text
stringlengths 4
179k
| predicted_class
stringclasses 5
values | confidence
float64 0
100
|
|---|---|---|
The first main finding lends support to results from other birth cohort studies; although the strength of evidence for associations found in our study are somewhat weaker than those in other birth cohorts. This could reflect our use of self-report rather than medical records whereby non-TBI events may be recalled as TBI diluting the true exposure. We found that 7% of the cohort had experienced a TBI by age 16 years, this lies between the rate of 3.8% in the Northern Finland cohort and 31.6% in the CHDS . Mild TBI was associated with 39–67% increased risk of six of the seven outcomes, which is comparable to the 18–52% increased risk of poor adult outcomes reported by Sariaslan and colleagues . The increased odds of higher levels of alcohol consumption amongst the TBI group here is in keeping with the more frequent intoxication reported by 14 year olds with a TBI in The Northern Finland Birth Cohort . In the CHDS, a TBI requiring hospitalisation was associated with increased odds of externalising disorders and substance abuse at age 14 to 16 years and with increased odds of alcohol and drug dependence, and criminal behaviour at age 16 to 25 years . In the current investigation, the association between TBI and criminal offences did not remain once substance use was added as a covariate. McKinlay and colleagues reported a similar association for those injured before age 5 years; however, for those injured from age 6 to 15 years, a strong association remained for arrests and property offences, but not violent offences. They concluded that a certain threshold of TBI may be required for these effects to be seen . However, in our study it was not possible to index the severity of the TBI.
|
study
| 29.88 |
The prospective birth cohort design is a major strength of this study. Each injury was reported in close proximity to the time it happened, minimising the issue of recall bias. The longitudinal nature of the study allows for causal inference based on the temporal relationship between exposure and outcome. Additionally, the lack of statistical support and weak associations between our negative control group and the main outcomes, with the exception of committing offences, adds to the strength of evidence for a causal association suggested by previous research. On the other hand, the findings from the direct comparison between TBI and OI showing that TBI was only associated with hazardous alcohol use highlights the importance of exercising caution when interpreting findings on mild TBI without inclusion of a negative control group. However, the study is not without limitations. In particular, we were unable to obtain any index of severity based on the TBI measure, meaning that some nuances in effects based on severity may have been missed. For example, increased alcohol use has previously been related to mild but not moderate-to-severe TBI . Nonetheless the items used to identify TBI are similar to existing research, skull fractures based on ICD codes have been used to classify mild TBI elsewhere and self-report questions asking about loss of consciousness have also been utilised [32–34].
|
other
| 28.98 |
Evidence from cross-sectional work suggests that there is a relationship between mild TBI and risk behaviour in youth [4–8]; however, there is a paucity of high quality longitudinal research investigating this association . We have attempted to further explore a potential causal link by using data from a representative birth cohort and including a non-brain related injury group as a negative exposure control. Overall we found that participants who sustained a mild TBI before age 16 years were more likely than those with no injury or with a history of OI to use alcohol to problematic levels at age 17 years. Additionally, sustaining either a mild TBI or OI before age 16 years increased the likelihood of an individual committing offences at age 17 years. The study adds evidence for a possible causal association between mild TBI in youth and later hazardous alcohol use, and highlights the importance of including an extra injury group in mild TBI research.
|
review
| 27.31 |
Second, we included a negative control exposure group to increase the confidence that the associations seen between TBI and risk behaviour may be causal, where several previous studies have only used an uninjured control group . We found that participants who had sustained an OI were not at increased risk of problematic substance use or conduct problems compared to the no injury group, providing further support to previous literature. Interestingly, in a direct comparison between the two injury groups, the TBI group was only found to have a higher likelihood of hazardous alcohol use. This finding has implications for the treatment and management of youth post-TBI as alcohol use has previously been linked with recurrent TBI and poorer recovery from TBI . The lack of evidence for an association between TBI and the other risk outcomes when directly compared with OI highlights the importance of exercising caution when drawing conclusions about mild TBI from research that does not take other injuries into account. The association between OI and committing offences was an unexpected finding; one plausible explanation is that there may be common risk factors for both committing crimes and for being involved in accidents that result in physical injury. For example, sensation-seeking has previously been linked with both criminality and spinal cord injuries in a case–control study of 140 male spinal cord injury patients and 140 matched controls . Although both TBI and OI were associated with committing offences, only those with a TBI were more likely to have been in trouble with the police. Previously it has been suggested that having a TBI may be a risk factor for criminal behaviour and it may place an individual at a disadvantage during legal proceedings . Our finding raises the possibility that having a TBI may also be a factor in the initial transition into the legal system. Future studies in prison populations should measure the incidence rate of OI in addition to TBI in order to further explore this relationship.
|
study
| 30.25 |
Third, there were some differences in risk of outcomes for childhood and adolescent TBI. Childhood TBI (aged 0–11 years) was associated with conduct problems, while adolescent TBI (aged 12–16 years) was associated with increased likelihood of problematic alcohol and tobacco use, as well as criminality. Adolescent OI was associated with problematic tobacco use and committing offences, further highlighting a possible role for common risk factors mentioned above. TBI in both age groups showed weak association with cannabis use. Between these age ranges there was a change from parent-reported to self-reported TBI; however, we feel that this change is unlikely to have impacted the findings as it is more appropriate for the offspring to report their own injuries once they have entered secondary education. There may be some differences in severity of the injuries reported from childhood to adolescence—elsewhere the injuries occurring after 15 years were more severe —it would be interesting to assess this possibility in future research.
|
other
| 28.3 |
Approximately 50% of cutaneous melanomas have BRAF (V600E) mutation, and most of them respond well to BRAF inhibitor (BRAFi) treatment (vemurafenib or dabrafenib) (Chapman et al., 2011; Flaherty et al., 2010). Although the combination of BRAFi and MEK inhibitor (MEKi) can increase response rate and duration of response, most patients developed resistance to these treatments (Hauschild et al., 2012; Sosman et al., 2012). Numerous studies have found that several alternative pathways, such as activated COT, CRAF, IGF‐1R, PDGFRβ, and mutation of RAS, can bypass BRAF inhibition and subsequently trigger activation of downstream ERK/AKT (Nazarian et al., 2010; Poulikakos and Rosen, 2011; Villanueva et al., 2010). Nevertheless, these mechanisms occur in only 60% of progressing melanoma tumors. Additionally, resistance heterogeneity within melanoma tumors in patients presents a challenging clinical problem (Rizos et al., 2014; Shi et al., 2014). We have previously reported several unique changes related to bioenergetics which act in concert and make BRAFi‐resistant (BR) cells extremely vulnerable to arginine deprivation, regardless of whichever alternative signal pathways they utilize to evade the antitumor effect of BRAFi (Li et al., 2016). In this report, we have investigated the mechanisms leading to these changes.
|
other
| 32.1 |
Our previous studies have illustrated that arginine deprivation achieved by treatment with arginine deiminase (ADI‐PEG20) can suppress tumor growth in 70% melanomas due to their low or no expression of argininosuccinate synthetase 1 (ASS1), an essential enzyme needed to generate arginine from citrulline (Feun et al., 2008). Therefore, these melanoma cells lacking ASS1 expression must acquire exogenous arginine to support biosynthesis of polyamines and other amino acids for protein synthesis. Acquisition of exogenous arginine can be accomplished through cationic amino acid transporters CAT‐1 and CAT‐2 (Closs et al., 2004; Lu et al., 2013). In this communication, we have investigated alterations of these transporters in BR cells.
|
other
| 34.44 |
For autophagy assay, 10 μm hydroxychloroquine (HCQ; Sigma‐Aldrich, St. Louis, MO, USA) was prepared in medium and used to treat parental and BR cells. To investigate the mechanism of proteasomal degradation, we treated the parental and BR cells with a proteasome inhibitor MG‐132 (Selleck Chemicals) or an inhibitor of protein synthesis [cycloheximide (CHX); Sigma‐Aldrich] to observe the turnover of AMPK‐α1. MEKi (trametinib/GSK1120212) and MK‐2206 were purchased from Selleck Chemicals, and SCH772984 was from APExBio (Houston, TX,USA). These compounds were utilized to inhibit phosphorylation of MEK, ERK, and AKT, respectively, to determine their role in RNF44 expression. The arginine‐free medium was generated by adding ADI‐PEG20 (100 ng·mL−1; Polaris Pharmaceuticals, San Diego, CA, USA) into complete MEM medium at 37 °C and incubating for 48 h.
|
other
| 35.5 |
For immunoblotting, antibody against ubiquitin (Ub), AMPK‐α1, p‐AMPK (Thr172), Akt, p‐Akt (Thr308), p‐Akt (Ser473), ERK1/2, LC3‐I/II, or p‐CREB (Ser133) was purchased from Cell Signaling Technology (Danvers, MA, USA). Anti‐MAGE‐A3, anti‐ASS1, anti‐p‐ERK, anti‐AMPK‐β, anti‐AMPK‐r, anti‐TRIM28, and anti‐CREB antibodies were separately, obtained from Abgent (San Diego, CA, USA), BD Biosciences (San Jose, CA, USA), Polaris, Sigma‐Aldrich, and GeneTex (Irvine, CA, USA). Anti‐RNF44 antibody was purchased from Abcam (Cambridge, MA, USA). All secondary antibodies conjugated with HRP were purchased from Promega (Madison, WI, USA). The immunoblots were developed using ultrasensitive enhanced chemiluminescent substrate and visualized by ChemiDoc MP System (Bio‐Rad, Hercules, CA, USA). The anti‐CREB antibody used for chromatin immunoprecipitation (ChIP) was purchased from Cell Signaling Technology.
|
other
| 33.78 |
ADI‐PEG20, a mycoplasma enzyme converting exogenous arginine to citrulline and ammonia, has shown antitumor effects in many cancer types (Feun et al., 2008; Kelly et al., 2012; Miraki‐Moud et al., 2015; Qiu et al., 2014), yet cancer cells survive after this treatment due to ASS1 re‐expression and undergoing autophagy (Tsai et al., 2012; You et al., 2013). Previously, it has been shown that ASS1 re‐expression is due to upregulation of positive regulator c‐Myc after ADI‐PEG20 treatment (Tsai et al., 2012). Conversely, attenuated c‐Myc‐mediated ASS1 expression occurs in BR cells and therefore increases sensitivity to ADI‐PEG20 treatment (Li et al., 2016). To further enhance their ability to survive, naïve melanoma cells undergo autophagy through AMP‐activated protein kinase (AMPK) activation (Savaraj et al., 2010; Yang et al., 2011). Activated AMPK phosphorylates ULK directly or through mechanistic target of rapamycin (mTOR) inhibition and subsequently triggers autophagy (Jeyabalan et al., 2012; Savaraj et al., 2010; Yang et al., 2011).
|
study
| 29.56 |
LKB‐AMPK axis has been regarded as a master energy sensor. Loss of LKB1 has been shown to induce mouse embryonic fibroblast (MEF) apoptosis in response to nutrient stress due to inability to activate AMPK‐mediated autophagy (Shaw et al., 2004). In BRAF‐mutant melanoma cells, activated BRAF constitutively phosphorylates ERK and ribosomal s6 kinase (RSK) which phosphorylate LKB1 (Ser325 and Ser428) and in turn suppress its capability to activate AMPK in melanomas (Esteve‐Puig et al., 2009; Zheng et al., 2009). In contrast, another study revealed that BRAFi cannot restore LKB1‐AMPK activation and triggers LKB1‐AMPK‐independent autophagy (Ma et al., 2014). Therefore, whether BRAF‐ERK suppresses LKB1‐AMPK activation remains unclear in BR cells. Another means to regulate AMPK activity is through its degradation via ubiquitin‐proteasome system (UPS) (Zungu et al., 2011). Currently, it is known that AMPK is composed of α, β, and γ subunits. Recent data suggested that AMPK‐α1 (possessing Thr172 phosphorylation site) can be degraded by MAGE‐A3/6‐TRIM28, which leads to tumorigenesis (Pineda et al., 2015). Besides autophagy, AMPK is also a vital regulator of metabolism. Activated AMPK phosphorylates acetyl‐CoA carboxylase (ACC) that triggers fatty acid oxidation to generate energy and terminates fatty acid biosynthesis (Hardie and Pan, 2002). Stimulation of AMPK using agonists can enhance glucose uptake and glycolysis through upregulation of GLUT1 (glucose transporter) and hexokinase II (HK II) (Hardie, 2011; Hardie et al., 2012; Wu and Wei, 2012). Overall, AMPK is a major regulator of both metabolism and autophagy.
|
study
| 29.94 |
This study uncovered that BRAFi resistance downregulates AMPK activity through UPS‐mediated AMPK‐α1 degradation. We further identified RING finger 44 (RNF44), a novel E3 ligase responsible for AMPK‐α1 degradation. Downregulation of AMPK‐α1 switches glucose dependence toward arginine dependence via attenuated GLUT1 and significantly upregulated arginine transporter CAT‐2 expression. Under arginine starvation, ASS1‐negative BR cells are unable to efficiently utilize glucose, synthesize arginine, and undergo autophagy to survive. Hence, they are more sensitive to arginine deprivation than their parental counterparts.
|
study
| 29.2 |
The BRAF‐mutant (V600E) melanoma cell lines were incubated with vemurafenib (Selleck Chemicals, Houston, TX, USA) over 30 weeks to generate BR cell lines. IC50 values of vemurafenib for parental and BR cells have been described in the previous study (Li et al., 2016). The parental cell lines (A375, A2058, UACC62, and SK‐MEL28) and skin fibroblast cell line BJ were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA); MEL‐1220 and MEL‐DA were established in our laboratory. BRAFi/MEKi‐resistant (BMR) melanoma cell lines (A375BMR SK‐MEL28BMR, and A2058BMR) were created by incubating their parental cells with the combination of vemurafenib and IC50 values of trametinib (20 nm for A2058, 10 nm for A375, and 1 nm for SK‐MEL28). Except for SK‐MEL‐28 (cultured in DMEM; Thermo Fisher Scientific), all melanoma cell lines were cultured in MEM supplemented with 10% FBS (Atlanta Biologicals, Flowery Branch, GA, USA) and streptomycin/penicillin (Thermo Fisher Scientific, Waltham, MA, USA) in CO2 incubator.
|
other
| 32.66 |
For CAT‐1 and CAT‐2 detection, melanoma cells were incubated with primary antibody (CAT‐1, Novus, Littleton, CO, USA, 1 : 20; CAT‐2, Santa Cruz Biotechnology, Santa Cruz, CA, USA, 1 : 20) for 20 min and then incubated with second antibody conjugated with Alexa Fluor® 555 (Thermo Fisher Scientific) for 20 min at room temperature. The samples were analyzed by FACS (BD Accuri™ C6).
|
other
| 39.03 |
The plasmid inserted with RNF44‐GFP or GFP gene was purchased from OriGene (Rockville, MD, USA) and then mixed with lipofectamine (Thermo Fisher Scientific) for transfection. Regarding knockdown experiment, individual nucleotides (siRNAs) targeting AMPK‐α1, RNF44, CAT‐2, and nontargeting (NT) control siRNAs were obtained from Dharmacon (Lafayette, CO, USA) and OriGene. These siRNAs were delivered into the cells using transfection reagent INTERFERrin (Polyplus, New York, NY, USA) according to the instruction of the manufacturer.
|
other
| 31.5 |
Total RNA was extracted using Trizole reagent (Thermo Fisher Scientific) and converted into cDNA using iScript cDNA synthesis kit (Bio‐Rad). The cDNA was mixed with SYBR Green SuperMix reagent (Bio‐Rad) and gene‐specific PCR primers and analyzed by real‐time PCR analysis (CFX96; Bio‐Rad). The gene expression was normalized by GAPDH.
|
other
| 42.06 |
Primer sequences for AMPK‐α1 are 5′‐GGTCCATAGAGATTTGAAACCTG‐3′ (forward) and 5′‐GCCTGCATACAATCTTCCTG‐3′ (reverse); primer sequences for CAT‐1 are 5′‐CTTCATCACCGGCTGGAACT‐3′ (forward) and 5′‐GGGTCTGCCTATCAGCTCGT‐3′ (reverse); primer sequences for CAT‐2 are 5′‐TTCTCTCTGCGCCTTGTCAA‐3′ (forward) and 5′‐TCTAAACAGTAAGCCSTCCCGG‐3′ (reverse); primer sequences for RNF44 are 5′‐CCTACTTCCTCTCGATGCTG‐3′ (forward) and 5′‐CTGCTCTATGTCTGCTTTGG‐3′ (reverse); primer sequences for GAPDH are 5′‐CTCTCTGCTCCTCCTGTTC‐3′ (forward) and 5′‐GGTGTCTGAGCGATGTGG‐3′ (reverse).
|
other
| 36.8 |
For site‐directed mutagenesis, the sequences of mutagenic primers recognizing CREB binding sites are 5′‐GCGGTTAAATGTCTCTGTGATAGGAGCGCGAGCAGGGC‐3′ (first CREB binding site) and 5′‐GCAGCTCTTGGGGGTGATAGGATCTCCGGGAAGGTG‐3′ (second CREB binding site). The assay was carried out using the kit purchased from Agilent Technology (Santa Clara, CA, USA) per manufacturer's instruction.
|
other
| 37.9 |
The ChIP assay kit was purchased from Millipore (Burlington, MA, USA). The PCR primer sequences for first CREB binding site are 5′‐CGCCGTGTAGTATAAACAAGGAG‐3′ (forward) and 5′‐ACAGGGTGCCGCCTGAGATACT‐3′ (reverse); primer sequences for second CREB binding site are 5′‐AGGGAGGTCTCCGCGGGGAC‐3′ (forward) and 5′‐ACGAGCTAACGTCTGCCGGGC‐3′ (reverse).
|
other
| 35.22 |
Before amplifying varying DNA fragments located in RNF44 promoter (−1785 to +120) using Bio‐Rad CFX PCR System and High‐Fidelity PCR SuperMix (Thermo Fisher Scientific), we designed seven primers including forward primers extended with KpnI and a reverse primer extended with NheI for DNA amplification. The forward primer sequence for region −33 to +122 is 5′‐TATGCAGGTACCCGAGTGATTGGCTCTCAGGG‐3′; the forward primer sequence for region −118 to +122 is 5′‐TATGCAGGTACCGGTCCCTTTAAGTGCGAAGGT‐3′; the forward primer sequence for region −423 to +122 is 5′‐TATGCAGGTACCGTTGTTAGAGCCAGCATAACCA‐3′; the forward primer sequence for region −530 to +122 is 5′‐TATGCAGGTACCCGCCGTGTAGTATAAACAAGGAG‐3′; the forward primer sequence for region −1122 to +122 is 5′‐TATGCAGGTACCATTATTCGACCTGTTCCAGCTC‐3′; the forward primer sequence for region −1785 to +122 is 5′‐TATGCAGGTACCTGACCTTGCTCTTGTGTTGCT‐3′; the common reverse primer sequence at +122 is 5′‐ATTCGTGCTAGCTTGATTCACAACATTCGAAGCGG‐3′. The PCR products and pGL3‐based vector carrying luciferase (LUC) gene (Promega) were, respectively, digested with enzyme KpnI/NheI. Afterward, DNA fragments and linear pGL3 plasmid were ligated together using T4 ligase (New England Biolab, Ipswich, MA, USA).
|
other
| 33.1 |
Regarding LUC activity assay, empty vector pGL3 and various RNF44 promoter constructs were delivered into melanoma cells using Lipofectamine (Thermo Fisher Scientific) for 6 h and then cultured in the presence of inhibitors for 24 h followed by detecting LUC activity using luciferase assay system kit (Promega).
|
other
| 41.6 |
Both parental and BR melanoma cells were cultured in 24‐well plates and transfected with siRNA against CAT‐2 (OriGene) overnight. Afterward, these transfectants were transferred to 96‐well plates (1 × 104/well) and incubated at 37 °C for 24 h. For ATP extraction, trichloroacetic acid (TCA, 1%) and 2 mm EDTA were added into transfectants and then were neutralized with 20 mm Tris/acetate (pH 7.8). Intracellular ATP concentration was detected using the ENLITEN® ATP assay system bioluminescence detection kit (Promega) and normalized by protein content.
|
other
| 40.1 |
5 × 103 melanoma cells were incubated with various doses of ADI‐PEG20 (0–1000 ng·mL−1) for 72 h. The cell proliferation was analyzed by MTT (Sigma‐Aldrich). The IC50 values of ADI‐PEG20 in both parental and BR cells have been shown in our previous study (Li et al., 2016). Apoptosis was determined using caspase activity assay kit (ApoSat apoptosis detection kit; R&D System, Minneapolis, MN, USA). The apoptotic proportion was analyzed by FACS (Accuri™ C6; BD Biosciences). Lysotracker Red (Thermo Fisher Scientific) and Cyto‐ID (Enzo Life Sciences, Farmingdale, NY, USA) were applied to autophagosome staining.
|
other
| 37.44 |
Cell lysates were collected from parental and BR cells after incubated with or without ADI‐PEG20 (100 ng·mL−1) in the presence of MG‐132 (10 μm) for 4 h. Immunoprecipitation was completed by adding anti‐AMPK‐α1 antibody (Santa Cruz Biotechnology) and Gammabind plus Sepharose bead slurry (GE Healthcare, Life Science, Marlborough, MA, USA) into the cell lysates. Immunoprecipitates were subjected to SDS/PAGE and immunoblotting. For proteomic analysis, we sent the immunoprecipitates of A2058 and A2058BR cells to Applied Biomics, Inc. (San Francisco, CA, USA), to identify the putative proteins interacting with AMPK‐α1 using two‐dimensional difference gel electrophoresis (2D‐DIGE) and matrix‐assisted laser absorption ionization‐time of flight mass spectrometry (MALDI‐TOF‐MS). The data were analyzed using GPS explorer equipped with search engine MASCOT.
|
other
| 41.6 |
The protocol of in vivo experiment has been reviewed and approved by the Institutional Animal Care and Use Committee (IACUC, #7715.63MR) at Miami VA Medical Center. 1 × 106 cells were injected subcutaneously into female athymic nude‐Foxn1nu mice (6‐8 weeks) purchased from Harlan Laboratories (Indianapolis, IN, USA). When the tumor volumes reached 100 mm3, the tumor‐bearing mice were randomly assigned to the control group or the experimental group. The experimental group received an intramuscular injection of ADI‐PEG20 (100 IU·kg−1), and the control group was treated with normal saline twice per week.
|
other
| 38.47 |
The tissue slides were dewaxed by xylene. Antigen retrieval was performed using citric acid (10 mm, pH 6.0). The tumor tissue slides were separately incubated with anti‐ASS1 (Polaris), anti‐RNF44 (Novus, 1 : 200), anti‐CAT‐1 (Novus, 1 : 50), anti‐CAT‐2 (Novus, 1 : 50), and anti‐AMPK‐α1 (Novus, 1 : 200) antibodies at 4 °C overnight. The slides then were stained with LSAB™2 Kits (DAKO, Carpinteria, CA, USA) and hematoxylin (DAKO) and visualized by a light microscope (Olympus, Center Valley, PA, USA). The levels of ASS1, RNF44, and AMPK‐α1 were randomly scored upon intensity scale ranging from 0 to 3+ and percentage of positive cells in tumor tissues. The outcome was based on scoring (H‐score) proposed by K.S. McCarty (McCarty et al., 1986). We were blinded to the allocation of pretreatment and post‐treatment groups in this experiment.
|
other
| 40.78 |
Briefly, cells were fixed in 2.5% glutaraldehyde at room temperature. The specimens were postfixed in 1% osmium tetroxide (OsO4) for 10 min, dehydrated using a graded ethanol series, en bloc stained with 2% uranyl acetate in 50% ethanol for 30 min, and embedded in Spurr's epoxy resin. Thereafter, ultrathin (< 90 nm) sections were cut using a Diatome 3‐mm diamond knife on the Leica EM UC6 ultramicrotome. The ultrathin sections were stained using lead citrate to be viewed under TEM on a Jeol 1400 EM at 80 kV.
|
other
| 39.53 |
Tumor tissues were obtained after obtaining informed consent approved by Institutional Review Board (#19990582) at the University of Miami Miller School of Medicine. With respect to explant assay, the tumor tissues obtained from patients were cut into small pieces (average diameter is < 0.2 cm) and then randomly seeded into transwells hanging on 24‐well plates (lower compartments) containing medium and drugs and incubated for 48–72 h.
|
other
| 37.12 |
Statistical analysis was performed by Student's t‐test and assessed using GraphPad Prism (La Jolla, CA, USA). All data were shown as mean ± standard error of mean (SEM). Every result was completed by three independent experiments. P‐value < 0.05 was regarded as significant difference.
|
other
| 36.44 |
We have generated five BR cell lines and three BMR cell lines from parental cell lines (A375, A2058, MEL‐1220, SK‐MEL‐28, and UACC‐62) harboring BRAF (V600E) mutation by long‐term exposure to IC50 values of vemurafenib or in combination with trametinib as stated in Table S1.
|
other
| 30.72 |
A previous study has shown that BR cells switch dependence on glycolysis to mitochondrial oxidative phosphorylation (Baenke et al., 2016). To examine whether low activity of glycolysis is also seen in our BR cells, we determined glucose uptake and key enzymes in the glycolytic pathway. The results substantiated that GLUT1 and hexokinase II (HK II) were downregulated in BR cells (Fig. 1A), corresponding to less glucose uptake assayed by 2‐NBDG compared to parental cells (Fig. 1B). Our previous study has reported that BR cells are extremely auxotrophic for arginine due to downregulated c‐Myc‐mediated ASS1 re‐expression (Li et al., 2016). Therefore, we hypothesized that these BR cells failing to re‐express ASS1 to generate arginine must increase the ability to acquire exogenous arginine through upregulation of arginine transporters CAT‐1 and CAT‐2 to survive. To test this hypothesis, we first examined whether ASS1 appearance affects expressions of CAT‐1 and CAT‐2 in A375 and A2058 cells (ASS1‐negative cells). The plasmid carrying ASS1 described in the previous publication (Long et al., 2013) was delivered to these two cell lines. The result showed that ASS1 overexpression significantly downregulated CAT‐2 rather than CAT‐1 in these melanoma cells (Fig. S1A,B). We next observed their expressions in parental cells and BR cells. As expected, BR clearly exhibited higher CAT‐2 expression compared to their parental cells (Fig. 1C and Fig. S2). To verify whether arginine acquisition mediated by CAT‐2 is critical for ATP synthesis, CAT‐2 expression in A2058BR and A375BR cells was silenced using individual siRNAs, and intracellular ATP concentration has been analyzed. Consistent with the previous study reported by Baenke et al. (2016), the basal levels of ATP in BR cells were much higher than those in parental counterparts due to the fact that BR cells utilized mitochondrial oxidative phosphorylation instead of glycolysis to generate ATP (Fig. S1C). Ablation of CAT‐2 remarkably reduced ATP synthesis as well as cell viability in BR cells (Fig. S1C,D). Collectively, our findings suggested that BR cells are dependent on arginine more than glycolysis as their energy source.
|
other
| 29.61 |
BRAFi resistance results in metabolic reprogramming toward arginine addiction and circumvention of autophagy due to downregulation of AMPK‐α1. Parental and BR cells were incubated with ADI‐PEG20 (100 ng·mL−1) for 72 h (A) or HCQ alone with or without for 0–72 h (D). Their cell lysates were subjected to immunoblotting. (B) Glucose uptake activity in parental and BR cells was analyzed by 2‐NBDG uptake using flow cytometry (FACS). (C) RNA levels of CAT‐1 and CAT‐2 were detected by qRT‐PCR. On the other hand, MEL‐1220 and A2058 cells were transfected with individual siRNAs (50 nm) against AMPK‐α1, nontargeting (NT) siRNA, or transfection reagent alone (vehicle, Veh). These transfectants were analyzed using immunoblotting (E,F), glucose uptake (G), MTT assay (# P < 0.05 and ## P < 0.01) (H) and qRT‐PCR (I). Data are represented as mean ± SEM (n = 3, *P < 0.05, **P < 0.01, and ***P < 0.005). Fig. 1A reproduced from our previous study (Li et al., 2016).
|
other
| 31.78 |
Our data showed that BR cells engaged in the acquisition of exogenous arginine instead of utilizing glucose. Both BR and BMR cells underwent apoptosis while encountering arginine depletion achieved by ADI‐PEG20 (100 ng·mL−1) treatment as shown by a drastic increase in caspase activity (Table S1). Moreover, ASS1 expression in some cell lines (e.g., A2058, SK‐MEL‐28, and UACC‐62) can be induced upon ADI‐PEG20 treatment and hence contributes to resistance to arginine depletion. An impaired ability to re‐express ASS1 in their BR/BMR cell lines increased sensitivity to ADI‐PEG20 treatment (Table S1) (Li et al., 2016). However, other cell lines, such as A375, MEL‐1220, and their BR/BMR cell lines do not have inducible ASS1, yet their BR/BMR cells were more sensitive to ADI‐PEG20 treatment. Thus, CAT‐2‐mediated arginine acquisition is critically important and may be regulated by other molecules in these ASS1‐negative BR/BMR cells. Besides dependency on arginine acquisition, attenuated autophagy in these BR/BMR cells could be the contributory factor of hypersensitivity to ADI‐PEG 20 treatment.
|
other
| 28.83 |
Low protein levels of AMPK‐α1 have been found in BR cells (Fig. 1A) and did not correlate with RNA levels (Fig. S3A). Thus, we speculated that more active AMPK‐α1 protein degradation occurs in BR cells. We treated parental and BR cells with proteasome inhibitor MG‐132 or a protein synthesis inhibitor CHX followed by immunoblotting of AMPK‐α1. As shown in Fig. 2A, compared with parental cells, BR cells displayed a delayed accumulation of AMPK‐α1 expression following treatment with MG‐132 but the faster degradation of AMPK‐α1 after CHX treatment. Furthermore, more ubiquitins were co‐immunoprecipitated with AMPK‐α1 in BR cells relative to parental cells, even in the presence of ADI‐PEG20 (Fig. 2B). Taken together, our results verified that BRAFi resistance enhances activation of AMPK‐α1 degradation through ubiquitin‐proteasome system (UPS).
|
other
| 29.12 |
The E3 ligase RNF44 is implicated in UPS of AMPK‐α1 degradation in BR cells. (A) Parental cell and BR cells were incubated with MG‐132 (10 μm) and CHX (80 μg·mL−1), respectively. Cell lysates were collected at different time intervals, and subsequently, AMPK‐α1 was assayed by immunoblotting. The levels of AMPK‐α1 were quantitated by ImageJ and presented as curves. (B) Parental and BR cells were incubated with or without ADI‐PEG20 in the presence of MG‐132 (10 μm) for 4 h. Ubiquitin (Ub) and AMPK‐α1 were separately detected by immunoblotting following immunoprecipitation of AMPK‐α1. (C) The precipitated proteins from A2058 cells and A2058BR cells were, respectively, labeled with Cy3 and Cy5 and then were subjected to 2D gel. Thereafter, the Cy5‐positive spot was identified as RNF44 (Q7L0R7) by MALDI‐TOF MS based on UniProtKB/Swiss‐Prot database. (D) The levels of different E3 ligases (RNF44, Cidea, MAGE‐A3, and MuRF1) in precipitated proteins (D) or total cell lysates (E) were detected by immunoblotting. (F) A2058 cells were constantly treated with vemurafenib (5 μm), and its cell lysates were collected at different time points. The levels of AMPK‐α1, RNF44, or actin were detected by immunoblotting.
|
other
| 31.47 |
Our previous data have shown that naïve melanoma cells undergo autophagy to survive arginine deprivation (Savaraj et al., 2010), whereas BR cells lose this ability (Li et al., 2016). To support these results, both parental and BR cells were treated with HCQ (10 μm) to inhibit autolysosome formation by increasing pH in lysosomes and result in autophagosome accumulation presented by an increase in LC3‐I/II conversion. Our data depicted that LC3‐ll expression increased over time, and LC3‐II accumulation was faster in parental cells rather than in BR cells following treatment with HCQ. Although the combination of HCQ and ADI‐PEG20 enhanced LC3‐I/II expression, LC3‐II expression remained lower in BR cells and was diminished after 48 h due to undergoing apoptosis (Fig. 1D). Furthermore, ADI‐PEG20 treatment promotes p‐AMPK and AMPK‐α1 expression in parental cells, but lower levels of p‐AMPK and AMPK‐α1 were seen in BR cells (Fig. 1A). Thus, our results suggested that lower levels of AMPK‐α1 disable BR cells to survive by undergoing autophagy under arginine deprivation, and hence, they succumb to apoptosis.
|
other
| 29.44 |
Our previous study reported that AMPK activation in BR cells is p‐LKB independent (Li et al., 2016). Low levels of p‐AMPK (Thr172) correlated with low levels of AMPK‐α1 (possessing Thr172), but not with AMPK‐α2, β, and γ subunits, resulting in downregulated p‐ACC expression (Fig. 1A). As AMPK has been reported to regulate autophagy, glucose uptake, and glycolysis (Hardie and Pan, 2002; Hardie et al., 2012), we next examined whether downregulation of AMPK‐α1 can abort glycolysis and ADI‐PEG20‐induced autophagy in parental melanoma cells as seen in BR cells. The results demonstrated that knockdown of AMPK‐α1 using #1 or #2 siRNA (Fig. 1E) in parental cells yielded greater sensitivity to ADI‐PEG20 treatment when compared to nontargeting (NT) siRNA or vehicle (Fig. 1H). These transfectants also had less ability to undergo autophagy as evidenced by decreased LC3‐II expression and autophagosome formation upon arginine starvation (Fig. S3B,C). In regard to metabolic regulation, silencing AMPK‐α1 attenuated glucose uptake and disturbed fatty acid metabolism through downregulation of GLUT1 and p‐ACC, but did not affect ASS1 expression (Fig. 1F,G). Interestingly, arginine transporter CAT‐2 expression is also greatly increased by silencing AMPK‐α1 in parental cells, while only slightly enhanced CAT‐1 expression is detected (Fig. 1I and Fig. S3D). In contrast, BR cells transfected with AMPK‐α1 (PRKAA1)‐GFP‐overexpressed plasmid possessed higher GLUT1 expression compared to vehicle (GFP groups) (Fig. S4A). Moreover, in vitro study also confirmed that PRKAA1‐GFP overexpression restored autophagy in BR cells and hence rendered BR cells resistant to arginine depletion (Li et al., 2016). This evidence was further confirmed by the xenograft model demonstrating that (PRKAA1)‐GFP overexpression subverted ADI‐PEG20‐induced apoptosis in A2058BR by enhancing autophagosome formation (Fig. S4C,D). Furthermore, PRKAA1 overexpression strikingly attenuated RNA and protein levels of CAT‐2 in A2058BR and MEL‐1220BR cells and xenografts (Figs S4 and S5B). Although AMPK‐α1 overexpression slightly suppressed CAT‐1 in A2058BR xenografts, the basal levels of CAT‐1 were low so that there was no significant reduction seen in RNA levels. Overall, our data suggested that downregulated AMPK‐α1 expression in BR cells not only abrogates the autophagy but also switches metabolism to arginine addiction.
|
review
| 31.06 |
Upregulated RNF44 expression is related to hyperactivation of ERK and AKT in BR cells. (A) Immunoblotting demonstrated that BR and BMR cells possessed downregulated AMPK‐α1 and upregulated RNF44. (B) RNA levels of RNF44 in parental and BR cell lines were determined by qRT‐PCR and normalized by GAPDH. (C) Higher protein levels of p‐AKT, p‐ERK, and RNF44 appeared in BR cells even in the presence of BRAFi (vemurafenib). (D) BR cell lines were treated with MEKi (trametinib, 5 nm), ERKi (SCH772984, 3 μm), or AKTi (MK‐2206, 2 μm) for 24 h, and their RNF44 levels were determined by qRT‐PCR. Data shown in a bar graph were represented as mean ± SEM (n = 3, *P < 0.05, **P < 0.01, and ***P < 0.005). (E) Protein levels of ERK, AKT, AMPK‐α1, and RNF44 in BR and BMR cells were determined by immunoblotting following treatment with AKTi or ERKi.
|
other
| 30.2 |
As we have demonstrated that resistance to BRAFi initiates RNF44 transcription, we next examined the underlying transcriptional regulation of RNF44. Firstly, we predicted that RNF44 promoter region contains E‐box, GC‐box (Sp1 binding site), and two CREB binding sites (Fig. 5A) based on decipherment of DNA elements (DECODE) database. Secondly, we constructed pGL3 LUC reporter vectors with various lengths of RNF44 promoter region and in turn detected transcriptional activity by LUC reporter assay. The result revealed that the LUC activity of pGL3‐530 containing first CREB binding site was highest among RNF44 promoter constructs in A2058BR cells incubated with BRAFi (Fig. 5B). Moreover, it has been known that CREB can be activated by RAF/ERK/RSK and AKT through phosphorylation at Ser133 (Chang et al., 2003; Du and Montminy, 1998). Hence, we speculated that BRAFi resistance‐induced hyperactivation of ERK/AKT elicits more p‐CREB (Ser133) to initiate RNF44 transcription. Thirdly, we determined p‐CREB (Ser133) expression in A2058 and A2058BR cells treated with or without BRAFi. As expected, the addition of BRAFi resulted in enhanced p‐CREB expression in A2058BR cells but downregulated p‐CREB in A2058 cells (Fig. 5D). To further test whether inhibition of ERK or AKT deters CREB from triggering RNF44 transcription, A2058BR transfected with various constructs was treated with AKTi or ERKi. The LUC activity of pGL3‐530 and the levels of p‐CREB were dramatically decreased by AKT or ERK antagonist (Fig. 5C,D).
|
study
| 28.16 |
We then sought for ad hoc proteins implicated in UPS using immunoprecipitation of AMPK‐α1 followed by proteomic analyses. Notably, our proteomic analyses identified a novel protein, RNF44, which was 4.9‐fold higher in A2058BR immunoprecipitates relative to A2058 immunoprecipitates (Fig. 2C). Even though RNF44 has been categorized in the RING finger family, its biological functions have not been identified yet. Hence, we searched for putative proteins sharing similar peptide sequences with RNF44 in UniProKB/Swiss‐Prot database and then found E3 ligases RNF38 and praja‐1 (< 30% similarity) (Fig. S6A). Additionally, higher RNF44 expression seen in BR cell lysates can be co‐immunoprecipitated with AMPK‐α1 compared to parental cells even in the presence of ADI‐PEG20 (Fig. 2D,E). The time‐course experiment showed that RNF44 levels increased stepwise and inversely correlated with AMPK‐α1 levels when A2058 cells were exposed to BRAFi and gradually became BR cells (Fig. 2F). Currently, several possible ubiquitin ligases of AMPK‐α1 including Cidea, MuRF1, and MAGE‐A3/A6‐TRIM28 have been reported to ubiquitinate AMPK‐α or β in muscle cells, adipose tissue, cervical cancer, lung cancer, and colon cancer cells (Pineda et al., 2015; Zungu et al., 2011). However, none of these candidate proteins correlated with AMPK‐α1 expression and cannot be co‐immunoprecipitated with AMPK‐α1 in BR cells (Fig. 2D,E).
|
study
| 28.17 |
To investigate whether RNF44 is involved in ubiquitination of AMPK‐α1, RNF44 expression in A2058BR cells was silenced using siRNAs. The result illustrated that silencing RNF44 expression enhanced AMPK‐α1 by abolishing its ubiquitination, yet it cannot be applied to the expression of α2, β, or γ subunit (Fig. 3A,B). As a result, these transfectants regained resistance to ADI‐PEG20 treatment by triggering autophagosome formation (Fig. 3C,D). For metabolic alteration, silencing RNF44 resulted in enhanced glucose uptake and fatty acid oxidation as evidenced by upregulation of AMPK‐mediated GLUT1 and HKII expressions and ACC phosphorylation, attenuated arginine transporter CAT‐2 expression, but did not affect ASS1 expression (Fig. 3A,E,F). Conversely, A2058 cells transfected with RNF44‐GFP‐overexpressed plasmid exhibited robust ubiquitination of AMPK‐α1 and consequently resulted in sensitization to arginine deprivation (Fig. S6B–D). Collectively, our data illustrated that RNF44 participates in AMPK‐α1 degradation in BR melanoma cells, resulting in downregulation of autophagy and glucose uptake but upregulation of CAT‐2. These biochemical alterations make melanoma cells hypersensitive to arginine deprivation.
|
other
| 27.9 |
Silencing RNF44 in BR cells abrogates AMPK‐α1 degradation, switches metabolism to glucose addiction, and restores autophagosome formation upon arginine deprivation. BR cells were transfected with individual siRNAs against RNF44, nontargeting (NT) siRNA (50 nm), or transfection reagent alone (Veh, as a control group). Total cell lysates of these transfectants were subjected to immunoblotting (A) and immunoprecipitation of AMPK‐α1 (B). (C) A2058BR cells were transfected with RNF44 siRNAs and treated with ADI‐PEG20. Their autophagosomes and nuclei were separately stained with Cyto‐ID (green) and Hoechst 33342 (blue) and then visualized using fluorescent microscope (scale bar = 50 μm). Autophagy‐positive cells were quantitated and presented as a bar graph. (D) Cell viability of these transfectants was analyzed by MTT after treatment with ADI‐PEG20 (100 ng·mL−1) or completed medium for 48–72 h. (E) Glucose uptake was analyzed by 2‐NBDG uptake with FACS. (F) CAT‐1 and CAT‐2 expressions were assessed by FACS, and data were shown in a bar graph (n = 3, *P < 0.05, **P < 0.01, and ***P < 0.005).
|
other
| 34.9 |
Currently, there is no published literature on how RNF44 is upregulated in BR cells. As higher levels of RNF44 seen in BR cells may be due to increased transcription or translation, we determined RNA levels of RNF44 by qRT‐PCR. Our results showed that higher RNF44 RNA levels contributed to higher RNF44 protein levels (Fig. 4A,B). As the majority of BR cells expressed high levels of phosphorylated ERK/AKT (Fig. 4C) (Welsh et al., 2016), we hypothesized that ERK/AKT hyperactivation in BR cells may upregulate RNF44 expression. To test this hypothesis, we treated BR cells with ERK inhibitor (ERKi, SCH772984), MEKi (trametinib), and AKT inhibitor (AKTi, MK‐2206) and determined RNF44 levels. The results revealed that AKT or ERK inhibition attenuated RNF44 expression in A375BR and A2058BR cells (Fig. 4D,E). This correlation can also be seen in A375BMR and A2058BMR cells. Our findings suggested that ERK/AKT hyperactivation may contribute to elevated RNF44 expression, leading to ubiquitination of AMPK‐α1.
|
other
| 29.6 |
ERK/AKT‐activated CREB triggers transcription via two cis‐regulatory elements of RNF44. (A) Delineation of RNF44 promoter region (−1785 to +120) and its deletions. (B,C) A2058 and A2058BR cells transfected with pGL3 luciferase vectors carrying various fragments of RNF44 promoter regions were treated with or without BRAFi (5 μμ), ERKi (3 μμ), or AKTi (2 μμ). The luciferase activity was shown in bar graphs (n = 3, *P < 0.05, **P < 0.01, and ***P < 0.005). (D) Immunoblotting displayed the levels of p‐CREB (Ser133) and CREB in A2058BR cells treated with BRAFi, ERKi, and AKTi. (E) Four different mutant constructs established from pGL3‐1122 vector were created by site‐directed mutagenesis and then were, respectively, delivered into A2058BR and SK‐MEL28BR cells. The bar graph represented luciferase activity normalized by wild‐type (Wt). (F) Chromatin immunoprecipitation (ChIP) assay. The cell lysates extracted from A2058/A2058BR and SK‐MEL28/SK‐MEL28BR cells were incubated with antibody against CREB and with nonspecific IgG, and their binding sites in RNF44 promoter region were analyzed by real‐time PCR. The data have been normalized by nonspecific IgG and their parental counterparts. (G,H) CREB was silenced using siRNAs, RNF44 RNA levels were determined by qRT‐PCR, and protein levels were analyzed by immunoblotting (n = 3, *P < 0.05, **P < 0.01, and ***P < 0.005).
|
other
| 33.12 |
To examine whether CREB binding sites mainly govern RNF44 transcription activity in BR cells, site‐directed mutagenesis of these binding sites was carried out and their pGL3 constructs are shown in Fig. 5E. The LUC activity disclosed that mutation (Mut) occurring either first or second CREB binding site resulted in lower transcription activity than wild‐type (Wt). Mutating two CREB binding sites further reduced more LUC activity compared to single mutant CREB binding site. To verify whether these vital binding sites are highly recognized by CREB in BR cells, antibody against CREB, or nonspecific IgG (as a negative control), was applied to ChIP followed by real‐time PCR assay. The result showed that more CREB binding sites can be co‐immunoprecipitated with CREB in BR cells (Fig. 5F). Moreover, knockdown of CREB resulted in lower RNA and protein levels of RNF44 in BR cells (Fig. 5G,H). Altogether, the results suggested that RNF44 expression is primarily controlled by CREB binding at its binding sites.
|
other
| 31.94 |
Our previous data showed that treatment with ADI‐PEG20 is able to abort growth of BR xenograft tumors but only slow the growth of parental xenograft tumors (Li et al., 2016) (Fig. 6A). Furthermore, increased apoptosis in A2058BR and A375BR cells treated with ADI‐PEG20 was verified by in vivo Annexin V analysis (Fig. 6B). Consistent with in vitro data, H‐scores representing AMPK‐α1 levels were 1.5‐ to 2.5‐fold lower in BR xenograft tumors, while RNF44 levels were 1.4‐ to 4‐fold higher compared to parental xenograft tumors (Fig. 6C,D).
|
other
| 28.25 |
Upregulated RNF44 and downregulated AMPK‐α1 sensitize BR xenograft tumors to ADI‐PEG20 treatment. (A,B) Corresponding to our previous study (Li et al., 2016), ADI‐PEG20 significantly reduced BR tumor sizes by inducing cell apoptosis. Female nude mice were inoculated subcutaneously with 1 × 106 melanoma cells. When tumor volume reached 100 mm3, tumor‐bearing mice received treatment of ADI‐PEG20 (100 IU·kg−1) twice per week. On day 32, single‐cell suspensions were prepared from tumor tissues and then were labeled with Annexin V and then analyzed by FACS. (C) (D) AMPK‐α1 and RNF44 expressions were detected by IHC staining, and the data were presented as H‐scores (scale bar = 100 μm) (n = 5, **P < 0.01, and ***P < 0.005).
|
other
| 35.22 |
Attenuated AMPK‐α1 and increased RNF44 levels were also found in melanoma tumors from patients who failed BRAFi or BRAFi/MEKi treatment. Primary cultures including MEL‐DA (pretreatment) and MEL‐DeA (postrelapse) were incubated with vemurafenib and/or ADI‐PEG20. The levels of p‐AMPK/AMPK‐α1 and ASS1 were upregulated in MEL‐DA, but downregulated in MEL‐DeA following vemurafenib treatment. Treatment with ADI‐PEG20 enhanced phosphorylation of AMPK, but levels of RNF44 were not detectable in MEL‐DA (Fig. 7B). In contrast, RNF44 expression was detected after vemurafenib treatment in MEL‐DeA, corresponding to decreased levels of p‐AMPK/AMPK‐α1. We suggested that BR primary cultures may lose this correlation over time in the absence of vemurafenib, but it could be evoked by the addition of vemurafenib. The explants from BR patient #2 and patient #3 also displayed similar results as MEL‐DeA (Fig. 7A,B). Immunohistochemistry (IHC) results also confirmed that BRAFi resistance and BRAFi/MEKi resistance attenuated the levels of AMPK‐α1 in patient #4 (BR) and patient #5 (BMR) (Fig. 7C). Prior to BRAFi treatment, low expression of ASS1 can be seen in tumor tissue from patient #4, but not from patient #5. However, ASS1 expression completely disappeared in relapsed tumor #4. Furthermore, expression of ASS1, AMPK‐α1, or RNF44 represented as H‐score confirmed that BR/BMR tumor samples had 2‐7‐fold lower ASS1 and AMPK‐α1 than the average levels but fivefold higher RNF44 compared to naïve melanoma samples (Fig. 7D). RNF44 expression can be seen in both nuclei and cytoplasm (Fig. S7) and also inversely correlated with AMPK‐α1 (Fig. 7E). These correlations also appeared in BMR cell lines (A2058BMR and SK‐MEL28BMR) established in vitro (Fig. 4A). These BMR cells were also sensitive to ADI‐PEG20 treatment (Table S1).
|
other
| 28.78 |
Attenuated ASS1 and AMPK‐α1 but increased RNF44 levels were seen in tumors from BR/BMR patients. (A) The explant assay was shown as schematic workflow, which has been described in Section 2.10. (B) Tumor explants from melanoma patients (patients #2 and #3) who relapsed following treatment with vemurafenib were placed in the inserts and exposed to drugs (in lower compartments) for 48–72 h. The cell lysates were subjected to immunoblot analysis. MEL‐DA and MEL‐DeA cells were primary cultures isolated from a melanoma patient #1 and represent pretreatment and BRAFi resistance (postrelapse), respectively. (C) ASS1, AMPK‐α1, and RNF44 expressions in paraffin‐embedded tumor tissues of patient #4 (resistant to BRAFi) and patient #5 (resistant to BRAFi/MEKi) were determined by IHC staining (scale bar = 100 μm). The intensities were scored and shown in upper right corner of each image. (D) The dot plots represented H‐scores of ASS1, AMPK‐α1, and RNF44 in tumor samples from melanoma patients with pretreatment (n = 10) and BRAFi or BRAFi/MEKi resistance (n = 10). (E) The correlation between RNF44 and AMPK‐α1 levels was assessed by a linear regression.
|
other
| 33.56 |
The combination of BRAFi and MEKi has been approved by the FDA as first‐line treatment for BRAF‐mutant melanoma patients. Although this combination can prolong progression‐free survival, patients eventually relapse due to MEK2 mutation‐ and BRAF amplification‐induced ERK reactivation (Sosman et al., 2012; Villanueva et al., 2013). Other combination treatments, such as BRAFi in combination with HCQ, HSP90 inhibitor, CDK4/6 inhibitor, and mTOR inhibitor, have been shown to improve antitumor effects in vitro and in vivo (Acquaviva et al., 2014; Fedorenko et al., 2015; Paraiso et al., 2012), yet their clinical efficacy remains unproven.
|
study
| 29.22 |
We have demonstrated that most BR cells are extremely vulnerable to arginine deprivation (Li et al., 2016). Our results further uncovered that BR cells have attenuated levels of c‐Myc and AMPK‐α1, which are crucial to governing ASS1 re‐expression, metabolic reprogramming, and autophagy. Additionally, c‐Myc and AMPK‐α1 have been shown to promote glucose uptake and glycolysis via upregulation of GLUT1 and HK II (Hardie, 2011; Miller et al., 2012). Indeed, our results substantiated that BR cells have lower glucose uptake as well as decreased HK II and GLUT1 expressions (Fig. 1A,B). To compensate for this, BR cells lacking ASS1 consume amino acids such as exogenous arginine by increasing arginine transporter CAT‐2 to generate ATP via oxidative phosphorylation (OXPHOS) (Baenke et al., 2016; Hernandez‐Davies et al., 2015) (Fig. 1C, Figs S1 and S2C,D), and hence, the cells are vulnerable to arginine‐depleting agents. In addition to arginine, other amino acids, such as glutamine, have been appreciated to be another energy source to sustain OXPHOS in mitochondria of melanoma cells when they develop resistance to BRAFi. Hence, glutamine transporter ASCT2 (anti‐neutral amino acid transporter) has been proposed as a potential therapeutic target in melanoma cells (Wang et al., 2014). Although in vitro study has shown that ASCT2 antagonist BenSer can suppress melanoma cell proliferation via cell cycle arrest and mTOR inhibition, this compound is not yet available in the clinic and its toxicity is unknown.
|
study
| 32.22 |
With respect to autophagic mechanism, our previous study elucidated that besides AMPK, a key autophagic protein involved in membrane elongation of autophagosome formation, Atg5, was also downregulated in BR cells (Li et al., 2016). Downregulation of Atg5 seen in BR cells may not sufficiently accomplish autophagy flux upon arginine depletion even in the presence of AMPK. However, our result demonstrated that restoring or overexpressing AMPK‐α1 in BR cells can regain the ability to undergo autophagy as evidenced by increased autophagosome formation under arginine starvation (Li et al., 2016) (Fig. 3C and Fig. S4D). Moreover, a previous study discovered that mammalian cells can utilize AMPK to trigger Atg5‐independent macroautophagy for mitochondrial clearance (Ma et al., 2015). In MEF, ULK1/2 which can be activated by AMPK triggers Atg5‐independent pathway through Rab9‐mediated lipidation instead of LC3‐II‐mediated lipidation (Nishida et al., 2009). It is possible that AMPK overexpression in BR cells initiates ULK1/2 and subsequently activates Atg5‐independent autophagy. Our previous study also disclosed that BR cells express equal levels of p‐ULK as parental cells; however, unlike parental cells, arginine depletion fails to enhance p‐ULK expression in BR cells due to the lack of AMPK‐α1 (Li et al., 2016). Taken together, our data suggest that low Atg5 expression in BR cells can contribute to its inability to undergo autophagy; nevertheless, the primary contributor to defective autophagy in BR cells is AMPK‐α1, a known master regulator of autophagy.
|
other
| 27.78 |
It has been reported that BRAF‐mediated ERK/RSK activation negatively regulates LKB through phosphorylation at Ser428 and subsequently inhibits activation and phosphorylation of AMPK at Thr172 (Zheng et al., 2009). However, LKB‐AMPK activation in melanoma cells cannot be restored by adding BRAFi (Ma et al., 2014). Therefore, AMPK‐α1 stability may govern AMPK activation. Indeed, a recent study has shown that ubiquitination on AMPK‐α1 can suppress LKB‐phosphorylated AMPK activation, and deubiquitinase USP10 can abort this process (Deng et al., 2016). Our previous data showed that elevated p‐LKB expression appears in BR cells, yet p‐AMPK and AMPK‐α1 attenuations are still present in BR cells following ADI‐PEG20 treatment (Li et al., 2016). Thus, our results illustrated that UPS rather than BRAF‐ERK regulates AMPK activity in BR cells.
|
other
| 29.95 |
Our results demonstrated that BR cells are much less capable of undergoing autophagy due to downregulation of AMPK‐α1 through active UPS (Fig. 2A,B). With regard to ubiquitination of AMPK in other tissue, MuRF1 has been experimentally demonstrated to add atypical ubiquitin chains on AMPK (Zungu et al., 2011). In brown adipose and heart tissues, Cidea regulates AMPK activity via interaction with AMPK‐β (Zungu et al., 2011). A recent study reported that a cancer‐specific ubiquitin ligase, MAGEA3/6‐TRIM28, participates in AMPK‐α1 degradation leading to inhibition of autophagy in HeLa, HEK‐293, and U2OS cells (Pineda et al., 2015). Nevertheless, none of these ubiquitin ligases can be detected to interact with AMPK‐α1 in BR melanoma cells (Fig. 2D). It is likely that ubiquitin ligases governing degradation of AMPK‐α1 are both tissue and tumor type specific. Our proteomic analyses have identified a novel protein, RNF44, which is a RING finger protein of C3H4 family. The actual structure needs to be confirmed by X‐ray crystallography. Notably, RNF44 can be detected in BR or BMR melanoma tissues rather than naive samples, and its expression also inversely correlates with AMPK‐α1 in tissue samples. Interestingly, BR cells, which have less ability to undergo autophagy due to attenuation of AMPK‐α1 secondary to higher levels of RNF44, also have high levels of p‐ERK/AKT. Blockade of AKT/ERK activation further confirmed that elevated RNF44 levels in BR cells are related to hyperactivation of AKT/ERK. The downstream transcription factor CREB, which has been reported to be activated/phosphorylated by AKT/ERK at Ser133, can interact with RNF44 promoter region and subsequently trigger RNF44 transcription (Fig. 5A–H).
|
study
| 32.47 |
Resistance to BRAFi or BRAFi/MEKi enhances RNF44‐mediated AMPK‐α1 degradation. Downregulation of AMPK‐α1 disables autophagy and results in less glucose uptake but increases arginine transporter CAT‐2 expression (graphical abstract, Fig. 8). In addition, c‐Myc attenuation is known to attenuate glucose addiction as evidenced by a decrease in GLUT1 and HK II expressions (Long et al., 2013; Miller et al., 2012), and abrogate arginine synthesis due to downregulated ASS1 expression. In this circumstance, exogenous arginine becomes the energy source for BR cells. Thus, depleting arginine coupled with an inability to trigger autophagy and ASS1 expression drives BR cells to undergo apoptosis. Therefore, ADI‐PEG20 treatment could be a good candidate for salvage therapy in BR/BMR melanomas.
|
study
| 28.9 |
The schematic diagram illustrates several mechanisms leading to sensitivity to arginine deprivation/ADI‐PEG20 treatment in BR cells. BRAFi resistance downregulates c‐Myc‐mediated ASS1 transcription; hence, BR cells are arginine auxotrophic. Moreover, upregulated E3 ligase RNF44 also promotes AMPK‐α1 degradation, which abrogates autophagic flux, impairs glucose uptake and glycolysis, and enhances expression of arginine transporter 2 (CAT‐2). These biological alterations switch metabolism toward exogenous arginine addiction and consequently give rise to an increase in vulnerability to arginine deprivation in BR cells.
|
other
| 26.42 |
YL and CW performed proliferation, apoptotic analyses, IP, immunoblotting, and creating BR cell lines. YL, SC, SS, and JP completed autophagosome and IHC staining. YL, SC, CW, and MW contributed to the results of animal study. XH and MK contributed to PGL3 plasmid and LUC activity assay. MS dictated H&E staining. YL, LF, NS, and SS participated in explant assay and composed and proofed the manuscript.
|
study
| 32.75 |
Neural oscillations reflect the functional state of the brain and are thought to dynamically orchestrate brain activity as a function of specific task requirements1–3. Oscillations in the developing brain have been studied both in animals using invasive recordings4–12, and in humans with electroencephalography (EEG)13–15. These rhythms have been implicated in the reorganization and refinement of the neural circuits throughout development16,17. In the developing brain, oscillations at different frequencies occur spontaneously4,5,7,18 and are also engaged by external stimuli or task demands4,19. However, relatively little is known about the interaction between LFP oscillations and spiking activity or between different frequency bands throughout brain development8,20,21. Further, it is not well known how endogenous oscillations are modulated by exogenous stimuli during early development.
|
other
| 31.28 |
Previous studies of brain development have revealed a critical period when external experience is crucial for the normal development of cortical functional organization22–24. However, the development of functional organization may be independent of sensory experience, as suggested by recent studies showing that self-organized, spontaneous activity patterns precede and prepare for the emergence of maps25–27. Furthermore, a transient state of prolonged sensory responses was found in visual cortex but disappeared after eye-opening4,21. To untangle the influence of sensory-independent and experience-elicited neural activity on the development of cortical networks, it is necessary to compare patterns of network activity across different periods throughout development.
|
other
| 35.75 |
To investigate the link between patterns of neural activity and functional development, we recorded local field potential (LFP) and multiunit-activity (MUA) in the primary visual cortex (area 17) of juvenile ferrets throughout different stages of development. Ferrets, a species relatively immature at birth, are a well-established model system for studying neural development, with most previous studies focusing on functional organization24,27–29, or spiking activity25,30–32. Here, we focus on the time periods before and after eye-opening as a critical time point that separates periods dominated by endogenously generated and externally driven activity patterns, respectively. We focus on alpha/beta/low-gamma/high-gamma (8–15 Hz, 15–30 Hz and 30–55 Hz and 65–100 Hz, respectively) frequency bands due to their known role in coordinating spiking activity during cognition in the adult animal33,34. Our results reveal profound changes in high frequency oscillations after eye-opening, including (1) increase in high frequency (especially >30 Hz) oscillation amplitude, (2) decreased spike coupling to low frequency phase (theta/alpha) but increase to high frequency phase (>50 Hz), (3) decreased theta (4–8 Hz)/alpha/beta band amplitude coupling to delta (1–4 Hz) band phase but increased high-gamma band amplitude coupling to theta/alpha band phase, and (4) increased entrainment of the LFP by periodic exogenous visual stimuli. Our results support the hypothesis that the development of higher frequency oscillations (especially >30 Hz) serves as a biomarker for the maturation of cortical circuits.
|
other
| 30.22 |
We asked how both local- and network-level patterns of neural activity in ferret visual cortex change around eye-opening, a critical time point in development. To answer this question, we recorded spontaneous and visually-elicited MUA and LFP activity in visual cortex of freely-moving juvenile ferrets (Fig. 1A,B, see methods for details) before eye-opening (postnatal age P22-P29, mean ± SD: 26.0 ± 2.6, n = 9) or after (P33-P50, 41.3 ± 4.5, n = 11). All animals in the before eye-opening group had both eyes completely closed and all animals in the after eye-opening group had both eyes completely opened at the time of recording. We found that in contrast to the more continuous activity patterns presented after eye-opening (Fig. 1F, example from a P50 animal, also refer to Fig. 2 for more examples with higher temporal resolution), spontaneous activity before eye-opening was characterized by short periods of pronounced LFP amplitude and spiking rate (bursts), interleaved with periods of relative silence (Fig. 1E, example from a P29 animal). The alternation of bursts and periods of low-amplitude activity in ferrets before eye-opening is consistent with previous observations in the pre-term infant13 and rodents and ferrets before eye-opening35–37.Figure 1Experimental setup, recording locations and example spontaneous activity. (A) Experimental setup. Spontaneous or visually-elicited extracellular activity was recorded via implanted electrode arrays from awake, unrestrained juvenile ferrets. Visual stimulation was delivered using four LED arrays positioned in the four corners of the test cage. (B) Illustration of the 2 × 8 electrode array implant location in one example P25 animal brain. Each electrode is represented by a red dot. (C) Nissl stained section of V1 in a P40 animal showing the location of implanted electrodes. Scale bar represents 1 mm. (D) Close-up of the region enclosed by the black box in figure C. Black arrows indicate the electrode tracks. Scale bar represents 400 um. (E) Example spontaneous activity from an animal before eye-opening. From top to bottom: LFP trace (1–300 Hz), MUA (>300 Hz), and spectrogram of the same 10 second epoch. The brown highlighted epochs in the MUA plot represent the detected bursting periods. The spectrogram around 60 Hz is left blank due to the applied notch filter. (F) Example spontaneous activity from an animal after eye-opening. Same conventions and plot order as E. Figure 2Additional example traces of spontaneous LFP and MUA before and after eye opening. (A) Example traces of three simultaneously recorded channels in another P29 animal. For clarity, raw signals were band-passed filtered and shown as low-frequency LFP (1–30 Hz, top in each subplot), high-frequency LFP (30–300 Hz, middle) and MUA (300–5000 Hz, bottom). Each short dashed line above the MUA traces indicates the detection of a spiking event. The insets in the bottom left of the MUA traces shows the shape of the detected spikes. (B) Example traces of three simultaneously recorded channels in another P43 animal. Same configuration as A.
|
review
| 26.44 |
Experimental setup, recording locations and example spontaneous activity. (A) Experimental setup. Spontaneous or visually-elicited extracellular activity was recorded via implanted electrode arrays from awake, unrestrained juvenile ferrets. Visual stimulation was delivered using four LED arrays positioned in the four corners of the test cage. (B) Illustration of the 2 × 8 electrode array implant location in one example P25 animal brain. Each electrode is represented by a red dot. (C) Nissl stained section of V1 in a P40 animal showing the location of implanted electrodes. Scale bar represents 1 mm. (D) Close-up of the region enclosed by the black box in figure C. Black arrows indicate the electrode tracks. Scale bar represents 400 um. (E) Example spontaneous activity from an animal before eye-opening. From top to bottom: LFP trace (1–300 Hz), MUA (>300 Hz), and spectrogram of the same 10 second epoch. The brown highlighted epochs in the MUA plot represent the detected bursting periods. The spectrogram around 60 Hz is left blank due to the applied notch filter. (F) Example spontaneous activity from an animal after eye-opening. Same conventions and plot order as E.
|
other
| 29.39 |
Additional example traces of spontaneous LFP and MUA before and after eye opening. (A) Example traces of three simultaneously recorded channels in another P29 animal. For clarity, raw signals were band-passed filtered and shown as low-frequency LFP (1–30 Hz, top in each subplot), high-frequency LFP (30–300 Hz, middle) and MUA (300–5000 Hz, bottom). Each short dashed line above the MUA traces indicates the detection of a spiking event. The insets in the bottom left of the MUA traces shows the shape of the detected spikes. (B) Example traces of three simultaneously recorded channels in another P43 animal. Same configuration as A.
|
other
| 30.5 |
To further quantify the difference in overall activity structure in these two periods, we compared the LFP power spectra and found that the power in all frequency bands was significantly increased after eye-opening (Fig. 3A, Mann-Whitney U-test, p < 0.001 for full frequency range, before eye-opening, n = 42 channels, after eye-opening, n = 102 channels). However, an increase in LFP power might merely be a consequence of the disappearance of the low-amplitude periods between bursts, which contribute little to the overall power. To control for this potential confound, we next compared LFP power during burst periods before eye-opening to LFP power after eye-opening. We defined the time windows of bursts as periods with a maximal inter-spike-interval of 100ms and minimal spike count of 10 (see Chiappalone et al., 200638, also see Methods for details). This method efficiently identified the bursts before eye-opening (brown highlighted epochs in MUA trace in Fig. 1E). The same criterion was applied to data after eye-opening for comparison. Before eye-opening, the burst occurrence was 1.92 ± 2.40 bursts per minute, the average duration was 350.3 ± 155.8 ms, and the mean spike rate in burst periods was 64.32 ± 31.20 spikes/second. Applying the same criteria in activity after eye-opening yields higher occurrence (8.2 ± 9.5 bursts per minute, p < 0.001) and duration (447.5 ± 121.5 ms, p < 0.001), but lower intra-burst spike rate (42.64 ± 16.79 spikes/second, p < 0.001). Focusing on the LFP specifically within burst periods, we found that the power spectrum before eye-opening already exhibited multiple peaks (Fig. 3B, black line). However, in burst periods, power was lower before eye-opening compared to after eye-opening, and the differences were significant in the high-gamma band (50–100 Hz in Fig. 3B), indicating that the elevation of the overall LFP power after eye-opening resulted from both a decrease in duration of quiescent periods, and an increase of high frequency LFP amplitude in active periods. It is notable that besides the increase of absolute LFP power in all frequency bands, the relative power was enhanced in higher frequency bands (the percentage to total power: alpha, 8–15 Hz, Before = 4.59% ± 1.87%, After = 7.76% ± 3.13%, p < 0.05; beta, 15–30 Hz, Before = 2.95% ± 1.47%, After = 10.62% ± 7.55%, p < 0.05; low-gamma, 30–55 Hz, Before = 1.56% ± 0.96%, After = 6.43% ± 3.10%, p < 0.001; high-gamma, 65–100 Hz, Before = 0.95% ± 0.62%, After = 2.15% ± 1.12%, p < 0.05).Figure 3LFP power of spontaneous activity (animal in the dark) before and after eye-opening. (A) Power spectra (1/f normalized) across the whole recording session in animals before (black, n = 42 channels) and after (red, n = 102 channels) eye-opening. Traces and shadows represent mean and s.e.m, respectively. The data around 60 Hz (dotted lines) are removed and interpolated between adjacent data points due to the applied notch filter. The dashed line marks the frequency range in which the power is significantly different between the two periods. (B) Power spectra (1/f normalized) in burst periods in animals before eye-opening (black) and in firing-rate matched epochs after eye-opening (red). The dashed line marks the frequency range in which the power is significantly different between the two periods. (C) LFP gamma band power of spontaneous activity as a function of postnatal day. Green dots and blue dots indicate low gamma band (30–55 Hz) and high gamma band (65–100 Hz), respectively. Each dot represents data from one animal. Error-bars indicate s.e.m. The period of eye-opening is marked by the brown shaded box. Before eye-opening the gamma power was relatively stable and after eye-opening the gamma power increased. *p < 0.05, **p < 0.01.
|
other
| 29 |
LFP power of spontaneous activity (animal in the dark) before and after eye-opening. (A) Power spectra (1/f normalized) across the whole recording session in animals before (black, n = 42 channels) and after (red, n = 102 channels) eye-opening. Traces and shadows represent mean and s.e.m, respectively. The data around 60 Hz (dotted lines) are removed and interpolated between adjacent data points due to the applied notch filter. The dashed line marks the frequency range in which the power is significantly different between the two periods. (B) Power spectra (1/f normalized) in burst periods in animals before eye-opening (black) and in firing-rate matched epochs after eye-opening (red). The dashed line marks the frequency range in which the power is significantly different between the two periods. (C) LFP gamma band power of spontaneous activity as a function of postnatal day. Green dots and blue dots indicate low gamma band (30–55 Hz) and high gamma band (65–100 Hz), respectively. Each dot represents data from one animal. Error-bars indicate s.e.m. The period of eye-opening is marked by the brown shaded box. Before eye-opening the gamma power was relatively stable and after eye-opening the gamma power increased. *p < 0.05, **p < 0.01.
|
other
| 27.39 |
To investigate whether eye-opening is an important event in the development of high frequency oscillations, we plotted the low- and high-gamma power as a function of postnatal day across all animals. The results (Fig. 3C) revealed that gamma power dynamics were distinct for the two stages of development, and were separated by eye-opening: gamma power was at a stable low level before eye-opening; and increased gradually after eye opening. A Chow-test revealed that the linear regression coefficients in the two age groups were significantly different (low-gamma, before eye-opening slope = 0.002, n = 9 animals, after eye-opening slope = 0.062, p < 0.05, n = 11 animals; high-gamma, before eye-opening slope = −0.022, after eye-opening slope = 0.093, p = 0.05). Due to the increased level of visual stimulation upon eye-opening, these results indicate that external sensory stimuli may play an important role in the development of high frequency oscillations.
|
other
| 34.28 |
In order to determine if the observed differences in the temporal structure of the LFP impact neuronal spiking, we next studied the link between the LFP and the spiking activity (MUA). It is well-known that structured neuronal activity is crucial for circuit maturation in the visual system (e.g. retinal waves). In agreement with previous reports37 we observed that MUA was synchronized across channels before eye-opening (Fig. 4A for example of synchronized channels). We hypothesized that before eye-opening intrinsically generated network oscillations guide spiking activity, whereas after eye-opening spiking activity is less coupled to rhythmic patterns of network activity. Indeed, the representative trace in Fig. 4C shows that before eye-opening, spikes were highly phase-locked to the alpha LFP phase. This phenomenon of elevated spike-LFP phase locking before eye-opening corresponds to a peak in the alpha band in the phase-locking value curve (black line in Fig. 4D). However, after eye-opening the spike-phase locking was significantly decreased in the theta and alpha bands, but was increased in high gamma band (red line in Fig. 4D, p < 0.01 in 5–15 Hz and 50–80 Hz, before eye-opening, n = 27, after eye-opening, n = 96). The change of spike-phase locking was also accompanied with decreased MUA synchronization measured between electrode contacts (Fig. 4B, correlation coefficient value Before = 0.177 ± 0.088, n = 7, After = 0.061 ± 0.039, n = 11, p < 0.01, see Fig. 4B inlet for distribution of correlation coefficients for all channel pairs). This observation matches a previous finding that spontaneous activity exhibited decorrelation soon after eye-opening (Fig. 4B in red). It is also notable that negative correlation values were only found after eye-opening (Fig. 4B inlet). Therefore, in agreement with our hypothesis, the overall link between the LFP and MUA was altered after eye-opening. The decreased coupling to theta/alpha phase indicates the spiking activity is less influenced by low-frequency rhythms, while the increased coupling to the high-gamma phase hints at the development of the inhibitory component of the local network.Figure 4Synchronization and spike phase-locking of spontaneous activities before and after eye-opening. (A) Example MUA traces from four simultaneously recorded channels in an animal before eye-opening. (B) Main plot: spike cross-correlograms in animals before eye-opening (black, n = 7) and after eye-opening (red, n = 11). Traces and shadows represent mean and s.e.m, respectively. Inlet: distribution of the spike cross-correlation coefficients for each channel pair. Data are pooled from all animals and sorted by the correlation coefficient value. (C) An example MUA trace in which the spikes were phase-locked to the alpha/ low beta (8–20 Hz) cycle of the simultaneously recorded LFP. The troughs of the alpha/ low beta phase are highlighted by the red shading. (D) Frequency-resolved spike LFP phase-locking value before eye-opening (black, n = 27 channels) and after eye-opening (red, n = 96). Traces and shadows represent mean and s.e.m, respectively. The dashed lines mark the frequency range in which the phase-locking value is significantly different between the two developmental periods. **p < 0.01.
|
other
| 27.9 |
Synchronization and spike phase-locking of spontaneous activities before and after eye-opening. (A) Example MUA traces from four simultaneously recorded channels in an animal before eye-opening. (B) Main plot: spike cross-correlograms in animals before eye-opening (black, n = 7) and after eye-opening (red, n = 11). Traces and shadows represent mean and s.e.m, respectively. Inlet: distribution of the spike cross-correlation coefficients for each channel pair. Data are pooled from all animals and sorted by the correlation coefficient value. (C) An example MUA trace in which the spikes were phase-locked to the alpha/ low beta (8–20 Hz) cycle of the simultaneously recorded LFP. The troughs of the alpha/ low beta phase are highlighted by the red shading. (D) Frequency-resolved spike LFP phase-locking value before eye-opening (black, n = 27 channels) and after eye-opening (red, n = 96). Traces and shadows represent mean and s.e.m, respectively. The dashed lines mark the frequency range in which the phase-locking value is significantly different between the two developmental periods. **p < 0.01.
|
other
| 29.98 |
Given this changing link between oscillations and neuronal spiking, we next asked if the interaction of oscillations at different frequencies, which is known to be present in the adult animal, similarly changes with eye-opening. Considering the shift of the frequency in functional coupling between spike and LFP, we hypothesize that with eye-opening the coupling of alpha/beta/low-gamma amplitude to the phase of lower frequency oscillations also decreased. We tested this hypothesis by computing the phase-amplitude coupling of the LFP. An example trace recorded before eye-opening (P29, Fig. 5A) indicated that low-gamma band amplitude was modulated by delta phase. The averaged phase-amplitude coupling map shows that the amplitude of a broad band of LFP frequencies was modulated by the delta band (2–4 Hz) phase before eye-opening (Fig. 5B). After eye-opening, however, the amplitude coupling to the delta phase was decreased except for the high-gamma band (Fig. 5C), which was increased. High-gamma amplitude was also more coupled to theta band phase. Before eye-opening the cross-frequency phase amplitude coupling (PAC) value was significantly larger in theta/alpha/beta/low-gamma amplitude to delta-phase coupling, as well as in beta amplitude to theta/alpha phase coupling (Fig. 5D shows the p value indicating PACs were significantly decreased after eye-opening). In contrast, after eye-opening the PAC value was significantly larger in high-gamma amplitude to theta/alpha phase coupling (Fig. 5E shows the p value indicating PACs were significantly increased before eye-opening). Thus, these findings agree with the above result of a decreased overall functional coupling of rhythmic activity patterns upon eye-opening, and support a conceptual model in which the maturation of network dynamics in visual cortex is associated with an overall decrease in the integration of cortical rhythms at different frequencies (except in the high-gamma band).Figure 5Spontaneous LFP phase-amplitude coupling before and after eye-opening. (A) An example trace of gamma (30–55 Hz) amplitude which was elevated at troughs of the LFP delta (2–4 Hz) cycle. Troughs of the delta oscillation are represented by gray shading. (B) Averaged phase-amplitude coupling before eye-opening (n = 9 animals). (C) Averaged phase-amplitude coupling after eye-opening (n = 11). (D) p value of Mann-Whitney U-test indicating whether phase-amplitude coupling is significantly stronger for before eye-opening compared to after eye-opening. The p values are without correction for multiple comparisons. (E) p value of Mann-Whitney U-test indicating whether phase-amplitude coupling is significantly weaker for before eye-opening compared to after eye-opening. The p values are without correction for multiple comparisons.
|
other
| 28.03 |
Spontaneous LFP phase-amplitude coupling before and after eye-opening. (A) An example trace of gamma (30–55 Hz) amplitude which was elevated at troughs of the LFP delta (2–4 Hz) cycle. Troughs of the delta oscillation are represented by gray shading. (B) Averaged phase-amplitude coupling before eye-opening (n = 9 animals). (C) Averaged phase-amplitude coupling after eye-opening (n = 11). (D) p value of Mann-Whitney U-test indicating whether phase-amplitude coupling is significantly stronger for before eye-opening compared to after eye-opening. The p values are without correction for multiple comparisons. (E) p value of Mann-Whitney U-test indicating whether phase-amplitude coupling is significantly weaker for before eye-opening compared to after eye-opening. The p values are without correction for multiple comparisons.
|
other
| 29.52 |
We then investigated how fast the brain oscillations developed by looking at the spike-phase locking and phase-amplitude coupling in the first week after eye-opening, in which period the functional properties of the visual cortex develops quickly under the influence of visual experience39,40. We found that both the spike-phase locking and the phase-amplitude coupling rapidly transformed within the first week after eye-opening (Supplemental Fig. 1), suggesting a possible linkage between the development of neuronal synchronizations and the development of the functional properties27.
|
other
| 34.3 |
We next asked if changes in spontaneous activity patterns are associated with similarly pronounced changes to sensory responses in cortex. While a detailed probing of visual responses is difficult in freely-moving juvenile animals, it has been previously established that visual stimuli evoke visual responses even before eye opening4,21,30. Specifically, we asked whether the visual cortex responds differently before and after eye-opening to bright 500 ms whole-field flash stimuli (See Methods). The flash stimuli elicited responses in visual cortex even before eye-opening (Fig. 6A for example LFP trace and spectrogram, 6 C for example spike raster and PSTH, 6E for population spectrogram and 6 G black trace for population PSTH), most likely via semi-transparent eye-lids30. Before eye-opening, response patterns were characterized by a second activity peak ranging from 500–1500 ms after stimulus offset, which were distinct from the response pattern confined to period of stimulus presentation after eye-opening (Fig. 6B,D,F, and red trace in Fig. 6G). Furthermore, the responses before eye-opening were also characterized by longer latency (Fig. 6H, Before: 215.2 ± 132.7 ms, n = 23; After: 104.8 ± 106.7 ms, n = 46, p < 0.001) and larger variance across trials (firing rate coefficient of variance, Fig. 6I, Before: 0.92 ± 0.38; After: 0.53 ± 0.18, p < 0.001). The prolonged responses after stimulus-offset, the higher response latency, as well as higher inter-trial firing rate variance indicates that the visual cortex was unable to respond to the sensory inputs in a temporally-constrained and stereotyped manner before eye-opening.Figure 6Visual responses before and after eye-opening. (A,B). Mean spectrograms of example recording session before (A) and after (B) eye-opening. LFP data (black traces) of one trial are superimposed on the top of spectrograms. The black lines above each spectrogram represent visual stimulus duration. The spectrogram around 60 Hz is left blank due to the applied notch filter. (C,D) Spike raster plots and peri-stimulus time histograms (PSTHs) from the same recording as A and B, respectively. Black lines above each plot represent visual stimulus duration. (E,F) Mean spectrograms across all recording sessions before (E, n = 7 animals) and after (F, n = 9) eye-opening. (G) Z-scored population PSTHs before and after eye-opening. Shaded regions represent 95% confidence intervals of the mean. (H,I) Response latency and coefficient of variance (CV) of firing rate (during the 500ms visual stimulation epochs) before (black) and after (red) eye-opening. Error-bar indicates s.e.m. ***p < 0.001.
|
review
| 27.47 |
Visual responses before and after eye-opening. (A,B). Mean spectrograms of example recording session before (A) and after (B) eye-opening. LFP data (black traces) of one trial are superimposed on the top of spectrograms. The black lines above each spectrogram represent visual stimulus duration. The spectrogram around 60 Hz is left blank due to the applied notch filter. (C,D) Spike raster plots and peri-stimulus time histograms (PSTHs) from the same recording as A and B, respectively. Black lines above each plot represent visual stimulus duration. (E,F) Mean spectrograms across all recording sessions before (E, n = 7 animals) and after (F, n = 9) eye-opening. (G) Z-scored population PSTHs before and after eye-opening. Shaded regions represent 95% confidence intervals of the mean. (H,I) Response latency and coefficient of variance (CV) of firing rate (during the 500ms visual stimulation epochs) before (black) and after (red) eye-opening. Error-bar indicates s.e.m. ***p < 0.001.
|
other
| 28.94 |
These changes in MUA responses prompted us to ask whether the visual-elicited LFP also changed after eye-opening. We hypothesized that the unit activity, which is less temporally-confined to visual stimuli is related to the immaturity of the inhibitory component of the neural circuit35, as reflected by lower amplitude high-frequency oscillations (especially gamma). Indeed, the spectrogram showed that visual stimuli elicited high-gamma (65–100 Hz) activation after eye-opening but not before (Fig. 6E versus 6 F). For a quantitative comparison, we calculated the difference of LFP power spectra between the baseline and visually-elicited responses (Fig. 7A). Before eye-opening, the LFP of visual responses had two local peaks at alpha and low-gamma (8–12 Hz and 30–40 Hz respectively as shown in the black line of Fig. 7A) but little activity above 40 Hz. After eye-opening, visual stimulation also elicited alpha activity, but the peak shifted to the high frequency band (~12–15 Hz as shown in the red line of Fig. 7A) compared to before eye-opening. In addition, higher-gamma activity was also substantially enhanced (40–100 Hz, p < 0.001, before eye-opening, n = 23, after eye-opening, n = 46). The overall visual responses were significantly higher after eye-opening in all frequency bands but, this increase was strongest in alpha/beta and high-gamma frequency bands. These results reveal that visual stimuli elicit larger amplitude high frequency LFP oscillations after eye-opening.Figure 7LFP power, spike LFP phase-locking, and phase-amplitude coupling of visually-elicited responses before and after eye-opening. (A) Power spectra of visually-elicited activity (subtracted from the baseline power) before (black, n = 7 animals) and after (red, n = 9) eye-opening. Traces and shaded regions represent mean and s.e.m, respectively. The data around 60 Hz (dotted lines) are removed and interpolated between adjacent data points due to the applied notch filter. The dashed lines mark the frequency ranges in which the power is significantly different between the two periods. (B) Frequency-resolved spike LFP phase-locking value before eye-opening (black, n = 13 channels) and after eye-opening (red, n = 43). Traces and shadows represent mean and s.e.m, respectively. The dashed lines mark frequency ranges in which the phase-locking value is significantly different between the two periods. (C–F) Phase-amplitude coupling of visually-induced oscillations before eye-opening (C, n = 7 animals) and after eye-opening (D, n = 9). P values indicate whether phase-amplitude coupling is significantly stronger (E) or weaker (F) before eye-opening compared to after eye-opening. *p < 0.05, **p < 0.01.
|
other
| 26.73 |
LFP power, spike LFP phase-locking, and phase-amplitude coupling of visually-elicited responses before and after eye-opening. (A) Power spectra of visually-elicited activity (subtracted from the baseline power) before (black, n = 7 animals) and after (red, n = 9) eye-opening. Traces and shaded regions represent mean and s.e.m, respectively. The data around 60 Hz (dotted lines) are removed and interpolated between adjacent data points due to the applied notch filter. The dashed lines mark the frequency ranges in which the power is significantly different between the two periods. (B) Frequency-resolved spike LFP phase-locking value before eye-opening (black, n = 13 channels) and after eye-opening (red, n = 43). Traces and shadows represent mean and s.e.m, respectively. The dashed lines mark frequency ranges in which the phase-locking value is significantly different between the two periods. (C–F) Phase-amplitude coupling of visually-induced oscillations before eye-opening (C, n = 7 animals) and after eye-opening (D, n = 9). P values indicate whether phase-amplitude coupling is significantly stronger (E) or weaker (F) before eye-opening compared to after eye-opening. *p < 0.05, **p < 0.01.
|
other
| 28.47 |
We then asked whether spike phase locking and phase amplitude coupling during visual responses develops in a similar way as observed for spontaneous activity. The results showed that, similar to patterns of the spontaneous activity, visual induced spiking activity was highly coupled to the low frequency LFP phase with two peaks in theta/alpha and beta bands, respectively. The phase locking value was significantly decreased in theta/alpha/beta bands but was increased in high-gamma band after eye-opening (Fig. 7B, p < 0.05, before eye-opening, n = 13, after eye-opening, n = 43). Accordingly, the phase amplitude coupling also shows similar changes from low frequency to high frequency: before eye-opening alpha amplitude was coupled to delta/theta phase, and low-gamma amplitude was coupled to theta/alpha phase (Fig. 7C,E shows PAC values which were significantly larger before eye-opening); while after eye-opening high-gamma amplitude was coupled to delta/alpha phase (Fig. 7D,F shows PAC values which were significantly larger after eye-opening). In summary, the spiking activity and LFP amplitude became less coupled to lower frequency phase and more coupled to higher frequency phase, suggesting the visual responses developed to be less synchronized by the slow intrinsic network rhythms but instead to be more synchronized by the higher-frequency network activity patterns that may be specific to the visual information processing.
|
other
| 29.69 |
In the light of the increased high frequency oscillations in both spontaneous and visual-elicited LFP after eye opening, we further dissected the role of these oscillations in visual information processing. We hypothesized that visual responses before eye-opening were unable to reliably follow the temporal structure of visual stimuli, as the visual response was not time-constrained to the stimulus duration (Fig. 6A,C,E) and had less gamma power. To test this, we used a set of visual stimuli consisting of square-wave light flashes with frequencies ranging from 0.25–29 Hz presented in either ascending or descending order to evoke visual responses and investigate the capability of the network to be entrained by stimulus frequency. Example traces show that before eye-opening, oscillations were able to follow the pace of the visual stimulus only for low frequencies (Fig. 8A, left plots, note there was also a power peak at harmonic frequency), but not when the flashing was fast (Fig. 8A, right plots). This was not the case after eye-opening (Fig. 8B), when network activity reliably followed the temporal structure of all stimulation frequencies. In fact, the population-normalized LFP power at the stimulation frequency was near 1 before eye-opening when the stimulus frequency was above 10 Hz (black line in Fig. 8C), indicating the absence of LFP entrainment to visual stimuli above 10 Hz in this developing period. After eye-opening, the normalized LFP power at the stimulation frequency was increased (p < 0.001 for all frequency, before eye-opening, n = 15, after eye-opening, n = 9) for all frequencies tested. We further investigated whether this enhanced response power really reflected entrainment or just an overall increase of the evoked power, by looking at the inter-stimulus phase coherence of the LFP for flashes presented at each stimulus frequency. The inter-stimulus phase coherence was decreased as the stimulus frequency increased before eye-opening, with minimal inter-stimulus phase coherence observed for stimulus frequencies above 10 Hz (Fig. 8D). The inter-stimulus phase coherence for visual stimuli above 10 Hz was significantly increased after eye-opening (p < 0.001 for 9–29 Hz), indicating the elevated normalized power observed in Fig. 8C was at least partially the result of enhanced entrainment. The changes of visual responses to different stimulus frequencies were also reflected in spiking activity: the firing rate dropped when stimulus frequency increased beyond 3 Hz before eye-opening, but the frequency tuning was relatively broad after eye-opening (Fig. 8E, p < 0.05 for 5–9 Hz). Interestingly, we found a rebound of the firing rate in response to stimuli above 10 Hz before eye-opening. These results suggest that in development, the visual system increases its ability to capture the temporal structure of visual inputs, which coincides with the development of high frequency oscillations and desynchronization of the activity at lower frequencies.Figure 8Visual entrainment to different stimulation frequencies before and after eye-opening. (A) Example neural responses to low frequency (1 Hz, left) and high frequency (25 Hz, right) visual stimuli from an animal before eye-opening. Top traces: visual stimuli pattern. Middle traces: broadband raw data from one channel. Bottom: Normalized power spectrum. Please note the time scales are different between the left and the right. (B) Example neural responses from an animal after eye-opening. Same configuration as A. (C) Normalized LFP power at the frequency of visual stimulation before eye-opening (black, n = 15) and after eye-opening (red, n = 9). LFP power was normalized by the session average. Traces and shadows represent mean and s.e.m, respectively. The dashed line marks the frequency range in which the LFP power is significantly different between the two periods. (D) Inter-stimulus phase coherence before eye-opening (black) and after eye-opening (red). Traces and shadows represent mean and s.e.m., respectively. The dashed line marks the frequency range in which the phase-locking value is significantly different between the two periods. (E) Normalized firing rate as a function of the stimulus frequency. Traces and shading represent mean and s.e.m, respectively. The dashed line marks the frequency range in which the firing rate is significantly different between the two periods. *p < 0.05, ***p < 0.001.
|
other
| 28.98 |
Visual entrainment to different stimulation frequencies before and after eye-opening. (A) Example neural responses to low frequency (1 Hz, left) and high frequency (25 Hz, right) visual stimuli from an animal before eye-opening. Top traces: visual stimuli pattern. Middle traces: broadband raw data from one channel. Bottom: Normalized power spectrum. Please note the time scales are different between the left and the right. (B) Example neural responses from an animal after eye-opening. Same configuration as A. (C) Normalized LFP power at the frequency of visual stimulation before eye-opening (black, n = 15) and after eye-opening (red, n = 9). LFP power was normalized by the session average. Traces and shadows represent mean and s.e.m, respectively. The dashed line marks the frequency range in which the LFP power is significantly different between the two periods. (D) Inter-stimulus phase coherence before eye-opening (black) and after eye-opening (red). Traces and shadows represent mean and s.e.m., respectively. The dashed line marks the frequency range in which the phase-locking value is significantly different between the two periods. (E) Normalized firing rate as a function of the stimulus frequency. Traces and shading represent mean and s.e.m, respectively. The dashed line marks the frequency range in which the firing rate is significantly different between the two periods. *p < 0.05, ***p < 0.001.
|
other
| 29.95 |
The synchronization of populations of neurons reflected by the LFP has emerged as a fundamental mediator of brain function and behavior. In recent years, the development of the circuit-level mechanisms that enable neuronal synchrony have received growing attention. The present study recorded both MUA and LFP in the visual cortex in freely moving, juvenile ferrets. When comparing our recordings from after eye-opening to before eye-opening we found that (1) both spontaneous activity and visual responses were increased in the high-frequency band (gamma oscillations), (2) spike-locking to the theta/alpha band phase was decreased and to high-gamma band phase was increased, (3) coupling of theta/alpha/beta amplitude to delta phase was decreased while coupling of high-gamma amplitude to theta/alpha phase was increased, and finally (4) the LFP was more reliably entrained by higher frequency visual stimulation.
|
other
| 26.05 |
We found that the amplitude of high-frequency rhythms (gamma band) was increased after eye-opening. Previous results in humans41 and more recent reports in mice12,20 showed that the resting-state gamma oscillations continuously developed for a relatively long period during childhood, which has been suggested to correlate to both typical42 and atypical development43 of higher-order brain function. Our results reveal that gamma oscillations (both spontaneous and visually-elicited) increase from as early as immediately after eye-opening in freely-moving ferrets studied under more natural conditions without anesthesia or head-fixation. Furthermore, visually-elicited high gamma (65–100 Hz) oscillations also increased. The mechanisms underlying the change of high gamma band is awaiting future studies, as there are multiple synaptic or network processes as candidates (see review, Uhlhaas et al.,44).
|
other
| 32.78 |
We revealed that the amplitude of gamma band increased dramatically soon after eye-opening, indicating sensory experience may contribute to the development of gamma oscillations necessary for many cognitive functions. This notion is supported by a recent study that dark-rearing impaired the normal expression of beta/gamma oscillations20, probably by disrupting the development of inhibitory synaptic transmission45. However, sensory-independent factors, mainly the maturation of cortical networks, may play an essential role46. As a result, the continuously developing gamma oscillation observed in this study and in previous reports in human and mice20,41 may be distinct from the gamma oscillation observed in somatosensory cortex in neonatal rat10, which is driven by bottom-up thalamo-cortical input. Taken together, this result – when combined with previous reports - suggests that gamma and other high frequency rhythmic activity in the visual cortex starts to develop from an early stage (around eye-opening), probably by a combination of experience-related and sensory-independent mechanisms. Future studies are needed to investigate the time course of development of gamma oscillation in other cortical regions and its potential causal relation to the development of brain function.
|
other
| 35.9 |
One novel finding of the current study is the decrease of synchronization in early development (in the level of spike-field and inter-LFP-frequency coupling). We found strong spike-phase locking in theta/alpha band as well as theta/alpha/beta amplitude coupling to delta frequency rhythms before eye-opening, which may indicate high correlation of neuronal spiking37 (also see Fig. 4A). The synchronization decreased after eye-opening, as revealed by decreased spike-phase locking and phase-amplitude coupling in our results. This is in agreement with previous findings of decreased noise-correlation with circuit maturation in both electrophysiology20,27 and calcium-imaging studies36,47. Interestingly, the decrease of spike locking to theta/alpha phase and the decrease theta/alpha/beta amplitude coupling to delta frequency is accompanied by increase of spike locking to high-gamma phase and increase of high-gamma amplitude coupling to theta/alpha phase. Such increases were even clearer in visual-induced activities. The decrease of spike locking to theta/alpha phase and the decrease of theta/alpha/beta amplitude to delta phase may reflect the transition of the network activity from high synchronization, which facilitates synaptogenesis48 by Hebbian plasticity to a less synchronized mode47 for reducing neural noise to enable efficient information processing49, to boost local and global network communication50, as well as to develop dynamic and optimal internal models representing the environment31,32. The increase of gamma power, spike locking to the high-gamma phase, and high-gamma amplitude coupling to theta/alpha may benefit the grouping and selecting in visual information processing51,52, modulate synaptic plasticity53,54, or reflect the maturation of the cortical excitatory-inhibitory balance55.
|
study
| 31.44 |
Our study also revealed that the visual responses became more temporally refined and that the visual cortex increased its capability to follow higher stimulus frequencies after eye-opening. Brain oscillations in adult humans are synchronized by temporally modulated visual and auditory stimuli and exhibit a preferred resonance frequency in the gamma band56,57. In human developmental studies, these gamma steady-state responses increase through childhood and adolescence58,59. This enhancement was correlated to the desynchronization of the background activity60, indicating that the maturation of networks is necessary for supporting the steady-state responses observed in adults. Our results in the juvenile ferret align with the above human studies and indicate that the development of sensory entrainment of brain oscillations can start as early as immediately after eye-opening. Future studies will need to examine the presence of entrainment at higher frequencies than those tested in the current study, and to elucidate the development of the network mechanisms sustaining such steady-state responses61. It is also an open question if similar stimuli presented to probe the brain activity of preterm infants would reveal features of network activity that may predict subsequent developmental trajectories.
|
other
| 35.3 |
The functional properties of the ferret visual cortex undergo dramatic changes after eye-opening26. Weak orientation and direction selectivity already exist before eye-opening but visual experience is needed for their full development39,40,62. Compared to orientation selectivity, direction selectivity relies more on visual experience in the first few days after eye-opening for its development39,40,62. This difference could be explained by the need of developing inhibition mechanisms to robustly discriminate the motions in opposite directions63–65. Our data revealed improved temporal properties of visual responses after eye-opening (Figs 6 and 8). This result, combined with our finding that the changes of synchronization properties could happen within few days after eye-opening (Supplemental Fig. 1), illustrates a coincidence between the development of the neuronal synchronization and the development of the direction selectivity. Our findings also support a recent suggestion that simultaneously activating (by visual or optogenetic stimulations) of neuron population with similar initial direction preference will permit the further strengthen of the direction maps66,67.
|
other
| 32 |
Although the present research focuses on the changes of rhythmic activities in visual cortex after eye-opening, the observed results of oscillations and synchronizations after eye-opening may not be limited to the visual cortex, nor should the eye-opening be viewed as an isolated event in development that has implications only for the development of the visual system. Recent research shows that the opening of the eyelid is controlled by the BMP/SMAD signaling pathways68. Interestingly, the BMP/SMAD pathways also play an active role in cortical neuronal development69, suggesting that eye-opening is just one of the landmarks reflecting the continuous unfolding of the developmental plan. The plan, determined by the cascades of gene expression but also later shaped by intrinsic and externally-driven brain activity, may lead to the changes in oscillations and synchronizations in the visual cortex as well as in other cortical regions. The notion that the changes of cortical oscillations around eye-opening is a general maturational process across cerebral cortex is supported by the finding of increased gamma oscillation in the rat barrel cortex after eye-opening (Fig. 2D,E, plots after P15 in Minlebaev et al.10). However, the exact form of the oscillatory patterns and the time course of their development may be different across the brain as suggested by the inter-regional difference of the patterns and timing of development in ferrets70. The appearance of increased oscillations across brain regions after eye-opening in ferrets is coincident with the emergence of new motor skills and behaviors71,72.
|
study
| 31.61 |
Our results show the development of gamma oscillations around the time of eye-opening, indicating the possible involvement of sensory experience in this process. A recent study also found that dark-rearing after eye-opening reduced but did not completely abolish the development of gamma oscillations in rat visual cortex20. However, sensory-independent factors46 may also play a role. Specifically in ferrets, several mechanisms may participate in the change of oscillations and synchronization observed around eye-opening. First, the segregation into local circuits through clustering of local horizontal connections in cortex starts around one week before eye-opening; and clustering refinement finishes around one to two weeks after eye-opening in ferrets73,74. This refinement of connections may reduce the synchronization of the MUA and LFP. Second, the AMPA and kainate receptors in ferret visual cortex increase around the time of eye-opening, and NMDA receptors constantly increase for 2 months after birth75. The enhancement of glutamate receptor level may contribute the increased high frequency LFP power observed in the present study. Third, the balance of inhibitory/excitatory neurons in the ferret visual cortex76,77 as well as the inhibitory synaptic connections78,79 continues to develop after eye-opening up until approximately postnatal day 60. The resulting increase in synaptic inhibition may shape the presence of gamma oscillations, in which synaptic inhibition plays an important role, at least in the adult animal80. Fourth, although having arrived at the cortex before eye-opening, thalamic axons keep modifying their targeting in visual cortex in ferrets up to 7 weeks81, suggesting that the bottom-up thalamic influences on cortical oscillations and synchronization may continue to develop after eye-opening. Finally, the differences in behavioral state between the two groups examined here82–84, e.g., increased locomotion, may correlate with the strength of gamma oscillations. A comparison of spike-phase locking and phase-amplitude coupling between spontaneous activities and visually-induced responses (Figs 4D, 5B–D, 7B–F) suggests that the oscillations and synchronizations before and after eye-opening may be state-dependent; however assays such as pupillometry in head-fixed animals are needed to adequately address this question.
|
other
| 33.47 |
One limitation is that the current study did not explore activities in a broader developmental range. So it is not clear whether gamma oscillations have already developed in very early age in ferret visual cortex like in other species and other sensory cortex10. Further studies are also needed to address how gamma oscillations develop in visual cortex after 8 weeks old. Another limitation is that the current study did not specify how the oscillations across different cortical layers develop before and after eye-opening. Brain oscillations differ in different layers in both adult and neonates. Depth-probes will be needed to be deployed to look at the layer specific difference in development. Finally, it is still unclear what causes the increase of gamma oscillation and the change of frequency coupling after eye-opening. More casual studies are required, e.g., using visual deprivation to study the role of visual experience.
|
study
| 31.77 |
In summary, our results revealed a profound change of higher-frequency LFP activity during development in the ferret visual cortex. The study of LFP in ferret visual cortex around eye-opening may provide a useful tool for the study of cortical oscillations during both typical and atypical development.
|
other
| 30.73 |
Seven pregnant ferrets (Mustela putorius furo) were acquired at the gestational age of 22–26 days from Marshall BioResources Inc. and housed individually until the delivery date (~E41). They were housed in a light cycle of 16 hr light/8 hr dark regimen throughout the pregnant and nursing periods to ensure the same breeding season cycle as maintained by the supplier. After birth, the kits were kept with their mothers (jills) from birth to the end of the experiment. Premium ferret diet (fat 15~20%, meat based protein 35~40%, Marshall Pet Products Inc., Wolcott, NY) was provided to the jills. The same food softened with warm water was used from 5 weeks of age to help wean the kits. A total of twenty kits (nine male and eleven female) were used in this study. All procedures were approved by the UNC – Chapel Hill Institutional Animal Care and Use Committee (UNC-CH IACUC) and in compliance with the guidelines set forth by the NIH (NIH Publications No. 8023, revised 1978) and USDA.
|
other
| 36.3 |
Electrode arrays were implanted in a brief surgical procedure one to three days before the electrophysiological recording session (recordings were performed at postnatal age P22-P50). Anesthesia was induced with four to five percent isoflurane in an induction chamber. Animals were maintained under anesthesia with 1.5%–3% isoflurane and 100% medical grade oxygen delivered through an anesthesia mask (David Kopf Instruments, Tujunga, CA). The depth of anesthesia was assessed by lack of toe-pinch response. Lidocaine (2%) was injected intramuscularly into the surgical site for local analgesia, and Furosemide (5%, 0.04 ml/kg) was injected intramuscularly to prevent cerebral edema. The electrocardiogram and breathing rate were monitored throughout the surgery to maintain deep general anesthesia (typical heart rate was 240 beats/minute and breathing rate was measured by a piezoelectric sensor). The body temperature was monitored with an infrared thermometer at least every 15 minutes. Hot-snap pads and a water heating blanket were used to maintain body temperature (36–38 °C) during surgery and subsequent recovery. Animals were fixed in a stereotaxic frame via ear bars (or ear cups if ear canals were closed) and a mouth bar (David Kopf Instruments). An initial scalp incision along the midline was made and a craniotomy was performed over primary visual cortex (area 17) located 1–3 mm anterior from lambda and 6–9 mm lateral from midline depending on animal age (Fig. 1B). The dura and pia were removed and a 2 × 8 electrode array (Innovative Neurophysiology, Durham, NC, 35 um tungsten electrodes with 200 um spacing, impedance of ~1 MΩs at 1 kHz, 0.5 mm shorter low impedance reference electrode) was lowered down into the cortex, oriented medial-laterally, until spikes or LFP signals were recorded. A silver ground wire was positioned on the surface of frontal cortex through an additional small craniotomy and was fastened to a nearby bone screw. The array was then fixed by dental cement which was anchored to the skull by bone screws. The incision site was sutured and triple antibody ointment was applied. Saline (10–20 ml/kg) warmed to body temperature was injected subcutaneously to prevent post-operative dehydration. After full recovery from anesthesia, the kit was returned to the litter. The body weight was measured twice a day for the following days to ensure proper recovery. Acetaminophen (Children’s Tylenol, 16 mg/kg) was administrated orally twice per day for at least 3 days after surgery for pain alleviation.
|
other
| 34.6 |
Recordings took place in the housing facility in a 68.5 × 51.5 × 49 cm (L X W X H) ferret cage which was light-insulated by thick black curtains. A blanket or bedding was added to the floor to keep the animal warm and comfortable. The animal was allowed to freely move within the dark cage for 10–15 minutes while spontaneous neural activity was recorded via a light-weight, wired head-stage (Intan,Los Angeles, CA, RHD2132; Fig. 1A). A subset of the animals (16/20) were then presented with visual stimuli of light flashes presented from four computer-controlled LED arrays (220 Lumen each array) positioned in each corner of the cage. The duration of each stimulus was 500 ms with an inter-stimulus interval 4.5–7.5 seconds. Flash stimuli were repeated 100–200 times. Of all visually-tested animals, 6 were further tested with another pattern of visual stimuli consisting of flashes at different frequencies (chirp stimuli). These recording sessions started with 50 seconds of continuous 10 Hz flashes (square wave of 50% duty cycle). This block of 10 Hz served to habituate the visual system, and to avoid a strong onset response for subsequent stimulation. Then periodic flash stimulation at the following frequencies was presented in either ascending or descending order, 0.25, 0.5, 1, 3, 5, 7, 9, 13, 17, 21, 25, 29 Hz. Stimulation at each frequency was presented for at least 6 cycles or 3 seconds, whichever was longer. Both ascending and descending frequency stimuli were repeated 10 times, and were intermingled and randomized. Between two ascending/descending stimulation blocks, there was 20 seconds of 10 Hz conditioning stimulus. Some animals (6/20, not used in the chirp stimuli test) were also tested with an acoustic stimulus that consisted of 100 ms 70 dB DSL broadband noise delivered by two speakers on both sides of the cage. However, no auditory responses were detected in visual cortex (aged P25 - P43, data not shown). Each recording session was limited to approximately one hour. We did not use individual animals for longitudinal studies because of the fast expansion of brain volume and the resulting challenges of maintaining a chronic implant, especially before eye-opening85.
|
other
| 33.97 |
A light-weight wired headstage (Intan, RHD2132) connected to the implanted arrays amplified and digitized the neural signal (sampling rate of 20 kHz), which was transmitted via a cable to the main control board of the electrophysiology acquisition system (Intan, RHD2000). A camera (Microsoft LifeCam Cinema 720p HD Webcam) was modified to be infrared sensitive and was mounted on the cage ceiling to track the position of the animal. A data acquisition device (USB-6212, National Instruments, Austin, TX) was used to control the LED arrays for visual stimulation and an infrared LED mounted within the field of view of the camera. The data acquisition system simultaneously sent a periodic “on” pulse signal to the electrophysiology acquisition system and to the LED in order to synchronize video and electrophysiology data for post-hoc analysis.
|
other
| 34.53 |
Immediately after recordings, animals were deeply anesthetized with an intramuscular injection of a ketamine/xylazine cocktail (40 mg/kg ketamine, 5 mg/kg xylazine). Electric currents (5 µA sine-wave, 10 second, by A395R Linear Stimulus Isolator, World Precision Instruments, Sarasota, FL) were passed between 2 to 3 sites on the multi-electrode array and the nearby reference electrode to produce lesions at the recording sites. The animal was then humanely euthanized with an overdose of sodium pentobarbital and immediately perfused with 0.1 M PBS followed by 4% formaldehyde. The brain was extracted and post-fixed for at least 4 days in 4% formaldehyde before being transferred to 30% sucrose. After sinking in sucrose, the brains were then cut into 50 µm sections using a cryostat (VT-1200, Leica Microsystems, Wetzlar, Germany) and processed for Nissl stainingto confirm the electrode tracks (Fig. 1C,D). These lesions were larger than the scale of the electrode size, which occurred when removing the array from the brain, and were exaggerated by subsequent histological processing.
|
other
| 36.06 |
Data were analyzed using custom-written scripts in Matlab (Mathworks, Natick, MA). Only data sections validated by video to be free of motion artifacts and in which the animal was awake (e.g. when body movements were observed) were used for further analysis. The raw data were notch-filtered at 60 Hz to remove line noise. A low-pass or high-pass filter with 300 Hz cut-off frequency (2nd order butterworth filter with the zero-lag) was used to extract LFP and MUA signals, respectively. The LFP spectrogram was computed by convolving LFP signals with a family of Morlet wavelets (0.5 to 120 Hz in 0.5 Hz steps). The mean power spectrum either across the whole recording session or in the burst periods (see below for the definition of burst periods) was estimated by averaging the square of the absolute value of the convolved signal across the time of interest. However, such estimation for low frequencies may be inaccurate if the time window is short. Thus, we only present results of the analysis on frequencies larger than 10 Hz for the power spectrum in burst periods. To account for the power law scaling of the LFP power spectrum, the power spectra were 1/f normalized by multiplying the each data point with its frequency. Spikes were extracted using a threshold of minus-five-times the standard deviation of the high-pass filtered signal. We used the method of a previous study38 to detect the burst events in neural data before eye-opening. Briefly, the inter-spike-interval (ITI) was calculated from the spike train of each spontaneous recording session (before eye-opening, ISI = 499.6 ± 1618.7 ms, mean ± SD; after eye-opening, ISI = 132.3 ± 418.8 ms). A burst was defined as a period within which: (1) the maximal ISI was less than 100 ms, and (2) a minimum of 10 spikes were counted.
|
other
| 38.97 |
To calculate spike cross-correlation, we computed the cross-correlogram of two spike trains binned into 20 ms windows. The lag time ranged from −250 ms to 250 ms with step length equals to the bin width. We found that for almost all channel-pairs, the peak of the cross-correlogram was at zero. Therefore, spike correlations were defined as the correlation coefficient of spike times. We also calculated the results using smaller bin widths of 5 ms or 2 ms and found that the choice of the bin width did not affect our finding.
|
other
| 35.2 |
To compute the phase-amplitude coupling and spike-field coherence, we convolved the LFP signal with a family of complex Morlet wavelets (2 to 128 Hz, with logarithmic step length of 21/8). The resulting signal was then used to estimate the instantaneous phase and amplitude LFP signals within specific frequency bands. To estimate the strength of spike locking to the phase of LFP oscillations, we computed the spike phase locking value (PLV)86. Spike PLV was defined as1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$PL{V}_{spk}({f}_{0})=\frac{1}{N}|\sum _{n=1}^{N}{e}^{i{\theta }_{n}^{spk}({f}_{0})}|$$\end{document}PLVspk(f0)=1N|∑n=1Neiθnspk(f0)|where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\theta }^{spk}$$\end{document}θspk is the instantaneous LFP phase at frequency f 0 at the time of spiking activity. This analysis projects the LFP phase at the time of each spike onto a unit circle. The magnitude of the mean of all unit vectors gives an indication of the strength of spike phase locking, with values close to 0 indicating no phase locking, and values close to 1 strong phase locking.
|
other
| 35.53 |
To control for the effect of different firing rates in different age groups on computing the spike-phase locking, we used a bootstrap method to estimate the spike PLV controlled for firing rate: 400 spikes were randomly selected with repetition to compute spike PLV. The same procedure was then repeated 5000 times and the mean resulting firing rate unbiased spike PLV was computed. To ensure the reliable estimation of spike PLV, channels where less than 400 spikes were detected for each recording session were discarded. To control for spike waveform contamination of the LFP signal, especially on high frequency components, we used the neighboring channel to estimate LFP phase for spike PLV analysis. Cross-frequency phase amplitude coupling (PAC) was computed to assess the degree with which high frequency oscillations are temporally organized by the phase of low frequency oscillations. PAC was defined as the phase locking value between a low frequency signal and amplitude fluctuations of a high frequency signal occurring at the lower carrier frequency. First, an LFP signal x(t) was convolved with a Morlet wavelet with carrier frequency f 1 2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$X(t,{f}_{1})=x(t)\,\ast \,w(t,{f}_{1})$$\end{document}X(t,f1)=x(t)∗w(t,f1)where * denotes the convolution operation and w(t, f 1) a Morlet wavelet with center frequency f 1. Then, the analytic amplitude of the same signal x(t) at a higher carrier frequency f 2 was convolved with a Morlet wavelet at the lower carrier frequency f 1.3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$X(t,{f}_{2})=|x(t)\,\ast \,w(t,{f}_{2})|\,\ast \,w(t,{f}_{1})$$\end{document}X(t,f2)=|x(t)∗w(t,f2)|∗w(t,f1)where X(t, f 2) captures fluctuations in the amplitude of the high frequency signal that occur at the lower carrier frequency. The PAC between carrier frequencies f 1 and f 2 is then defined as the phase locking value between X(t, f 1) and X(t, f 2).4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$PAC({f}_{1},{f}_{2})=\frac{1}{N}|\sum _{n=1}^{N}{e}^{i({\theta }_{n}^{diff})}|$$\end{document}PAC(f1,f2)=1N|∑n=1Nei(θndiff)|where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\theta }^{diff}$$\end{document}θdiff represents the phase lag between X(t, f 1) and X(t, f 2).
|
clinical case
| 28.25 |
For visual responses, spectrograms were computed per trial using wavelet convolution as described above, and then averaged across trials. The power spectrum of the elicited LFP was computed by averaging the spectrograms in the visual stimulus periods. The baseline power spectrum was estimated by averaging the spectrograms rate in the baseline period (500–1000 ms before stimulus onset). To compute the spike-phase locking and phase-amplitude coupling of visual responses, we used induced LFP (raw LFP subtracted by the event evoked potential on a trial-by-trial basis) to compute the phase and amplitude. We also used trial-shuffling to control the effect of phase reset by the visual stimulus. However, we found that subtraction of evoked potentials was sufficient to remove nearly all of the phase reset effect.
|
other
| 31.2 |
Peri-stimulus time histogram (PSTH) was made using a 100 ms smoothing window. The PSTH was then z-scored by subtracting the mean firing rate in the baseline period then dividing by the standard deviation. The population PSTH was then computed by averaging across all units. The 95% confidence interval was estimated using a bootstrap procedure on the individual PSTHs with 10,000 repetitions.
|
other
| 35.53 |
Visual response latency was estimated using a previous method87. Briefly, after binning the spike counts in 1ms bins, we fitted the Poisson distribution parameter of the baseline period (500 ms immediate before stimulus onset) across all bins and all trials. The response latency was taken as the first bin after stimulus onset that: (1) exceeded a P = 0.01 difference from background distribution and (2) the subsequent 2nd and 3rd bin exceeded a P = 0.05 level. The method relies on the accurate estimation of baseline distribution, thus it works well with a large number of trials (100 and more) as in the present study.
|
other
| 38.1 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.