calculations are based on different best practice frontiers whose differences are not known. Consequently, even the efficiency comparison of the same set of units in two different time periods is questionable. In order to overcome this subtle point of DEA that hampers the comparison of efficiency scores over time (i.e. efficiency changes), one could move from the so-called contemporaneous perspective, where a unique frontier is derived for each time period, to the so-called intertemporal perspective where a single common frontier which spans the whole period is defined . The basic idea within this latter framework is to regard each unit as if it were a different unit in each of the reporting periods. Thus, the performance of a unit in a particular period is compared with its own performance in other periods as well as with the performance of other units. Although, within this latter perspective, a year-to-year comparison can be carried out, one has to bear in mind that this approach implicitly assumes that there are no substantial technical changes over the entire time period. (i.e. the technological frontier is fixed). This assumption, however, cannot always be considered valid, especially when long time periods are analyzed, since production conditions may have substantially altered between distant years.
|
study
| 98.8 |
A compromise between contemporaneous and inter-temporal analyses is the so-called window analysis where DEA is applied successively on overlapping time periods of constant width (called a window). Once the window width has been specified all observations within it are viewed and examined in an inter-temporal manner referred to as locally inter-temporal analysis . The method was initially proposed by Charnes et al. in order to measure efficiency in cross sectional and time varying data. Furthermore, when window-DEA is applied, the number of observations taken into account is multiplied essentially by a factor equal to window’s width, which is quite useful when dealing with small sample sizes as it increases the discrimination capability of the method . Therefore, two factors should be reconciled when choosing window width. The window should be wide enough to incorporate the minimum number of DMUs for the required discrimination but it should also be narrow enough to ensure that technological change within it is negligible and therefore it will not allow misleading or unfair comparisons between DMUs belonging to distant apart time periods .
|
other
| 56.88 |
The assessment of productivity change over time together with its decomposition into efficiency changes and technology changes can be carried out using the so-called Malmquist index which was first introduced by Caves et al. based on an idea of Malmquist . Following Fare et al. the input-based adjacent Malmquist index between time periods t and t+1 is given by MPI(xt+1,yt+1,xt,yt)=[dCRSt(xt+1,yt+1)dCRSt(xt,yt) dCRSt+1(xt+1,yt+1)dCRSt+1(xt,yt)]1/2, Where dCRSt(xt,yt) and dCRSt+1(xt,yt) represent the distance functions of the production bundle (xt, yt) from the CRS technology frontiers in periods t and t+1 respectively whilst dCRSt(xt+1,yt+1) and dCRSt+1(xt+1,yt+1) represent the corresponding distance functions for the production bundle (xt+1, yt+1), i.e. dCRSτ(xφ,yφ)=max{θ>0:(xφ/θ,yφ) ∈ TCRSτ}, where TCRSτ={(xτ,yτ): ∑j=1nλjτxj,iτ≤xiτ , i=1,2,…m,∑j=1nλjτyj,rτ≥yrτ ,r=1,2,…,s, λjτ≥0, j=1,2,…,n}, for τ = {t, t +1} and φ = {t, t +1} and are equal to the reciprocal of the Farell measures of technical efficiency.
|
study
| 77.5 |
The first term is equal to the Farell technical efficiency measure at period t divided by the Farell technical efficiency measure at period t+1 and therefore reflects the efficiency change component in productivity change. In other words, the term indicates whether the hospital has moved closer to the CRS-frontier (i.e. catching-up to the frontier). The second term is equal to the geometric mean of the shifts in the CRS technology observed at the production bundles (xt, yt) and xt+1, yt+1 respectively and therefore reflects the technological change component in productivity change (i.e. shift in the frontier).
|
other
| 99.9 |
The component of technical efficiency change can be further decomposed into pure technical efficiency change and scale efficiency change. Assuming the variable returns to scale technologies TVRSt and TVRSt+1 for the two-time periods τ = {t, t+1}, i.e. TVRSτ={(xτ,yτ): ∑j=1nλjτxj,iτ≤xiτ , i=1,2,…m,∑j=1nλjτyj,rτ≥yrτ ,r=1,2,…,s,∑j=1nλjτ=1,λjτ≥0, j=1,2,…,n}, and the corresponding distance functions dVRSτ(xφ,yφ)=max{θ>0:(xφ/θ,yφ) ∈ TVRSτ}, where τ = {t, t +1} and φ = {t, t +1}
|
other
| 99.9 |
Thus, the Malmquist productivity index can be thought of as the product of three terms representing the changes attributed to pure technical efficiency, scale efficiency and technology. Taking the reciprocal of the indices calculated above , values greater than unity are meant to indicate progress whilst values smaller than unity indicate regress. It should be noted however that technological change can be calculated in a different way using the so-called base period Malmquist index introduced by Berg et al. .
|
other
| 99.9 |
One of the disadvantages of DEA is that statistical inference is difficult to be applied due to the implied assumption of the method that the whole distance of a DMU from the efficient frontier reflects solely its inefficiency. In reality, this distance reflects inefficiency as well as sampling variability and noise because input and output data are normally subject to errors. Furthermore, given the assumption that noise does not exist, the estimated empirical technology can only be a subset of the true but unknown technology, and therefore DEA scores will be upwards biased . In order to overcome this shortcoming of biased DEA scores due to sampling variability Simar and Wilson proposed a methodology, which is based on bootstrap techniques and allows determining the statistical properties of DEA estimators.
|
other
| 78.4 |
In the present study, their bootstrapping approach was applied in order to calculate the bias-corrected Malmquist indices together with estimations of their confidence intervals. Thus, the method allows for examination whether increases or decreases in productivity are significant in a statistical sense. In other words, the method allows for the conclusion of whether a result indicates real progress/regress or is a coincidence due to sampling variation.
|
study
| 99.9 |
The study examined the efficiency of a sample of 107 Greek NHS hospitals over the five-year period, 2009–2013. Twenty-four out of the 131 hospitals of the Greek NHS were excluded due to the idiosyncratic nature of health care services they provide [59–61]. More specifically, nine psychiatric, four anticancer/tumor, two dermatological, one maternity, one ophthalmological, one special diseases, one thoracic diseases, one pathological and four pediatric hospitals were excluded from the analysis in order to increase homogeneity of the remaining units. Going one step further towards increasing homogeneity and to permit their comparability, these 107 hospitals were additionally classified into four groups according to their size and the mixture/range of services they provide [59, 62–63]. Thus, under this classification the four groups A, B, C, D were assigned with NA = 21, NB = 33, NC = 30 and ND = 23 members respectively with corresponding sizes spanning within the ranges of <85, 85–190, 190–400, >400 beds. At this point, it should be noted that three hospitals could not be classified as belonging always to the same group throughout the whole five-year period of the study due to reductions/augmentations in their bed capacity. In order to maintain exactly the same members within each group, those three hospitals were considered to belong to the group into which they were assigned most of the time (i.e. 3 or 4 out of the 5 years).
|
study
| 100.0 |
All relevant data were collected from sources of Ministry of Health and Welfare. The data were given in (annual) summarized form and therefore no medical records or any other information concerning patients/caregivers/staff were used in the study. The study was also approved by the Ethical Committee of the Hellenic Open University
|
other
| 99.94 |
In DEA, the set of variables (inputs and outputs) that need to be included in the model should meet the following criteria: inputs should cover the full range of resources used; outputs should capture all activity levels and performance measures; furthermore, both, input and output variables, should constitute a set of factors common to all units under evaluation .
|
other
| 99.9 |
Thus, in accordance with published research [60,64–67] the actual variables selected for this study are among some of the most commonly used inputs and outputs affecting hospital efficiency. On the input side, labor and capital inputs were aggregated as follows: labor inputs were measured in terms of absolute numbers of staff (in full time equivalents- FTE) classified separately as physicians and other hospital employees. The separation was considered necessary since the former are fundamentally different as they enjoy the primary “decision rights” for patient care. Physicians are the most dominant and influential components in the entire production process, with little or no interference from management and it has been reported that their decisions, directly or indirectly, may eventually account for as much as 80–90% of the total health care expenditure in any system. Capital input, as in most relevant studies [60,64–67], was proxied by the number of hospital beds assuming that invested capital per bed is similar throughout the hospitals in the sample. This assumption, in the present study, is expected to be valid considering the high degree of homogeneity of the four groups in the whole sample.
|
study
| 100.0 |
As regards outputs, patients’ health gain is the ultimate measure of output against which hospital activity should be assessed. Since practical difficulties limit this outcomes approach , output is usually measured as an array of health care services that supposedly improve patients’ health. Thus, in this study the multiplicity of hospital services was aggregated into three main outputs: inpatient cases, surgeries and outpatient visits. As far as inpatient services are concerned the number of cases was chosen rather than the (alternative choice of) number of inpatient days in order to avoid distortions imposed by variations in average length of stay due to higher/lower occupancy rates . Furthermore, surgeries were counted separately as they constitute a fundamentally different part of inpatient services (compared to those of general medical) which usually account for a substantial volume of the total inpatient work. Thus, the total number of surgeries was included as a separate output variable in order to capture the workload of this special inpatient component. Finally, the number of outpatient visits was included as the third output variable representing the volume of outpatient services of the hospitals.
|
study
| 100.0 |
Finally, a 2-year window width was chosen and therefore four overlapping windows were analyzed over the 5-year study period. The narrow window width was decided since it provides the minimum common ground that allows the year-to-year efficiency comparisons without the possible distortion that would have been imposed if a wider one had been chosen. At the same time, the 2-year width is considered sufficiently large -as far as the discrimination of the method when applied on small groups is concerned—since, it doubles the number of DMUs in each window. It is evident that the choice of a wider window could not be justified –especially within the years of the financial crisis of the country that are characterized by substantial changes in the health care sector affecting, among others, the applied technology. Thus, a wider window would most likely have led to unfair or non-realistic comparisons among hospitals in distant apart years.
|
study
| 99.9 |
Average year-specific and window-specific VRS efficiency scores are presented in Table 3 which has been constructed in the following way . Columns represent years and rows represent windows. Thus, the intersection of a row with a column represents a value in the context of a year within a specific window. It is clear that a hospital can have different efficiency scores for the same year in the context of different windows. More specifically, for a 2-year width window, each year will participate twice in two adjacent windows (excluding the first and last years of the study period which participate only once). Consequently, when reading the table vertically (following the so-called column view) one can see possible efficiency alterations for a hospital in the same year measured against the efficient frontiers of the two different windows it participates. Thus, any efficiency difference, in essence, reflects the impact on the efficient frontier due to changing half of the units (by adding a later year and removing the earliest one from the window) and therefore vertical fluctuation provides an indication for the stability of efficiency results for each year across the two different windows it participates. According to Cooper et al. , a hospital that is efficient in one year regardless of the window is said to be stable in its efficiency rating.
|
study
| 100.0 |
On the other hand, when reading the table horizontally one can see how the efficiency of a hospital changes from one year to another within the same window. Thus, these row views, in essence, make it possible to determine efficiency trends (i.e. whether a hospital exhibits improving, steady or deteriorating efficiency in the second year of a window against to the corresponding one in the first year). As noted in the observation of “stability” and “trend” in window analysis reflects simultaneously both the absolute performance of a hospital over time and the relative performance of that hospital in comparison to the others in the sample. Average values of row and column values, for each section, are presented in the bottom row (labeled “common year”) and in the rightmost column (labeled “entire window”) respectively.
|
study
| 55.72 |
The results in Table 3, for average year-specific and window-specific efficiency scores suggest that there is considerable room for improvement. More specifically, the lowest and highest year-specific mean efficiency scores are [85.5%-91.4%], [85.5%-91.6%], [85.9%-92.6%] and [88.2%-93.9%] for groups A, B, C and D respectively. The detailed results (not presented here) showed that the hospitals that exhibit the lowest efficiency scores for each group are A1 (49.8%, year 2013, window 2012–13), B14 (57.7%, 2010, window 2009–10), C2 (55.2%, 2009, 2009–10) and D9 (57.6%, 2010, window 2010–11). In the same table, the number of fully VRS efficient hospitals is also presented. Thus, year 2011 within window 2010–11 accommodates the highest number of fully efficient units for groups B, C and D. More precisely there are 14, 16 and 15 fully efficient units accounting for 42.4%, 53.3% and 65.2% of the total number of units for groups B, C and D respectively. In group A, the highest number (11) of fully efficient units appears in year 2013 within windows 2012–13. In addition, the VRS efficiency scores for groups A and B show the highest and lowest spread respectively, as evidenced by their corresponding standard deviations (and minimum values not presented in the table). As it can also be seen from the Table 3, the window-specific efficiency scores appear to have an increasing trend until window 2011–12 after which they are reversed and start to deteriorate.
|
study
| 100.0 |
Efficiency trends within windows, on individual hospital basis, are shown in Table 4 which has been constructed in the following way: There are four columns per group corresponding to the four windows. Three symbols “↗”, “↘” and “↔”, are used to denote improvement, deterioration or steadiness of a hospital’s efficiency respectively, when going from the first to the second year within the same window. In addition, the symbol “+” is used in conjunction with these three symbols to denote a fully efficient unit in the implied year on the left or right hand side of the arrow. It is clear that only the three combinations “↗+”, “+↘”, and “+↔+” may arise. The first one, “↗+”, will be used when efficiency improvement starts from a non-efficient unit and leads to a fully efficient unit. Symmetrically, the second one, “+↘”, will be used when efficiency deterioration starts from a fully efficient unit and leads to a non-efficient unit. Finally, the third one, “+↔+”, will be used to denote the transition from a fully efficient unit to an also fully efficient unit. An extra summary column to the right of the four window-specific columns (labeled “inter-temporal trend”) is used in order to designate a constant inter-temporal (all-windows) behavior of a unit whenever there is one. Thus, if the same trend symbol appears for a hospital in all four windows the corresponding symbol (↗, ↘, or ↔) is placed in this summary column. In addition, the extra symbol “√” is placed in this column whenever a hospital manages to be fully efficient in at least one year within every window. Therefore, it is evident that the combined presence of the “√” together with the symbol ↔ (i.e. ↔√) indicates a fully efficient hospital throughout all years within all windows. Similarly, the presence of symbol “√” either sole or accompanying the symbols ↗ and ↘, indicates an (inter-temporal) almost fully efficient unit which we shall call a “semi-efficient” unit. Finally, at the bottom of the table, a summary of column totals shows the number of occurrences of each trend symbol. It is worth noting that the total number of occurrences of the “+” symbol on the left (or right) side of a trend arrow coincides with the total number of fully efficient hospitals in the year implied on the corresponding side of the arrow. For example, as can be seen in summary section of group B in Table 4, the symbol “+” appears 14 times on the right of trend arrows (9 x “↗+” and 5 x “+↔+”) in window (column) 2010–11. There are, therefore, 14 VRS fully efficient hospitals (within window 2010–11) in the year which is implied on the right of the particular trend arrow which, in this case, is 2011. Similarly, there are 7 (5 x “+↔+” and 2 x “+↘”) fully efficient hospitals in year 2010 within window 2010–11. Furthermore, it is evident that 5 hospitals (5 x “+↔+”) managed to be fully efficient in both years within window 2010–11.
|
study
| 99.8 |
Some interesting points emerge from a careful examination of Table 4. In terms of inter-window efficient units, as can be seen from the contents of the summary column for each group, the following apply: There are nine semi-efficient and two inter-efficient hospitals in group A. In group B, there is only one inter-efficient hospital and three semi-efficient ones. This is also the group with the lowest number of inter-temporal efficient units (i.e. semi-efficient plus inter-efficient). Group C has five semi-efficient hospitals and two inter-efficient ones while in group D there are seven semi-efficient and two inter-efficient units. In terms of inter-window trend (i.e. constantly upward/downward units) there are only three hospitals that appear to be trending always upward. These are hospitals B23, C15 and D23. It should be noticed also that the two smaller groups (A and D) are the ones with the highest number of inter-window efficient units.
|
study
| 83.4 |
Table 3 summarizes also the results of average year-specific and window-specific scale efficiency. As can be seen, year-specific scale scores are relatively high and lie within the ranges [89.3–94.2], [85.9–93.6], [92.1–94.6] and [86.4–92.0] for groups A, B, C and D respectively. The diagrams in the second column of Fig. 1 depict graphically the average scale efficiency trend form year-to-year. It is evident that there is a downward dominating tendency for all groups in window 2009–10 (all groups deteriorate slightly, with the exception of group B that shows an inappreciable increment). In the following window, 2010–11, all groups appear to have a sharp upward trend which is followed by a more or less downward trend in window 2011–2012. More precisely one can see that groups B and D in window 2011–12 show a relatively sharp deterioration whilst groups A and C seem to remain almost constant.
|
study
| 100.0 |
Finally, in the last window 2012–13, although there is no clear common trend for all groups it is evident that there is an upward tendency since, with the exception of group B, all other groups show a clear improvement in their average scale efficiency scores when going from year 2012 to 2013. The sixth section of Table 3 presents the number of scale efficient hospitals (CRS). As can be seen the maximum number of scale efficient hospitals is achieved (on average) in the two instances of year 2011 for all groups (with the slight exception of group A). Finally, as far as scale inefficiency is concerned, one should be able to distinguish whether it is attributed to increasing or decreasing returns to scale. The last section in Table 3 summarizes the detailed findings concerning the nature of scale returns for each hospital. The contents of this section show that, with the exception of group A, for which relevant findings are almost equally divided, there is a clear dominance of hospitals operating at increasing returns to scale (IRS).
|
study
| 99.94 |
Table 5 presents the (geometric) average change of the estimations of the input-based Malmquist productivity index together with its decomposition into the components of efficiency change “Eff” (i.e. catching- up to the frontier) and technical change “Tech” (i.e. shift in the frontier). The index of the efficiency change component “Eff” is further decomposed into its two constituting indices of pure technical efficiency change “Pure.eff” and scale efficiency change “Scale”. The results are presented along with their 95% confidence intervals that were calculated applying the bias-corrected and accelerated (BCα) bootstrap methodology introduced by Efron and Tibshirani . Since Malmquist indices are given in geometric rather than arithmetic means, the methodology was applied in the way described by Atkinson and Wilson where the bootstrap procedure is initially used to calculate the confidence intervals for the arithmetic mean of the log(index) values and subsequently take their exponential values in order to derive the confidence intervals for the geometric mean.
|
study
| 100.0 |
As can be seen from the empirical results of the Table 5, all groups of hospitals achieved statistically significant (at the 5% level) increases in productivity between 2009 and 2013 with the exception of group B that is shown to exhibit a (non-statistically significant) slight decrease. It is interesting to note that the larger hospital set (group D) is shown to exhibit the greater (statistically significant) productivity change by 26.8%. However, when the total productivity change is decomposed into efficiency and technical change, it appears clearly that the growth in productivity is primarily attributed (20.2%) to a progressive shift of the efficient frontier and secondarily (5.5%) to efficiency improvement. The results indicate that the dominance of the technology versus efficiency change is reversed for the other groups of hospitals. Specifically, in groups A and B, the contribution of efficiency change in productivity is positive by 11.8% whilst the corresponding contribution of the technology change component is negative by 2.1% and 11.1% respectively. Similarly, the statistically significant productivity improvement by 7.1% in group C is mainly due to efficiency improvement (4.7%) rather than to technology progress (2.4%).
|
study
| 100.0 |
Summing up, the results in Table 5 reveal that all indices have a positive contribution (>1) the only exception being the technology change index in groups A and B where it appears to participate negatively. However, only in group B does a negative contribution lead to a marginal (non-significant) productivity decline by 0.7%.
|
other
| 99.56 |
On an individual basis, the values of the Malmquist productivity indices for groups A, B, C and D are depicted in the four diagrams depicted in Figs 1–4. The hospitals have been arranged clock-wisely in ascending order according to their MPI values (blue line) whilst the first occurrence of an MPI value >1 is denoted by the yellow dots. The MPI line is surrounded by two spirals that correspond to the bootstrap calculated for upper and lower values for the 95% confidence interval. Furthermore, the geometric MPI mean for each group is depicted as a yellow circle. From the diagrams, it is easily deducted that the number of hospitals that experienced overall productivity progress (MPI>1) for the groups A, B, C and D are 14/21 (= 66.6%), 17/33 (= 51.5%), 21/30 (= 70%) and 20/23 (= 87.0%) respectively. The MPI values span in the range from 0.7–2.06 (group A), 0.39–1.64 (group B), 0.56–2.72 (group C) and 0.56–2.32 (group D).
|
study
| 100.0 |
The main objective of this study was to apply DEA to measure the efficiency of the Greek NHS hospitals during the period 2009–2013. The hospitals were categorized and allocated into four separate groups with common characteristics in order to increase their homogeneity. The DEA-Windows method was chosen for the assessment of their efficiency since (a) it leads to increased discrimination on the results especially when applied on small samples and (b) provides a means for the year-to-year comparison of the results. At the same time, the number of DMUs is doubled since, in essence, every actual hospital participates twice in each window and therefore the discrimination of the method is improved.
|
study
| 100.0 |
The findings showed that all four groups of hospitals were operating at relatively high levels of technical and scale efficiency over the whole period. In general, mean pure technical efficiency (VRS) for all groups was found to span approximately a range of values from 85.5% up to almost 94%. Mean scale efficiency was also found to lie within the range 85.9%-94.6%.
|
study
| 96.5 |
The analysis also concluded that there are only seven hospitals that can be considered all-windows best performers. Two of them belong to group A (A6, A15), one (B28) is a member of group B and two (C12, C26) are members of group C and finally two (D5, D13) belong to group D. These seven hospitals are the ones that managed to maintain their full pure technical efficiency in all eight instances of their evaluation against the four different window frontiers. This finding, therefore, suggests that hospital managers and policy makers should pay attention and examine closely and thoroughly their applied practices and technology profile since they can serve as benchmarks for the others. In addition, apart from this small number of the seven all-window best performers, the DEA-window analysis identified also a respectable number of all-window very good performers which we called “semi-efficient”. These are the hospitals that managed to remain fully efficient at least once in every window. It is worth noting that when their results were scrutinized on an individual hospital basis it was found that almost all of them managed to maintain a very high VRS efficiency score in the years that they were not fully efficient. On the other end of the spectrum, the analysis also reported 33 hospitals as being consistently inefficient over all windows. These are 33 hospitals that never managed to achieve a pure technical efficiency score of 100% in any of the years and they are distributed as 6, 14, 7 and 6 in the four groups respectively. Although not all of them can be considered as all-windows worst performers the worst ones are indeed among them. From a managerial point of view, it is clear that this group of hospitals needs also to be scrutinized closely in order to identify peculiarities and possible sources of inefficiency. Hospitals A2, B14, C2 and D9 are the ones with the lowest average efficiency scores (65.8%, 73.3%, 57.7%, 62.9% respectively) over all windows. Among the factors that may explain this finding is the fact that all output/input ratios of these hospitals are among the lowest of their groups throughout all years. Furthermore, hospital A2 is shown to have the lowest occupancy rate (29%) whilst hospitals C2 and D9 exhibit almost a twofold ALoS compared to the average values of their groups. Hospital A2 is located on a rather small island and therefore the low occupancy rate is expected due to the low population. On the other hand, hospitals C2 and D9 are big hospitals located in Athens and are among those covering the more difficult and complicated cases that require longer stays.
|
study
| 99.94 |
As far as efficiency trend is concerned, the finding that emerged from window analysis is the following. Although the results do not reveal the existence of a clear and consistent inter-temporal trend in any of the four groups there is, however, an indication that year 2011 constitutes a turning point in the whole 5-year period since the upward efficiency trends for all groups in window 2010–11 are turning downwards in window 2011–12. A similar finding is documented in another study according to which the middle-sized hospitals (100–400 beds) of the Greek NHS exhibit efficiency improvement in the period 2009–2011.
|
study
| 99.94 |
Empirical results also showed that the hospitals under study were operating at respectable levels of scale efficiency from 85.9% up to 94.6%. Detailed results, summarized in Table 3, show that a small number of hospitals were operating at optimal size (characterized by constant returns to scale), though many others were operating close to their optimal size. Furthermore, by examining individual scale efficiency findings, it was concluded that the hospitals under study exhibit a mix of decreasing and increasing returns to scale at current levels of output. Interestingly, the pattern of scale inefficiency indicates that most of the hospitals are operating in an area of increasing returns to scale implying that they could benefit from increasing the scale of their operations. At this point two things need also to be taken into account. First, the size of each hospital should be interpreted relatively in the context of the group it belongs to and therefore it should not be seen as an absolute “big” or “small” number. Second, scale inefficiency for many hospitals account for as little as 5% or even less. This implies that the potential for their improvement through resizing is rather limited. These findings suggest that hospital managers and policy makers should primarily focus on addressing the technical inefficiency issues before examining ways for a possible restructure of their scale of operations. Besides, it should not be overlooked that technical improvement is more controllable and can be addressed in the short-term without requiring the prior change of scale.
|
study
| 99.94 |
Finally, our findings showed that technical and scale efficiencies of all groups were improved at the end of the 5-year period although these were difficult years of the financial crisis. At this point, it is interesting to examine these results against the ordinary indicators used by hospital managers and policy makers. As can be seen in Table 1 for instance, the indicators of average length of stay and occupancy rate have been improved (-18.8% and +6.8% respectively). The same applies to the indicators for medical and other staff resources per inpatient case which were increased by 55% and 47% respectively. Going one step further, we examined all nine possible indicators that can be formed by the ratios of the three outputs when combined with the three inputs (i.e. the six variables that were used in DEA model). Simple comparisons of year’s 2013 values, against those of 2009, showed that eight out of nine indicators significantly improved, the only exception being the (meaningless) ratio of outpatient visits per bed, which was reduced by 10%. Thus, it can be argued that the findings of the study are in line with the whole “macro-picture” depicted in Table 1. The fact that the efficiency improvement between 2009 and 2013, as measured by DEA, is not of the extent implied by the large increment of ordinary indicators can be mainly attributed to its inherent characteristic of relativity in comparisons. The method is capable to perform only relative measurements of efficiency which as such are valid only inside the "borders" of the particular sample. For instance, the improved performance that would be expected by a plain +5% increment throughout all units' outputs, while keeping all their inputs constant, would not be captured by DEA. On the other hand, the overall augmented DEA scores in year 2013, compared with the ones five years before, indicate that eventually some inefficient hospitals in 2009 managed to operate closer to 2013’s efficient frontier defined by their counterpart efficient ones regardless of the relative possible shift and final distance between the two borders which cannot be accounted for by the method. It is evident therefore that DEA findings are complementary to the ones that can (and should) be derived by other means of performance evaluation.
|
study
| 100.0 |
In this context, the productivity of the hospitals between the first (2009) and last (2013) years of the study was also assessed by means of the Malmquist productivity index so that the changes in technology (reflected by shifts in the technology frontier) can be captured as well. Thus, the overall productivity change was subsequently decomposed into technological and efficiency changes whilst the latter was further analyzed into scale and pure technical efficiency changes. The relevant findings were in line with the ones from window analysis as all efficiency related indices were found to be greater than unity meaning that both scale and pure technical efficiency were improved in 2013. As far as the technological change is concerned, the findings revealed progress for the groups of larger hospitals (B and C) and regress for the smaller (groups A and B). The group D with the largest hospitals (>400 beds) experienced the highest technological progress (+20.2%) whilst the group B with the medium to small hospitals (85–190 beds) exhibited the highest regress (-11.1%). On the other hand, the comparisons of efficiency changes showed a twofold improvement for the medium-to-small sized hospitals (groups A+B) compared to the medium-to-large ones (groups C+D). It is therefore deducted that productivity change in larger hospitals is mainly related to changes in technology, as opposed to the smaller ones for which changes in efficiency appears to have higher contribution. It could be argued that the finding is compatible with the fact that larger hospitals are usually equipped with the latest medical technological innovations and personnel and therefore their productivity growth is heavily dependent on them. A similar finding is also reported by Gannon who examined the productivity growth and efficiency in the production of hospital care in Ireland from 1995 to 1998.
|
study
| 100.0 |
Summing up, it can be concluded that there has undoubtedly been a significant performance improvement due to the large increments of outputs and the corresponding simultaneous large decrements in inputs. At the same time, there was also a respectable efficiency improvement meaning that technically the “center of efficiency mass” of the whole system, viewed as a unit, was elevated positively.
|
other
| 99.94 |
The present study has some limitations which should be taken into account. First, patient severity which may influence utilization and therefore measured efficiency has not been incorporated directly in the study. The lack of relevant data did not allow adjusting outputs for variability in case mix across hospitals. This means that efficiency results may be biased since all outputs are considered equivalent across all facilities. Although it is clear that not all of them are equivalent, it can be considered that this limitation has been adequately remedied as all "special" hospitals were excluded from the study whilst the rest of them were further allocated into four fairly homogenous samples in terms of their size and mixture of services they provide (same clinics/specialties). Consequently, the four groups can be considered as consisting of hospitals with similar case mix. Second, it is worth noting that although the technical aspect of efficiency variations has been covered adequately, the equally important efficiency component that relates to quality variations has been left out in this study due to lack of appropriate and reliable data. It is clear that the establishment of a mechanism for collecting more detailed data in a centralized and systematic way will resolve the afore mentioned limitations and undoubtedly improve the completeness and quality of future efficiency studies.
|
study
| 100.0 |
The fact that health care expenditures represent a substantial proportion of total public expenditure places an enormous pressure to cut costs. Towards this objective in the last five years, the Ministry of Health has an ongoing program of reforms in the public hospital sector. On the other hand, cost containment in the health care sector should be carried out without compromising the quantity and quality of services produced. This is even more pertinent in Greece, given the current economic crisis. One way to compensate for reduced budgets in order to maintain or even improve the level of health care services offered is through identification and elimination of possible sources of inefficiency. It is evident, for instance, that once inefficiency has been eliminated, the saved resources could be devoted to cover other areas of the health care system (improved quality care to patients, innovative technology, staff training). Hence, the establishment of specific and thoroughly researched criteria for measuring efficiency is extremely important. In this context, the present study constitutes an inter-temporal efficiency map for the technical performance of hospital activity in the context of the Greek NHS throughout the five-year period 2009–2013. The empirical findings can be used as a useful guide for both hospital managers and policy makers in their process to improve the health care system, in favor of the true beneficiaries, which are the patients and society in general.
|
study
| 99.9 |
Tau is a microtubule-associated protein (MAP), the main function of which is to stabilize microtubules implicated in axonal transport and axon structure1. In pathological conditions, and for reasons that remain to be elucidated, tau detaches from microtubules and assembles into paired helical filaments (PHF) to form intracellular inclusions2,3. In physiological conditions, tau is considered to be an unfolded protein mainly associated to microtubules. However, post-translational modifications, such as hyperphosphorylation and conformational changes, promote the transition towards pathological forms of tau. Several neurodegenerative diseases are characterized by the progressive accumulation of modified tau, including Alzheimer’s disease and frontotemporal dementia and FTDP-174. In FTDP-17, several autosomal dominant missense mutations identified in the tau gene demonstrate that the modification of tau protein can cause neurodegeneration and dementia5. In Alzheimer’s disease, tau pathology is associated with its hyperphosphorylation, conformational changes and deposition in NFT, in the absence of any mutation in the tau gene. However, the development of the tau pathology correlates with neuronal loss and progressive cognitive deficits. This indicates that the post-translational modifications of normal tau confer a crucial role for this protein in neuronal dysfunction and neurodegeneration6–8.
|
review
| 98.7 |
In order to devise effective therapies, it is however important to identify critical steps in the gain of toxic functions of tau in neurons. Tau missorting, often associated with the compromised interaction of the protein with microtubules, is considered as an initial step in this process9. Several factors determine the affinity of different tau isoforms for microtubules, such as the number of repeats in the microtubule-binding domain (3 R versus 4 R), or the extent and site of tau phosphorylation. The phosphorylation-induced dissociation of tau from tubulin has been proposed to cause the loss of the microtubule-stabilizing function of tau, possibly contributing to neuronal degeneration10–12. When dissociated from microtubules, tau can then acquire various conformations, such as a compact “paperclip-like” conformation bringing together the N- and C-terminal portions of the protein13. Furthermore, the microtubule-binding domain contains hexapeptide motifs, which have a propensity to form β-sheets14. The transition to β-sheet structure is considered to be an important step, which promotes the formation of oligomers that can be competent for further aggregation into larger fibrillar structures such as PHF15. It is therefore important to determine how variations in the amino-acid sequence of the microtubule-binding domain critically control the kinetic of this process16–18.
|
study
| 100.0 |
To address this question, we used an in vivo model of tau pathology based on intracerebroventricular injections of AAV2/6 vectors in mouse neonates, to induce widespread overexpression of various forms of human tau in the forebrain. Tau pathology developed in the cortex and hippocampus, and induced motor behavioral deficits. To test the hypothesis that the propensity of the microtubule-binding domain to form β-sheets is a key factor in tau toxicity, we compared three variants of human 4R0N tau. The wild-type (WT) form was first compared to the pathogenic P301S mutant linked to FTDP-17. Conversely, to reduce the propensity of the microtubule-binding domain to form β-sheets, we introduced β-sheet-breaking proline residues in WT tau. The results show that the neuronal dysfunction caused by tau overexpression depends on the propensity of the microtubule-binding domain to form β-sheets. In addition, the observed behavioral defects correlate with differences in tau hyperphosphorylation and aggregation, as well as with the propensity of tau to protect neurites exposed to the microtubule-destabilizing agent vinblastine. These findings indicate that amino acid substitutions, or other modifications that may affect tau conformation in the microtubule-binding repeats, determine the progression and the type of tau-induced pathology in neuronal cells in vivo.
|
study
| 100.0 |
In order to model tau pathology in mice and explore behavioral changes induced by tau overexpression, we injected mouse neonates in the lateral ventricles (ICV injection, postnatal day 2) with AAV2/6-pgk vectors either encoding the WT form of human 4R0N tau (AAV-WT), or encoding the P301S mutant tau associated with frontotemporal dementia (AAV-P301S) (Fig. 1a). To modulate the propensity of tau to form β-sheets, we generated a vector encoding another form of the protein (2P tau) with amino acid substitutions in the domain that controls binding to microtubules (Fig. 1a). Two Ile-to-Pro mutations were introduced between R1 and R2 (I277P), and between R2 and R3 (I308P), two regions of the microtubule-binding domain that are critical for the transition towards a β-sheet conformation. These mutations have been previously reported to act as β-structure breakers in the highly aggregating ΔK280 mutant tau19. As a control group, mouse neonates were injected with the same dose of a vector encoding the fluorescent maxFP-Green protein (AAV-maxFP). This technique of AAV injection has been reported to efficiently transduce widespread regions of the mouse forebrain20. Indeed, at three months post-injection, overexpression of human tau was detected by immunohistochemistry on histological brain sections using an antibody specific for human tau (HT7). The human tau protein was broadly expressed in neuronal cells throughout the mouse forebrain (Fig. 1b), the highest expression being observed in the cortex and hippocampus (Fig. 1c). Expression was most prominent in large-sized neurons, such as pyramidal cells in cortical layers II/III and V, or hippocampal CA1 region (Fig. 1b and c). Regardless of the form of tau overexpressed (WT or P301S), the distribution of the protein was found to be both somatodendritic and axonal (Fig. 1c). As expected, there was no HT7 signal in mice injected with AAV-maxFP (Fig. 1c). Following ICV injection of AAV-2P, HT7 immunostaining showed similar expression of the tau protein in both the mouse cortex and hippocampus (Fig. 1c).Figure 1Intracerebroventricular injection of AAV-tau leads to widespread expression of human tau in the mouse forebrain. (a) Schema of the three tau-encoding sequences with amino acid substitutions. (b) Representative immunolabeling of human tau (HT7 antibody) in sagittal brain section, 3 months after injection of AAV-WT shows broad expression of human tau, mainly in cortex and hippocampus. (c) Representative HT7 immunolabeling in sagittal sections shows human tau-positive neurons in cortex and hippocampus (CA1) of mice injected with either AAV-WT, AAV-P301S or AAV-2P, 3 and 7 months after vector injection. Negative HT7 staining in a control mouse injected with AAV-maxFP is shown for comparison. Scale: 100 µm.
|
study
| 100.0 |
Intracerebroventricular injection of AAV-tau leads to widespread expression of human tau in the mouse forebrain. (a) Schema of the three tau-encoding sequences with amino acid substitutions. (b) Representative immunolabeling of human tau (HT7 antibody) in sagittal brain section, 3 months after injection of AAV-WT shows broad expression of human tau, mainly in cortex and hippocampus. (c) Representative HT7 immunolabeling in sagittal sections shows human tau-positive neurons in cortex and hippocampus (CA1) of mice injected with either AAV-WT, AAV-P301S or AAV-2P, 3 and 7 months after vector injection. Negative HT7 staining in a control mouse injected with AAV-maxFP is shown for comparison. Scale: 100 µm.
|
study
| 100.0 |
Next, we sought to evaluate and compare the effect of each of these forms of tau on mouse behavior. As tau was highly expressed throughout the mouse cortex including the somatosensory and motor regions, we assessed exploratory locomotion and motor coordination, as such behavioral readouts are typically impaired in mouse models of tauopathies21–23. We first evaluated the exploratory behavior of the mice in the open field, at 3 and 7 months after vector injection (Fig. 2a). At three months, there was no significant difference between groups in the distance travelled by the mice. However, when the same test was performed with 7-months old mice, the group injected with the AAV-P301S vector showed a significant increase in travelled distance compared to both the control and AAV-WT groups (Fig. 2a). This effect may denote the progressive development of a hyperactive behavior in these mice, which is consistent with previous observations in mice injected with an AAV1 vector overexpressing the P301L tau mutant24,25.Figure 2AAV-tau injection induces behavioral deficits: the effects of the P301S and 2P modifications. (a) Open field test performed at 3 and 7 months after AAV-tau injection, compared to the control group injected with AAV-maxFP. Data represent the distance travelled by each mouse on average, for each experimental group. (b) Rotarod test performed at 1.5 and 7 months after AAV-tau injection, compared to the control group injected with AAV-maxFP. The graph shows the average time on the rotating rod for each experimental group. AAV-maxFP: n = 15 mice; AAV-WT: n = 12; AAV-P301S: n = 13. Statistical analysis: two-way ANOVA (time × group) with repeated measures, followed by Newman-Keuls post-hoc test. Open field: group effect F(2,37) = 4.29; Rotarod: group × time effect F(2,37) = 4.38, *p < 0.05; **p < 0.01; ***p < 0.001. (c) Rotarod test to compare AAV-maxFP, AAV-WT and AAV-2P at 1.5, 3 and 7 months after vector injection. The graph shows the average time on the rotating rod for each experimental group. AAV-maxFP: n = 4 mice; AAV-WT: n = 16; AAV-P301S: n = 13. Statistical analysis: two-way ANOVA (time × group) with repeated measures, followed by Fisher’s LSD post-hoc test. Group effect F(2, 30) = 4.18, *p < 0.05.
|
study
| 100.0 |
AAV-tau injection induces behavioral deficits: the effects of the P301S and 2P modifications. (a) Open field test performed at 3 and 7 months after AAV-tau injection, compared to the control group injected with AAV-maxFP. Data represent the distance travelled by each mouse on average, for each experimental group. (b) Rotarod test performed at 1.5 and 7 months after AAV-tau injection, compared to the control group injected with AAV-maxFP. The graph shows the average time on the rotating rod for each experimental group. AAV-maxFP: n = 15 mice; AAV-WT: n = 12; AAV-P301S: n = 13. Statistical analysis: two-way ANOVA (time × group) with repeated measures, followed by Newman-Keuls post-hoc test. Open field: group effect F(2,37) = 4.29; Rotarod: group × time effect F(2,37) = 4.38, *p < 0.05; **p < 0.01; ***p < 0.001. (c) Rotarod test to compare AAV-maxFP, AAV-WT and AAV-2P at 1.5, 3 and 7 months after vector injection. The graph shows the average time on the rotating rod for each experimental group. AAV-maxFP: n = 4 mice; AAV-WT: n = 16; AAV-P301S: n = 13. Statistical analysis: two-way ANOVA (time × group) with repeated measures, followed by Fisher’s LSD post-hoc test. Group effect F(2, 30) = 4.18, *p < 0.05.
|
study
| 100.0 |
We also explored the impact of tau overexpression on motor performance using the rotarod test. Remarkably, already at 1.5 months after injection the AAV-P301S group showed a significant decrease in latency to fall from the rod rotating at 35 rpm when compared to the AAV-maxFP control group (Fig. 2b). At 7 months, the effect of tau was even more pronounced, as the latency to fall was 14.0 ± 1.8 sec in the AAV-P301S group, and 31.3 ± 5.2 sec in the AAV-WT group, both significantly different from the control group (52.5 ± 3.2 sec). In particular, expression of P301S tau worsened the motor performance of the mice, which was found to be significantly lower than the two other groups (Fig. 2b). Whereas the deficits caused by WT tau remain stable, the P301S mutant tau leads to a time-dependent deterioration of the motor phenotype, indicating a progressive pathology.
|
study
| 100.0 |
In a second experiment, we compared the effects of AAV-WT and AAV-2P on mouse behavior using the rotarod test (Fig. 2c). Again, as compared to the AAV-maxFP, AAV-WT induced a significant decrease in the latency to fall already at 1.5 months post-injection, and which remained stable at 3 and 7 months after vector injection (Fig. 2c). Remarkably, however, the presence of the two β-structure breaking prolines in tau (AAV-2P) reversed this effect, as the mice injected with the AAV-2P vector performed as well as the control AAV-maxFP mice in the rotarod test (Fig. 2c). Over the whole course of the experiment, their latency to fall was significantly increased for the AAV-2P mice compared to the group injected with AAV-WT (Fig. 2c).
|
study
| 100.0 |
Next, in order to correlate the observed behavioral effects with possible alterations of the tau protein, we performed a biochemical characterization of human tau in AAV-WT and AAV-P301S mice. The expression level of tau in the mouse forebrain was assessed at different ages by western blotting in total brain homogenates. Using an antibody that recognizes both mouse and human tau (tau5), we observed an increase in the expression of a tau species with an apparent molecular weight of approximately 55 kDa, which indeed corresponds to 4R0N human tau (Supplementary Fig. S1). This isoform is also abundantly present in protein extracts from the neocortex of an Alzheimer’s patient and of an unaffected control. Compared to the level of endogenous tau in the forebrain of a mouse injected with the control AAV-maxFP vector, the amounts of total tau found in the AAV-WT and AAV-P301S mice were nearly doubled.
|
study
| 100.0 |
For a more specific analysis of overexpressed tau between the WT and P301S mice, we used the HT7 antibody that specifically detects human tau. As shown in Fig. 3a and b, the amount of human tau in brain protein homogenates appeared stable up to 7 months post-injection in the AAV-WT group of mice. In contrast, in mice injected with the vector encoding P301S mutant tau, we observed a time-dependent accumulation of human tau whereby P301S tau was significantly increased at 7 months when compared to 1.5 and 3 months (Fig. 3a,b). P301S tau reached nearly a two-fold higher level than WT tau in 7-months old mice. Notably, we noticed the presence of high-molecular weight species of the tau protein in brain extracts of the AAV-P301S mice (Fig. 3a).Figure 3The level of human tau increases over time in the brain of mice injected with AAV-P301S. (a) Detection of overexpressed tau using the human tau-specific HT7 antibody in mouse forebrain homogenates at 1.5, 3 and 7 months post-AAV injection. Actin is used as loading control. *Indicates higher molecular weight human tau species observed in the P301S group of mice. (b) Relative human tau levels measured by western blotting of mouse forebrain homogenates (1.5, 3 and 7 months post- injection). Blots are shown in Supplementary Figure 1. For each animal, human tau values are normalized to actin and plotted as percentage of the mean value for the AAV-WT group at 1.5 months. Statistical analysis: two-way ANOVA followed by Tukey’s post-hoc test, group × time effect F(2,30) = 4.89, **p < 0.01; ***p < 0.001. (c) The graph compares the level of human tau (HT7/TAU-13 AlphaLisa) in total soluble proteins from forebrain homogenates, as well as in the soluble and insoluble fractions following sarkosyl extraction. Samples are derived from AAV-WT and AAV-2P mice, 7 months after AAV injection; n = 8 mice per group.
|
study
| 100.0 |
The level of human tau increases over time in the brain of mice injected with AAV-P301S. (a) Detection of overexpressed tau using the human tau-specific HT7 antibody in mouse forebrain homogenates at 1.5, 3 and 7 months post-AAV injection. Actin is used as loading control. *Indicates higher molecular weight human tau species observed in the P301S group of mice. (b) Relative human tau levels measured by western blotting of mouse forebrain homogenates (1.5, 3 and 7 months post- injection). Blots are shown in Supplementary Figure 1. For each animal, human tau values are normalized to actin and plotted as percentage of the mean value for the AAV-WT group at 1.5 months. Statistical analysis: two-way ANOVA followed by Tukey’s post-hoc test, group × time effect F(2,30) = 4.89, **p < 0.01; ***p < 0.001. (c) The graph compares the level of human tau (HT7/TAU-13 AlphaLisa) in total soluble proteins from forebrain homogenates, as well as in the soluble and insoluble fractions following sarkosyl extraction. Samples are derived from AAV-WT and AAV-2P mice, 7 months after AAV injection; n = 8 mice per group.
|
study
| 100.0 |
In the experiment comparing AAV-WT and AAV-2P, the amount of soluble human tau present in forebrain homogenates was similar in these two groups of mice, at 7 months after vector injection (Fig. 3c). Moreover, a sarkosyl extraction demonstrated similar levels of human tau in both the sarkosyl-soluble and insoluble fractions, indicating that the 2P modification did not have any effect neither on the overall tau expression nor on the deposition of insoluble human tau (Fig. 3c).
|
study
| 100.0 |
Altogether, these results showed that injection of AAV-tau vectors leads to sustained levels of human tau in the mouse forebrain, which are comparable across the different forms of tau. However, the P301S mutant form of tau tends to accumulate over time. Next, we further analyzed tau misfolding and phosphorylation as a function of time, as these modifications are associated with tauopathies.
|
study
| 100.0 |
Conformational changes in the tau protein, including transition towards a “paperclip-like” conformation, or towards the formation of β-structures aggregating into PHF, occur in the development of tauopathies26, and can precede the formation of NFTs27. We analyzed the presence of misfolded tau in the brain of AAV-injected mice using conformation-specific tau markers. Immunohistochemistry for MC1, an antibody specific for early changes in tau conformation, revealed the presence of abnormally folded tau in the brain regions overexpressing human tau, including the cortex and hippocampus (Fig. 4a). The staining was observed in the AAV-WT, AAV-P301S and in the AAV-2P groups of mice, 7 months after vector injection.Figure 4Aggregated forms of human tau progressively accumulate in the brain of mice injected with AAV-P301S. (a) Representative images of the hippocampus immunolabeled with the conformational MC1 antibody. Note the presence of MC1-positive neurons and neurites, 7 months after injection of either AAV-WT, AAV-P301S or AAV-2P. Scale bar: 100 μm. (b) Quantification of MC1-positive misfolded tau in forebrain protein homogenates using an HT7/MC1 AlphaLisa at 1.5 and 7 months post-injection. Statistical analysis: Kruskal-Wallis test, H(2, N = 38) = 28.25. (c) HT7/MC1 values are normalized for total human tau determined with a HT7/TAU-13 AlphaLisa. (d) The amount of human tau multimers is determined with a HT7/HT7 single-epitope AlphaLisa and normalized for total human tau. For c and d: data represent relative arbitrary units. Statistical analysis: two-way ANOVA followed by Tukey’s post-hoc test, group × time effect (c) F(2, 36) = 6.55, (d) F(2, 36) = 3.88, *p < 0.05; **p < 0.01; ***p < 0.001. (e) Significant correlation between the amount of MC1-immunoreactive human tau and human tau multimers (Pearson). (b–d) 1.5 and 3 months: AAV-maxFP: n = 2; AAV-WT and AAV-P301S: n = 6. 7 months: AAV-maxFP: n = 6; AAV-WT: n = 8; AAV-P301S: n = 10 mice. (f–h) Representative histological analysis, 7 months after AAV-P301S injection: (f) AT100 immunolabeling of phosphorylated tau in the hippocampus (CA1). Scale bar: 100 µm. (g) Thioflavine-S positive neurofibrillary deposits in cortical neurons. Scale bar: 25 µm. (h) Gallyas silver staining of argyrophilic filamentous inclusions in cortical neurons. Scale bar: 25 μm. (i) Detection of misfolded tau (HT7/MC1 AlphaLisa) in mouse forebrain homogenates, at 3 and 7 months after either AAV-WT or AAV-2P injection. Single mouse values were normalized for total human tau (HT7/TAU-13 AlphaLisa); n = 8 mice per group. (j) Detection of tau multimers (HT7/HT7 single-epitope AlphaLisa); n = 8 mice per group. Statistical analysis: two-tailed Student’s t test.
|
study
| 100.0 |
Aggregated forms of human tau progressively accumulate in the brain of mice injected with AAV-P301S. (a) Representative images of the hippocampus immunolabeled with the conformational MC1 antibody. Note the presence of MC1-positive neurons and neurites, 7 months after injection of either AAV-WT, AAV-P301S or AAV-2P. Scale bar: 100 μm. (b) Quantification of MC1-positive misfolded tau in forebrain protein homogenates using an HT7/MC1 AlphaLisa at 1.5 and 7 months post-injection. Statistical analysis: Kruskal-Wallis test, H(2, N = 38) = 28.25. (c) HT7/MC1 values are normalized for total human tau determined with a HT7/TAU-13 AlphaLisa. (d) The amount of human tau multimers is determined with a HT7/HT7 single-epitope AlphaLisa and normalized for total human tau. For c and d: data represent relative arbitrary units. Statistical analysis: two-way ANOVA followed by Tukey’s post-hoc test, group × time effect (c) F(2, 36) = 6.55, (d) F(2, 36) = 3.88, *p < 0.05; **p < 0.01; ***p < 0.001. (e) Significant correlation between the amount of MC1-immunoreactive human tau and human tau multimers (Pearson). (b–d) 1.5 and 3 months: AAV-maxFP: n = 2; AAV-WT and AAV-P301S: n = 6. 7 months: AAV-maxFP: n = 6; AAV-WT: n = 8; AAV-P301S: n = 10 mice. (f–h) Representative histological analysis, 7 months after AAV-P301S injection: (f) AT100 immunolabeling of phosphorylated tau in the hippocampus (CA1). Scale bar: 100 µm. (g) Thioflavine-S positive neurofibrillary deposits in cortical neurons. Scale bar: 25 µm. (h) Gallyas silver staining of argyrophilic filamentous inclusions in cortical neurons. Scale bar: 25 μm. (i) Detection of misfolded tau (HT7/MC1 AlphaLisa) in mouse forebrain homogenates, at 3 and 7 months after either AAV-WT or AAV-2P injection. Single mouse values were normalized for total human tau (HT7/TAU-13 AlphaLisa); n = 8 mice per group. (j) Detection of tau multimers (HT7/HT7 single-epitope AlphaLisa); n = 8 mice per group. Statistical analysis: two-tailed Student’s t test.
|
study
| 100.0 |
To quantitatively assess tau misfolding and multimerization, we first developed three AlphaLisa assays to analyze tau in protein homogenates from the mouse forebrain. The assays were conceived using the human-specific HT7 monoclonal antibody as capture antibody. For the first assay, we used MC1 as reporter antibody for early conformational changes of tau (Fig. 4b,c). Already at 1.5 months after injection, we detected a significantly higher MC1 signal in the mice injected with the tau-expressing vectors as compared with mice injected with the control vector (Fig. 4b). These data indicate that human tau protein rapidly acquires a pathology-prone conformation when overexpressed in the mouse forebrain. In order to effectively compare the misfolding of tau between animals, the signal obtained with the HT7/MC1 assay was normalized to the total amount of human tau, determined for each sample in a parallel AlphaLisa with the HT7/TAU-13 monoclonal antibody pair. As shown in Fig. 4c, the normalized MC1-positive tau remained constant over time in the AAV-WT mice. In contrast, the injection of the AAV-P301S vector induced a progressive increase in the amount of misfolded MC1-tau, which was increased by 4.6- and 5.9-fold compared to AAV-WT at 3 and 7 months post-injection, respectively.
|
study
| 100.0 |
Tau multimers were measured using the same HT7 antibody for capture and detection in a single-epitope immunodetection design that cannot detect monomeric tau forms (Fig. 4d). When measuring HT7/HT7 tau multimers, we made very similar observations (compare Fig. 4c with d), highly correlated with the levels of the MC1 signal (Fig. 4d,e). Again, while aggregated tau remained stable in the AAV-WT group of mice, a progressive accumulation of tau multimers occurred in the forebrain of the AAV-P301S group. Compared to WT tau, the level of P301S tau multimers was increased by 3.1- and 3.3-fold at 3 and 7 months post-injection, respectively. Overall, we find a prominent effect of the P301S mutation on the misfolding and aggregation of the tau protein in vivo.
|
study
| 100.0 |
Biochemical detection of aggregated P301S tau indicates that fibrillar tau may accumulate over time in these animals. Therefore, we performed additional histological analyses using markers for more advanced tau pathology. The AT100 antibody was used to detect abnormal tau phosphorylation on residues Thr 212/Ser 214, also present on PHF (Fig. 4f)28. To further substantiate the presence of tau deposition, we used the thioflavine-S and Gallyas silver stainings, which reveal neurofibrillary deposits of PHF. We observed the presence of AT100-positive neuronal cell bodies and processes only in the group of mice injected with the AAV-P301S vector (Fig. 4f). These neurons were mainly distributed in the cortex and hippocampus 7 months after vector injection. In the same samples, we also observed the presence of neurons positive for thioflavine-S in the cortex (Fig. 4g), as well as the deposition of argyrophilic filamentous inclusions stained with Gallyas (Fig. 4h). The filamentous inclusions were found in the neuronal soma or appeared as structures resembling neuropil threads (Fig. 4h). Of note, we did not detect any sign of advanced tau pathology in any of the mice injected with AAV-WT. Altogether, these data confirmed the progressive modification of tau leading to misfolding and aggregation, which was most evident for the P301S pathogenic mutation of tau.
|
study
| 100.0 |
When comparing WT and 2P tau, we did not find any significant effect of the 2P modification neither on the amount of misfolded tau in the mouse forebrain (HT7/MC1 AlphaLisa normalized to total human tau, Fig. 4i), nor on the amount of tau multimers (HT7/HT7 AlphaLisa normalized to total human tau, Fig. 4j), at 3 and 7 months post-injection.
|
study
| 100.0 |
The deposition of abnormal tau hyperphosphorylated on several residues is a typical feature of several tauopathies. The phosphorylation pattern of tau was therefore assessed in the brain homogenates obtained from the mice at different ages. For quantitative determination of tau phosphorylation at specific sites associated with tauopathies29, several additional AlphaLisa assays were established. As before, we used the HT7 or TAU-13 capture antibodies coupled to the donor beads, whereas detection was performed with monoclonal antibodies specific to different phosphorylated tau residues coupled to the acceptor beads. For each sample, the abundance of the phosphorylated epitopes was normalized to the total amount of human tau, in order to also take into account the different levels of WT and P301S tau as a function of age. Overall, the data demonstrated tau phosphorylation as early as 1.5 months after vector injection in both the AAV-WT and AAV-P301S groups. However, phosphorylation extent varied for most residues, as a function of age and presence of the P301S mutation (Fig. 5a).Figure 5Human WT tau is more subjected to phosphorylation in the mouse forebrain than P301S and 2P tau. (a) AlphaLisa assays are performed on forebrain protein homogenates to determine the levels of human tau phosphorylation at the indicated residues, as function of age (1.5, 3 and 7 months after injection) and tau variant (AAV-WT or AAV-P301S). Data represent relative levels of phosphorylated tau, normalized to the total amount of human tau determined in a parallel AlphaLisa. Note that WT tau is more heavily phosphorylated than P301S tau at residues Ser396, Ser 202/Thr 205, Thr 181 and Thr 231. For all conditions: n = 6. Statistical analysis: two-way ANOVA followed by Tukey’s post-hoc test, *p < 0.05; **p < 0.01; ***p < 0.001. (b) Representative images of phosphorylated tau deposition, 7 months after injection of AAV-WT and AAV-P301S. Immunolabeling with anti-phospho tau antibodies AT8 (Ser202/Thr205) in the cortex (left panels) and with PHF1 (Ser396) in the hippocampus (right panels) are shown. Note the localization of P301S tau in the neuronal soma and dendrites. Scale bar: 100 μm. (c) AlphaLisa quantification of human tau phosphorylation at residues Ser 202/Thr 205, Thr 231, Ser 396 and Thr 181 in mouse forebrain homogenates, 7 months after injection of either AAV-WT or AAV-2P. The graphs represent the levels of phosphorylated tau relative to total human tau; n = 8 mice per group. Statistical analysis: two-tailed Student’s t test. *p < 0.05; **p < 0.01; ***p < 0.001. (d) Representative immunodetection of phosphorylated tau using the AT8 (Ser 202/Thr 205) and PHF1 (Ser 396/Ser 404) antibodies in sagittal sections of the mouse cortex 7 months after injection of either AAV-WT or AAV-2P. Note the decreased immunoreactivity for phosphorylated tau in AAV-2P-injected mice. Scale bar: 100 μm.
|
study
| 100.0 |
Human WT tau is more subjected to phosphorylation in the mouse forebrain than P301S and 2P tau. (a) AlphaLisa assays are performed on forebrain protein homogenates to determine the levels of human tau phosphorylation at the indicated residues, as function of age (1.5, 3 and 7 months after injection) and tau variant (AAV-WT or AAV-P301S). Data represent relative levels of phosphorylated tau, normalized to the total amount of human tau determined in a parallel AlphaLisa. Note that WT tau is more heavily phosphorylated than P301S tau at residues Ser396, Ser 202/Thr 205, Thr 181 and Thr 231. For all conditions: n = 6. Statistical analysis: two-way ANOVA followed by Tukey’s post-hoc test, *p < 0.05; **p < 0.01; ***p < 0.001. (b) Representative images of phosphorylated tau deposition, 7 months after injection of AAV-WT and AAV-P301S. Immunolabeling with anti-phospho tau antibodies AT8 (Ser202/Thr205) in the cortex (left panels) and with PHF1 (Ser396) in the hippocampus (right panels) are shown. Note the localization of P301S tau in the neuronal soma and dendrites. Scale bar: 100 μm. (c) AlphaLisa quantification of human tau phosphorylation at residues Ser 202/Thr 205, Thr 231, Ser 396 and Thr 181 in mouse forebrain homogenates, 7 months after injection of either AAV-WT or AAV-2P. The graphs represent the levels of phosphorylated tau relative to total human tau; n = 8 mice per group. Statistical analysis: two-tailed Student’s t test. *p < 0.05; **p < 0.01; ***p < 0.001. (d) Representative immunodetection of phosphorylated tau using the AT8 (Ser 202/Thr 205) and PHF1 (Ser 396/Ser 404) antibodies in sagittal sections of the mouse cortex 7 months after injection of either AAV-WT or AAV-2P. Note the decreased immunoreactivity for phosphorylated tau in AAV-2P-injected mice. Scale bar: 100 μm.
|
study
| 100.0 |
The level of phosphorylation of human tau at residues Ser 409, Ser 396 and Ser 202/Thr 205 was dependent on the age of the animals, progressively increasing between 1.5 months and 7 months post-injection (Fig. 5a). Remarkably, tau phosphorylation on residues Ser 396, Ser 202/Thr 205, Thr 181 and Thr 231 was more abundant on WT tau than on P301S tau (Fig. 5a). In 7 month-old mice, this effect was particularly evident on the phosphorylation of the Ser 396, Ser 202/Thr 205 and Thr 181 residues. Therefore, although P301S tau accumulates as a function of age, the relative contribution of phosphorylation on certain residues is more pronounced for WT tau. In contrast, age and mutation effects were different for other phosphorylation sites. The level of phosphorylation of Ser 409 and Thr 212 was nearly identical for WT and P301S tau (Fig. 5a). In particular, phosphorylation of Thr 212 showed no dependency on the presence of tau mutation and age.
|
study
| 100.0 |
In addition to these biochemical assays, the presence of hyperphosphorylated tau was investigated by immunohistology using two phospho-tau specific antibodies: AT8 (phosphorylated Ser 202/Thr 205) and PHF1 (phosphorylated Ser 396/Ser 404). In 7-month-old mice, a strong immunoreactivity for AT8 and PHF1 antibodies was found in the cortex and hippocampus, in the AAV-WT group and to a lesser extent in the AAV-P301S group (Fig. 5b). Whereas immunoreactivity for phospho-tau epitopes was mainly observed in the axons of neurons overexpressing WT tau, phosphorylated P301S was more evident in the somatodendritic compartment. Altogether, these results indicate that overexpression of WT tau is characterized by a more extensive hyperphosphorylation pattern, which tends to further develop over time when compared to P301S tau.
|
study
| 100.0 |
Next, we compared the phosphorylation status of WT and 2P tau. The level of phosphorylation was assessed on sites Ser 202/Thr 205, Thr 231, Ser 396 and Thr 181, which were found highly phosphorylated on WT tau (see Fig. 5a). In contrast to the absence of effects for misfolding and aggregation, the 2P modification clearly affected the pattern of tau hyperphosphorylation. When compared to WT tau, phosphorylation was significantly decreased in the 2P tau group for all the residues analyzed, 7 months after vector injection (Fig. 5b,c). The effect of the 2P modification was particularly dramatic on the phosphorylation of Ser 202/Thr 205, which was almost completely suppressed (Fig. 5c). Moreover, histological analysis confirmed these findings. Indeed, the immunoreactivity for AT8 (phosphorylation at Ser 202/Thr 205) was nearly absent, whereas PHF1 immunoreactivity (phosphorylation at Ser 396/Ser 404) was partly decreased in the cortex of the mice overexpressing 2P tau, as compared to WT tau (Fig. 5b). Overall, the main effect of the 2P modification appears to be a significant decrease in the extent of human tau hyperphosphorylation.
|
study
| 100.0 |
Altogether, the comparison between WT and 2P tau suggests that the behavioral impairments caused by the overexpression of human WT tau are likely to be conferred by increased burden of hyperphosphorylation, rather than by the accumulation of misfolded tau protein. Remarkably, amino acid substitutions that affect the conformation of the microtubule-binding region of 4R0N tau have major effects on the downstream accumulation of hyperphosphorylated tau, and prevent the impairments of motor behavior observed in this mouse model.
|
study
| 100.0 |
The phosphorylation of tau affects the association of the protein to microtubules. To determine if the observed differences in the phosphorylation status of overexpressed tau correlate with changes in microtubule stability, we used transmission electron microscopy (TEM) to explore possible ultrastructural changes in the mouse neocortex.
|
study
| 100.0 |
Following injection of the WT tau-encoding vector, microtubule arrangement appeared normal and indistinguishable from the normal situation (Fig. 6a,b). However, in both the P301S and 2P tau groups, local increases in the density of the microtubule network were observed mainly in axons (Fig. 6c,d). This includes the axonal boutons, where the microtubules could also be seen close to the synaptic vesicles, and active zone (Fig. 6d). Microtubules were often organized in bundles with hexagonal symmetry, similar to the tau-induced microtubule fascicles described previously30. To quantify this effect, the number of microtubules was determined in individual axons. In both the P301S and 2P conditions, the number of microtubules per axon was significantly increased compared to either non-injected or AAV-WT injected mice (Fig. 6e). When the number of microtubules was normalized to the cross-sectional area of the axon, we also observed a significant increase in density in the P301S and 2P mice versus the control non-injected mice (Fig. 6f). In contrast, the microtubule density was not changed in the dendrites (Fig. 6g). Therefore, we find that the P301S and 2P variants of tau lead to clear changes in microtubule number and density. These results suggest that these two forms of tau, which have a modified amino-acid sequence in the microtubule-binding domain, are more likely to interact with tubulin than WT tau, which is consistent with the reduced level of tau phosphorylation observed in these conditions.Figure 62P-modified and P301S tau increase microtubule number and density in axonal profiles in the mouse cortex. Representative electron micrographs from layer II/III of mouse neocortex neuropil, 7 months post-vector injection. Images are taken close to the site of injection. (a,b) In the mouse injected with AAV-WT, the neuropil shows normal morphology with microtubules scattered in axon terminals and dendrites, similar to the ultrastructure observed in a non-injected mouse (normal cortex). (c,d) In mice injected with either AAV-P301S or AAV-2P, there is an abundance of cytoskeletal elements, with high densities of aligned microtubules (indicated by white arrowheads) located in axons. These are also seen in the axonal boutons (indicated by *). Scale bars: 500 nm. (e) Quantification of the number of microtubules per axon in each condition. (f,g) Quantification of microtubule density in individual axons. Note the significant increase in microtubule number and density in the AAV-P301S and AAV-2P conditions. Statistical analysis: one-way ANOVA followed by Tukey’s post-hoc test, n = 29–30 axons per condition, *p < 0.05; **p < 0.01; ***p < 0.001. (h) Representative electron micrographs from the neocortex, 7 months post-injection of the AAV-P301S tau vector. Enlarged axons containing high densities of aligned microtubules (indicated by #) are present, as well as axonal swellings containing large accumulations of degenerated organelles including whorls, membrane inclusions and degenerating mitochondrial profiles (white arrows). Scale bars: 1 µm (left panels) and 500 nm (right panel).
|
study
| 100.0 |
2P-modified and P301S tau increase microtubule number and density in axonal profiles in the mouse cortex. Representative electron micrographs from layer II/III of mouse neocortex neuropil, 7 months post-vector injection. Images are taken close to the site of injection. (a,b) In the mouse injected with AAV-WT, the neuropil shows normal morphology with microtubules scattered in axon terminals and dendrites, similar to the ultrastructure observed in a non-injected mouse (normal cortex). (c,d) In mice injected with either AAV-P301S or AAV-2P, there is an abundance of cytoskeletal elements, with high densities of aligned microtubules (indicated by white arrowheads) located in axons. These are also seen in the axonal boutons (indicated by *). Scale bars: 500 nm. (e) Quantification of the number of microtubules per axon in each condition. (f,g) Quantification of microtubule density in individual axons. Note the significant increase in microtubule number and density in the AAV-P301S and AAV-2P conditions. Statistical analysis: one-way ANOVA followed by Tukey’s post-hoc test, n = 29–30 axons per condition, *p < 0.05; **p < 0.01; ***p < 0.001. (h) Representative electron micrographs from the neocortex, 7 months post-injection of the AAV-P301S tau vector. Enlarged axons containing high densities of aligned microtubules (indicated by #) are present, as well as axonal swellings containing large accumulations of degenerated organelles including whorls, membrane inclusions and degenerating mitochondrial profiles (white arrows). Scale bars: 1 µm (left panels) and 500 nm (right panel).
|
study
| 100.0 |
Ultrastructural analysis of the mouse cortex by TEM also revealed typical features of degenerating neurons. In particular, in the AAV-P301S mouse, we noticed axonal structures containing large abnormal accumulation of defective organelles or autophagic material at 7 months post-injection (Fig. 6h). For histological examination of neurodegeneration in the mouse forebrain, cresyl violet staining was used to reveal neuronal cell bodies in sagittal brain sections. In the animals injected with AAV vectors encoding either WT tau (4 out of 6 animals) or P301S tau (all the animals), there was an evident thinning of the cortical regions near the site of vector injection (Supplementary Fig. S2a). The hippocampus remained mostly intact. Degeneration of the mouse cortex was already observed at 1.5 and 3 months after vector injection. In contrast, there were no apparent morphological changes in the mice injected with the control AAV-maxFP vector at month 3, indicating that this effect was mainly caused by the human tau protein and not by mechanical injury (Supplementary Fig. S2a). In the AAV-2P injected group, only one out five mice showed a noticeable degeneration of the cortex. In the mice from the AAV-WT and AAV-P301S groups which displayed cortical neurodegeneration, it appeared that the most superficial layers of the cortex (layer I-V) were the most affected, as shown with a NeuN immunostaining to reveal cortical layers (Supplementary Fig. S2b).
|
study
| 100.0 |
Next, we sought to further determine the effects of the different tau variants on microtubules in the neurites of primary cortical neurons in vitro. To assess the protective effects of tau on microtubules, we induced defects by exposing the neurons to vinblastine, which interacts with tubulin heterodimers in the same location as tau and competes with tau binding31. When cultures of mouse cortical neurons overexpressing tau were exposed to 1 nM vinblastine for 24 h, we observed a dramatic increase in the number of neuritic filipodia, a clear indicator of microtubule severing (Fig. 7a and b). We did not observe any significant difference in the number of primary or secondary neuritic branches.Figure 72P-modified tau has protective effects on the microtubule network in primary cortical neurons exposed to vinblastine. Overexpression of Tau 2P prevents the formation of filopodia in mouse cortical neurons exposed to vinblastine. (a) Immunodetection of human tau (HT7) and GFP in cortical neurons (DIV5) infected with a bicistronic AAV-GFP/tau vector to overexpress each of the different human tau variants (WT, P301S or 2P). Nuclear expression of GFP is used to identify the transduced neurons. Upper panels show representative images of neurons in the control condition. Lower panels show cortical neurons after 24 h exposure to 1 nM vinblastine. Arrowheads indicate the presence of filopodia on the neuronal processes. Scale bars: 40 µm and 20 µm (lower panels). (b) Quantification of the number of filopodia normalized to neurite length. Exposure to 1 nM vinblastine increases the number of filopodia. Note that the overexpression of 2P tau significantly reduces the number of neuritic filopodia compared to WT and P301S tau. (c) Seven days exposure to 1 nM vinblastine leads to the formation of neuritic swellings. Experimental conditions and immunodetection are similar to panel (a). Lower panels show higher magnification of HT7 immunostaining in the neurites of neurons overexpressing human tau. Arrowheads indicate the presence of neuritic swellings. Scale bars: 100 µm and 20 µm (lower panels). (d) Quantification of the number of swellings normalized to neurite length. Exposure to 1 nM vinblastine increases the number of swellings in neurons overexpressing P301S tau. For all conditions: n = 3. Statistical analysis: two-way ANOVA followed by Tukey’s (b) or Fisher’s LSD (d) post-hoc tests, *p < 0.05; **p < 0.01; ***p < 0.001.
|
study
| 100.0 |
2P-modified tau has protective effects on the microtubule network in primary cortical neurons exposed to vinblastine. Overexpression of Tau 2P prevents the formation of filopodia in mouse cortical neurons exposed to vinblastine. (a) Immunodetection of human tau (HT7) and GFP in cortical neurons (DIV5) infected with a bicistronic AAV-GFP/tau vector to overexpress each of the different human tau variants (WT, P301S or 2P). Nuclear expression of GFP is used to identify the transduced neurons. Upper panels show representative images of neurons in the control condition. Lower panels show cortical neurons after 24 h exposure to 1 nM vinblastine. Arrowheads indicate the presence of filopodia on the neuronal processes. Scale bars: 40 µm and 20 µm (lower panels). (b) Quantification of the number of filopodia normalized to neurite length. Exposure to 1 nM vinblastine increases the number of filopodia. Note that the overexpression of 2P tau significantly reduces the number of neuritic filopodia compared to WT and P301S tau. (c) Seven days exposure to 1 nM vinblastine leads to the formation of neuritic swellings. Experimental conditions and immunodetection are similar to panel (a). Lower panels show higher magnification of HT7 immunostaining in the neurites of neurons overexpressing human tau. Arrowheads indicate the presence of neuritic swellings. Scale bars: 100 µm and 20 µm (lower panels). (d) Quantification of the number of swellings normalized to neurite length. Exposure to 1 nM vinblastine increases the number of swellings in neurons overexpressing P301S tau. For all conditions: n = 3. Statistical analysis: two-way ANOVA followed by Tukey’s (b) or Fisher’s LSD (d) post-hoc tests, *p < 0.05; **p < 0.01; ***p < 0.001.
|
study
| 100.0 |
In neurons transduced with AAV-tau, HT7 immunostaining was used to reveal the neuronal processes expressing human tau, and determine the frequency of filopodial protrusions in neurons expressing each of the different tau variants. A similar number of filopodia was observed in neurons overexpressing WT and P301S tau (Fig. 7a and b). Remarkably, the effect of vinblastine on the formation of filopodia was significantly diminished in neurons overexpressing 2P tau, compared to both WT and P301S tau. In neurons overexpressing 2P tau, the number of filopodia was not significantly different from the control neurons not exposed to vinblastine. This result indicates that this form of tau has enhanced protective effects against microtubule destabilization caused by vinblastine.
|
study
| 100.0 |
Following longer exposure to vinblastine (7 days), we also observed the formation of neuritic swellings in tau-overexpressing cortical neurons, and quantified this effect (Fig. 7c and d). The number of these neuritic swellings was significantly increased in neurons overexpressing P301S tau when treated with 1 nM vinblastine, as compared to neurons expressing either WT or 2P tau (Fig. 7d).
|
study
| 100.0 |
Overall, these results indicate that in presence of vinblastine, cortical neurons overexpressing either WT or P301S tau develop signs of microtubule severing and axonal damage including formation of filopodia (WT and P301S tau) and neuritic swellings (P301S tau). These effects are significantly decreased in neurons overexpressing 2P tau. The insertion of β-sheet breaking proline residues in regions flanking the second microtubule-binding repeat enhances the ability of tau to stabilize microtubules, most likely by increasing the ability of tau to compete with vinblastine for microtubule binding.
|
study
| 100.0 |
We compared the pathogenic effects of three variants of human tau with point mutations in the microtubule-binding domain that control the propensity of the protein to acquire a β-sheet conformation. Following ICV injection of AAV-tau in the mouse forebrain, we showed motor deficits in the rotarod task caused by tau overexpression. The pro-aggregant P301S tau mutant led to motor impairments that worsened over time. The WT form of tau also induced behavioral deficits, which however did not significantly progress. In contrast, tau overexpression had no effects on motor behavior when β-sheet breaking proline residues were introduced in the microtubule-binding region.
|
study
| 100.0 |
To address the possible mechanisms underlying tau-induced pathology, we determined the main tau species produced in each condition. P301S tau leads to the progressive accumulation of misfolded forms, as well as tau multimers, which correlates with the progression of motor deficits. In contrast, WT tau overexpression produces only low levels of aggregated tau. However, compared to P301S tau, the protein appeared to be more heavily phosphorylated at several specific residues, such as Ser 396, Ser 202/Thr 205, Thr 181 and Thr 231. With respect to WT tau, the 2P variant does has similar abundance of misfolded tau species. Instead, it is characterized by reduced the levels of tau phosphorylation. Consistent with its low levels of aggregation and phosphorylation, 2P tau promotes the stabilization of microtubules both in vivo and in cultures of cortical neurons exposed to vinblastine, which suggests that this form of tau is more prone to interact with microtubules.
|
study
| 100.0 |
The pathologic deposition of tau filaments mainly composed of hyperphosphorylated forms of the protein appears to correlate with the symptoms observed in FTDP and Alzheimer’s disease6–8. In the molecular processes leading to the tau pathology, the dissociation of the tau protein from the microtubules appears as a primary event, whereas tau hyperphosphorylation may prevent the re-association of the protein with the microtubule network. The amount of tau, with respect to the available sites for microtubule binding, determines the level of free tau protein in the cytosol, which is the starting point towards the formation of pathological tau, either hyperphosphorylated or misfolded into PHF, or both. Therefore, tau overexpression appears as a rational approach to promote the deposition of different forms of tau and assay how they contribute to pathology.
|
study
| 99.94 |
Here, tau was overexpressed following injection of AAV2/6 vectors in the lateral ventricles of neonatal mice. The main advantage of this approach is to induce widespread expression of the protein throughout the mouse forebrain, which is most prominently, but not exclusively observed in the long-projection pyramidal neurons located in the cortex and hippocampus. It is therefore possible to directly compare the pathogenic effects of various forms of tau, when overexpressed at similar levels. Remarkably, this model system leads to a prominent pathology, characterized by tau hyperphosphorylation and misfolding, which is already observed after a few weeks. In the normal brain, the tau protein appears to be transiently hyperphosphorylated during the early postnatal period, although the pattern of tau isoforms expressed during this period is different from the adult brain32–35. The overexpressed 4R0N tau isoform might as well be subject to hyperphosphorylation during the early postnatal period, initiating pathological changes that may persist or further progress during adulthood.
|
study
| 100.0 |
The pathology observed in the adult mice overexpressing each of the three variants of the 4R0N tau protein is summarized in Fig. 8. Remarkably, hyperphosphorylation is most prevalent on the WT form of tau, and appears to be reduced when proline residues are introduced in the microtubule-binding region of the tau protein (2P tau), or in presence of the pathogenic P301S mutation. In particular, the residues Ser396 (PHF1 epitope), Ser202/Thr 205 (AT8 epitope), Thr 181 (the most frequent phosphorylation site measured in clinical AD samples) and Thr 231 are more heavily phosphorylated on WT tau than on P301S and 2P tau. Notably, phosphorylation of Thr 231 can promote the dissociation of the tau protein form microtubules, via conformational changes induced by trans-to-cis isomerization36. P301S and 2P mutations in the microtubule-binding region reduce the propensity of the protein to become hyperphosphorylated and are therefore likely to facilitate tau interaction with microtubules, as suggested by the effect of 2P and P301S tau on microtubule bundling in vivo.Figure 8Motifs in the tau microtubule-binding domain affect the development of the tau pathology in vivo. Schematic representation of the pathology observed in the mouse forebrain following overexpression of each tau variant. Background colors represent the relative abundance and arrows indicate possible transitions between the different states.
|
study
| 100.0 |
Motifs in the tau microtubule-binding domain affect the development of the tau pathology in vivo. Schematic representation of the pathology observed in the mouse forebrain following overexpression of each tau variant. Background colors represent the relative abundance and arrows indicate possible transitions between the different states.
|
study
| 91.94 |
Via a β-sheet breaker effect, the 2P modification reduces the propensity of the ΔK280 pro-aggregation variant to accumulate in the sarkosyl-insoluble fraction37. The proline residues therefore decrease the toxicity of 2P ΔK280 tau overexpressed in transgenic mice. In the context of WT 4R0N tau, which poorly aggregates in physiological conditions, the 2P modification does not have any apparent effect on the level of MC1-positive or insoluble tau. In the absence of any pro-aggregation mutation, the amount of tau present in these fractions remains low and may not solely depend on β-sheet formation. However, our results support the possibility of a higher interaction of 2P tau with microtubules, possibly facilitated by its low phosphorylation status and/or the presence of these proline residues in key regions of the microtubule-binding domain. This is supported by the observation made in primary cultures of cortical neurons that 2P tau prevents the formation of filopodia in neurites exposed to vinblastine, a microtubule-destabilizing agent. Furthermore, this result suggests that 2P tau may compete more effectively than WT tau with vinblastine, to occupy the tau-binding site at the interface between α- and β-tubulin heterodimers31. Therefore, the conformation of the microtubule-binding domain might be critical for the interaction of tau with microtubules38.
|
study
| 100.0 |
Remarkably, the behavioral impairments observed in mice overexpressing WT tau are significantly reduced when two proline residues are introduced near the second repeat in the microtubule-binding region. Compared to 2P tau, the enhanced toxic effects of WT tau correlate with higher levels of tau phosphorylation on several residues. Indeed, it has been already demonstrated that the accumulation of hyperphosphorylated forms of tau, most likely dissociated from microtubules, can have detrimental effects in neurons39,40. However, these toxic effects are unlikely to be caused by aggregated tau, which remains at low levels in mice overexpressing WT 4R0N tau. Similarly, a mouse model overexpressing 4R1N tau displays cognitive and motor deficits in the absence of any neurofibrillary degeneration, suggesting that phospho-tau and pre-tangle species can cause neuronal dysfunction22. In the Drosophila, overexpression of 3R0N tau can destabilize microtubules via the formation of soluble hyperphosphorylated species that sequester endogenous tau11. Here, we have not observed any significant loss of axonal microtubules in the cortex following AAV-WT injection (see Fig. 6). It is therefore plausible that other mechanisms are implicated, such as disrupted mitochondrial transport, calcium dyshomeostasis and synaptic dysfunction, that may contribute to the induced behavioral deficits41.
|
study
| 100.0 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.