MicroRNA control of the myogenic cell transcriptome and proteome: the role of miR-16

Abstract

MicroRNAs (miRs) control stem cell biology and fate. Ubiquitously expressed and conserved miR-16 was the first miR implicated in tumorigenesis. miR-16 is low in muscle during developmental hypertrophy and regeneration. It is enriched in proliferating myogenic progenitor cells but is repressed during differentiation. The induction of miR-16 blocks myoblast differentiation and myotube formation, whereas knockdown enhances these processes. Despite a central role for miR-16 in myogenic cell biology, how it mediates its potent effects is incompletely defined. In this investigation, global transcriptomic and proteomic analyses after miR-16 knockdown in proliferating C2C12 myoblasts revealed how miR-16 influences myogenic cell fate. Eighteen hours after miR-16 inhibition, ribosomal protein gene expression levels were higher relative to control myoblasts and p53 pathway-related gene abundance was lower. At the protein level at this same time point, miR-16 knockdown globally upregulated tricarboxylic acid (TCA) cycle proteins while downregulating RNA metabolism-related proteins. miR-16 inhibition induced specific proteins associated with myogenic differentiation such as ACTA2, EEF1A2, and OPA1. We extend prior work in hypertrophic muscle tissue and show that miR-16 is lower in mechanically overloaded muscle in vivo. Our data collectively point to how miR-16 is implicated in aspects of myogenic cell differentiation. A deeper understanding of the role of miR-16 in myogenic cells has consequences for muscle developmental growth, exercise-induced hypertrophy, and regenerative repair after injury, all of which involve myogenic progenitors.

INTRODUCTION

MicroRNAs (miRNAs) destabilize mRNAs and/or prevent their translation and are key regulators of myogenic stem cell (satellite cell) biology (1). These small noncoding RNAs control nearly every aspect of myogenic cell function including maintenance of quiescence (2–5), activation (6, 7), proliferation (8, 9), self-renewal (10), migration (11), and differentiation (8, 12, 13). The depletion of miRNAs from myogenic cells results in swift cell death (2, 8, 14). Dysregulation of miRNAs in satellite cells in vivo may negatively affect muscle development, adaptability to exercise, the muscle microenvironment with aging, and regenerative capacity after injury; all these processes depend on satellite cells (15–24). Furthermore, miRNAs are released from satellite cells in extracellular vesicles (EVs) and taken up by recipient cells throughout muscle in vivo (14, 25, 26). Satellite cell communication throughout muscle via miRNAs in EVs contributes to muscle adaptation during loading (27, 28). Understanding the role of miRNAs in myogenic progenitors may accelerate the discovery of miRNA-based therapeutics that affect muscle plasticity via fusion-dependent and/or -independent mechanisms (27, 29).

miR-16 is highly conserved and ubiquitously expressed (30–32). It was the first miRNA implicated in tumorigenesis and is a highly influential miRNA in numerous cellular processes, specifically as it relates to tumor growth (32, 33). In skeletal muscle tissue that contains both mature muscle fibers and myogenic satellite cells, miR-16 content rises during avian embryonic development but begins to wane at birth concomitant with secondary myogenesis (34). Low miR-16 levels during development are associated with muscle hypertrophy (35, 36). We reported that muscle miR-16 levels decline after resistance exercise following a week of acclimation to training and that knockdown in myotubes increases muscle protein synthesis (37). Skeletal muscle miR-16 is also lower in regenerating relative to uninjured muscle (6). The repression of miR-16 in muscle is therefore associated with postnatal myofiber growth and regeneration as well as the exercise-induced hypertrophic response. In myogenic precursor cells, miR-16 is induced upon activation (2, 6, 38) and enriched during proliferation (2, 14) but gradually declines in culture (2) and with differentiation (34, 36, 39). Some evidence suggests that miR-16-5p is high in quiescent satellite cells but still drops precipitously during differentiation (1). miR-16 overexpression prevents myogenic cell differentiation and myotube formation in vitro (34), whereas knockdown enhances these processes (34, 40). MyoD contributes to myogenic cell differentiation (41–48). miR-16 knockdown in murine myofiber-associated myogenic cell culture may increase the proportion of MyoD+ satellite cells by 3 days but reduces it by 5 days, pointing to an effect on myogenic cell behavior and fate (6). The literature collectively points to miR-16 having a key function in myogenic cells and in determining skeletal muscle mass. Despite its importance, how miR-16 exerts its effects in myogenic cells is incompletely defined.

The purpose of this investigation was to provide detailed information on how miR-16 influences myogenic cell biology. First, we evaluated miR-16 levels in mechanically overloaded muscle tissue in vivo, which is a well-established model of rapid growth associated with satellite cell differentiation and fusion (49). We then inhibited miR-16-5p in proliferating C2C12 myoblasts and performed RNA-sequencing (RNA-seq) and discovery proteomics. C2C12s express the satellite cell marker Pax7 (50) and can model the behaviors of primary myogenic progenitors in vitro (51–53). Our experiments point to a multifaceted role for miR-16 repression in myogenic cell differentiation. We thus shed light on potential mechanisms whereby miR-16 influences satellite cell-mediated muscle growth and regenerative potential in vivo (6, 36). We also provide general information on miR-16 targets, which may be informative in nonmuscle cell types where miR-16 is enriched (30).

MATERIALS AND METHODS

Murine Experiment

Female C57BL6/J mice (2–3 mo of age) were utilized for in vivo overload experiments. These experiments were performed for a prior publication from our laboratory (54), and the mice were injected with 5-ethenyl uridine 5 h before being euthanized. Experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Kentucky. Mice were housed in a temperature- and humidity-controlled room maintained on a 14:10-h light-dark cycle, and food and water were provided ad libitum throughout experimentation. Briefly, synergist ablation surgery to overload the plantaris muscle was performed as described by Murach et al. (54). Under isoflurane anesthesia, a portion of the gastrocnemius and soleus was removed, leaving the plantaris to be mechanically overloaded during reambulation (n = 4). Sham-operated mice (no removal of muscle) were controls (n = 3). Seventy-two hours after surgery, animals were euthanized in the morning via a lethal dosage of pentobarbital sodium injected intraperitoneally followed by cervical dislocation. Plantaris muscle tissue was harvested and frozen in liquid nitrogen and stored for downstream analyses. miR-16 level in tissue was evaluated with an unpaired directional t test in GraphPad Prism (Boston, MA).

Cell Culture Experiments

C2C12 murine myoblasts (CRL-1772, ATCC, Manassas, VA) were plated in six-well plates (5 × 10⁴ cells/well) with 2 mL of Dulbecco’s modified Eagle medium (DMEM) combined with 20% fetal bovine serum (FBS) and 1% penicillin-streptomycin (P/S). Cells were incubated at 37°C with 5% CO₂, and medium was changed every 48 h. Upon 40–50% confluence, cells were washed with PBS for transfection (37). For proteomics five technical replicates were utilized, and for transcriptomics three control and two knockdown wells were employed. miR-16 levels in cells were evaluated in Prism with an unpaired directional t test with n = 5 technical replicates.

Transfection of C2C12 Myoblasts

Plasmids encoding either an empty vector control (pCMVmiR; PCMVMIR, Origene, Rockville, MD) or miR-16-1 (miR-16-5p) inhibitor were transferred into DH5-a Escherichia coli as we have previously described (37). Plasmid DNA was amplified and isolated from bacteria with the PureLink HiPure Plasmid Filter Maxiprep kit (K211017, Life Technologies, Carlsbad, CA). For miR-16 inhibition, we utilized anti-miR inhibitor (AM17000, Ambion, Austin, TX). Plasmid DNA (1 μg) was diluted in 50 μL of Opti-MEM reduced serum medium (31985088, Life Technologies), combined with 4 μL of Lipofectamine 2000 (11668019, Life Technologies) diluted in 50 μL of Opti-MEM, and incubated for 20 min to allow lipid/DNA complexes to form. Medium was replaced with Opti-MEM, and lipid/DNA complexes were added and incubated for 5 h at 37°C with 5% CO₂. After 5-h incubation, medium was replaced with normal growth medium for 18 h. The efficiency of the inhibition of miR-16 was tested and validated previously (37).

RNA Isolation, Quality Check, and qPCR for miR-16

Myoblast and muscle tissue RNA were isolated with TRIzol reagent, which was then homogenized with a Polytron or bullet homogenizer. After homogenization, RNA was isolated with the phase separation method by addition of chloroform or bromochloropropane followed by centrifugation. The aqueous phase was transferred to a new 1.5-mL tube, and an equal amount of 70% of diethyl pyrocarbonate (DEPC)-treated ethanol was added. RNA isolation was further processed with the Invitrogen RNA isolation kit (K145002, Invitrogen, Carlsbad, CA) or the Zymo Direct-zol kit (Zymo Research, Irvine, CA). RNA concentration was determined with a BioTek Take3 microplate with a BioTek PowerWave XS microplate reader as previously described (55) or NanoDrop (ThermoFisher Scientific). RNA samples were only accepted if 260 nm-to-280 nm ratio was >2.0. Samples were stored at −80°C until further use.

Reverse transcription (RT) of miR-16 was performed with the TaqMan MicroRNA Reverse Transcription Kit (4366596, Applied Biosystems, Waltham, MA). Briefly, 200 ng of total miRNA was added to a master mix comprised of 1 µL of a 10× TaqMan probe (PN4427975, Applied Biosystems) for U6 and miR-16 each (RT:001973 and RT:000391, respectively, Applied Biosystems), 0.3 µL of dNTPs, 3 µL of MultiScribe RTase, 1.5 µL of 10× RTase buffer, 0.19 µL of RNase inhibitor, and nuclease-free water added up to 15 µL of total volume. RT reaction occurred with a hold step of 30 min at 16°C, followed by 30 min at 42°C and finally 5 min at 85°C. Samples were held at 4°C until further analysis. RT-qPCR was performed with the QuantStudio 3 Real-Time PCR system (Applied Biosystems). A 20-µL reaction composed of an adequate amount of TaqMan probes plus TaqMan Fast Advanced Master Mix (4444556, Applied Biosystems) was used to amplify cDNA. Samples followed a protocol consisting of incubation at 95°C for 4 min, followed by 45 cycles of denaturation, annealing, and extension at 95°C and 60°C. TaqMan probes were measured at the end of the extension step of each cycle. Fluorescence-labeled 20× TaqMan probes (PN4427975, Applied Biosystems) included U6 and miR-16 (TM:001973 and TM:000391, respectively, Applied Biosystems). Results were analyzed with QuantStudio Software. Cycle threshold (C_T) was determined, and the ΔC_T value was calculated as the difference between C_T value and U6 C_T value. U6 C_T values were not different between experimental conditions. Final quantification of gene expression was calculated with the ΔΔC_T method. Relative quantification was calculated as $2^{\mathrm{-}\Delta \Delta {\mathrm{C}}_{\mathrm{T}}}$ and expressed as arbitrary units.

RNA-Sequencing and Transcriptomic Analyses

RNA-sequencing was performed by NovoGene as previously described by us (54). Standard 150-bp paired-end sequencing was performed, and read counts were >20 million. Raw counts from RNA-sequencing were used as inputs into Partek Flow. Alignment was performed with STAR with mmu39. After low-expressed genes were filtered, DESeq2 (version 1.34.0) was used for normalization and differential analyses to identify differentially expressed genes (DEGs) with pairwise comparisons (56). DEGs were identified with a false discovery rate (FDR, step-up procedure) adjusted P value < 0.05. DEGs with adjusted P value < 0.05 were used for downstream functional analysis using ConsensusPath DB (57). Pathway analysis was conducted using the mouse overrepresentation feature, up- or downregulated DEGs, and the Reactome database with default settings. For pathway analysis, q values were used to determine significance based on DEGs with adjusted P < 0.05.

Proteomics: FASP bHPLC-Orbitrap Fusion

Proteomics was performed at the University of Arkansas for Medical Sciences Proteomics Core. Protein samples were reduced, alkylated, and digested by filter-aided sample preparation (58) with sequencing-grade modified porcine trypsin (Promega). Tryptic peptides were separated into 46 fractions on a 100 × 1.0-mm Acquity BEH C18 column (Waters) with an UltiMate 3000 UHPLC system (Thermo) with a 50-min gradient from 99:1 to 60:40 buffer A-to-B ratio under basic pH conditions and then consolidated into 12 superfractions; buffer A = 0.1% formic acid, 0.5% acetonitrile, and buffer B = 0.1% formic acid, 99.9% acetonitrile. Each superfraction was then further separated by reverse-phase XSelect CSH C18 2.5-μm resin (Waters) on an in-line 150 × 0.075-mm column with an UltiMate 3000 RSLCnano system (Thermo). Peptides were eluted with a 45-min gradient from 98:2 to 65:35 buffer A-to-B ratio. Eluted peptides were ionized by electrospray (2.4 kV) followed by mass spectrometric analysis on an Orbitrap Fusion Tribrid mass spectrometer (Thermo). MS data were acquired with the FTMS analyzer in profile mode at a resolution of 240,000 over a range of 375 to 1,500 m/z. After HCD activation, MS/MS data were acquired with the ion trap analyzer in centroid mode and normal mass range with normalized collision energy of 28–31% depending on charge state and precursor selection range. Proteins were identified by database search using MaxQuant (Max Planck Institute) label-free quantification with a parent ion tolerance of 2.5 ppm and a fragment ion tolerance of 0.5 Da. Scaffold Q + S (Proteome Software) was used to verify MS/MS-based peptide and protein identifications. Protein identifications were accepted if they could be established with <1.0% false discovery and contained at least two identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm (59). Protein abundance is presented with intensity-based absolute quantification (iBAQ) values. To determine differential expression and pathway analysis between conditions, P values based on unpaired nondirectional t tests were employed with a cutoff of P < 0.05. Benjamini–Hochberg-adjusted P values were subsequently calculated from these P values. Untranslated region (UTR) sequences for genes that resulted in proteins of interest were downloaded from UCSC Genome Browser and used for prediction of miR-16 binding sites. The potential binding site was identified with RNAhybrid software (60, 61), which combines thermodynamic and seed sequence information.

RESULTS

miR-16 Is Lower after 72-h Synergist Ablation Mechanical Overload of the Plantaris Muscle

Previous work suggests that muscle miR-16 levels are inversely related to postnatal muscle growth in chickens (34, 36) and are lower during regeneration in mice (24). We showed that miR-16 is lower in muscle tissue after a bout of resistance exercise in rats (37). Previous work showed significant changes in global muscle miRNA levels in the early phase of mechanical overload-induced hypertrophy (62). To determine whether miR-16 declines during the early phase of a well-characterized model of rapid loading-induced hypertrophy in mice (49, 63, 64), we compared miR-16-5p levels in 72-h mechanically overloaded (MOV) plantaris muscle to sham-operated muscles. miR-16 was 28% lower in muscle tissue during MOV (P = 0.06) (Fig. 1A). Overall, low miR-16 appears characteristic of growing muscle in several settings and species (6, 35–37).

Figure 1.

miR-16 levels in overloaded muscle in vivo and RNA-sequencing (RNA-seq) analysis in proliferating C2C12 myoblasts 18 h after miR-16 knockdown (KD). A: miR-16 levels in plantaris skeletal muscle after 72 h of synergist ablation mechanical overload of the plantaris (MOV) relative to sham operation. B: in vitro study design. C: miR-16 levels in control and KD myoblasts. D: levels of ribosomal protein genes in control and KD myoblasts. E: levels of p53- and Pol II transcription-related genes in control and KD myoblasts. For RNA-seq experiments, n = 3 control and n = 2 KD technical replicates. *adj. P < 0.05.

miR-16 Influences Ribosomal-, p53-, and Pol II Transcription-Related mRNA Abundance

The in vitro study design is found in Fig. 1B. The transfection of a miR-16 inhibitor was effective at knocking down miR-16 (miR-16 KD) in proliferating C2C12 myoblasts (Fig. 1C). Inhibition of miR-16 in myoblasts upregulated mRNA levels of Rps12, Rps26, Rplp1, Rpl22l1, and Rpl35 (adj. P < 0.05), which encode ribosomal proteins (Fig. 1D). Additional ribosomal protein genes tended to be higher after miR-16 knockdown (Rpl27 and Rpl27a, adj. P < 0.10) (Supplemental Table S1). The major processes downregulated by miR-16 inhibition were Regulation of TP53 Activity through Methylation (Reactome, q = 0.0014) and RNA Polymerase II Transcription (Reactome, q = 0.0026) (Supplemental Table S2). In the former pathway, Atm, Ep300, and Mdm2 were lower (adj. P < 0.05). In addition to those three genes, Aff4, Crebbp, Kmt2a, Kmt2c, Mga, Nr2c2, and Sox9 were lower after miR-16 knockdown in the latter pathway (adj. P < 0.05) (Fig. 1E). The long noncoding RNA (lncRNA) H19 was elevated with miR-16 inhibition (adj. P = 0.0035), and the lncRNAs Malat1 (adj. P = 0.059) and Xist (adj. P = 0.009) were lower (Supplemental Table S1).

miR-16 Knockdown Increases Metabolism-Related Proteins and Specific Markers of Myogenic Cell Differentiation

Proteomics was performed in five technical replicates per condition. After miR-16 knockdown, five proteins were upregulated (EEF1A2, HIST1H1B, NPC2, PRPH, and TSEN54) and seven proteins were downregulated (GLRX2, GOLPH3, GRK6, LAPTM4A, NDC1, THAP4, and ZC3H11A) with an adjusted P value <0.05, and 899 proteins achieved significance at P < 0.05 (Supplemental Table S3). Proteins involved in different aspects of Metabolism (Reactome, q = 0.0178) were upregulated (assessed using proteins with P < 0.05) (Fig. 2A) (Supplemental Tables S3 and S4). Within the broad Metabolism pathway, various tricarboxylic acid (TCA) cycle proteins were selectively enriched (Reactome, q = 0.00293). Proteins related to Metabolism of RNA (Reactome, q = 0.000456) were most downregulated by miR-16 inhibition (Supplemental Table S4). Markers of myogenic cell maturation, including myosin light chains (MYL6B, MYL9, MYL12A, P < 0.05) as well as smooth muscle actin (ACTA2, P = 0.04, adj. P = 0.41) (Fig. 2B), were upregulated with miR-16 KD. Specific proteins that have a defined role in satellite cell behavior were also altered. In addition to muscle-enriched EEF1A2 (P = 0.0000164, adj. P = 0.018) (Fig. 2C) (65–67), OPA1 (P < 0.05, adj. P = 0.25 and 0.30) (Fig. 2D) (68, 69) and PTEN (P < 0.05, adj. P = 0.41) (Fig. 2E) (70–72) were elevated by miR-16 inhibition; the latter two proteins are of interest but did not achieve significance according to adjusted P values. OPA1 was detected twice (canonical 111 kDa and truncated 34 kDa), and the levels of both isoforms were higher with knockdown (P < 0.05) but not according to adjusted P values.

Figure 2.

Proteomic analysis in proliferating C2C12 myoblasts 18 h after miR-16 knockdown (KD). A: Reactome pathway analysis using upregulated differentially expressed proteins (P < 0.05). B–E: thermodynamic and seed sequence target prediction for miR-16 and protein levels of ACTA2 (B), EEF1A2 (C), OPA1 (D), and PTEN (E) in control and KD myoblasts. iBAQ, intensity-based absolute quantification. For all experiments, n = 5 technical replicates were utilized. *P < 0.05. Created with BioRender.com.

An overlap in gene and protein expression would not necessarily be expected at a discrete time point since protein translation lags behind changes in mRNA expression. Nevertheless, comparison of gene to protein levels revealed agreement for Bmpr2, Cltc, Lama5, Lpp, Mga, Nptx1, and Sema5a, all of which were lowered by miR-16 knockdown (P < 0.05). Rpl27a, which was trending to increase at the gene level, was elevated at the protein level (P = 0.01, adj. P = 0.28). Since we observed upregulation of proteins with miR-16 knockdown, and miRNAs can prevent protein translation without altering transcript levels (14, 73), we performed a 3′ UTR binding affinity analysis of miR-16 for ACTA2, EEF1A2, OPA1, and PTEN (Fig. 2, B–E), using RNAhybrid (60, 61). Based on free energy, the results collectively suggested that miR-16 could regulate the levels of these proteins. The highest complementarity between the miR-16 seed sequence and 3′ UTR sequence was for EEF1A2, OPA1, and PTEN (Fig. 2. C–E).

DISCUSSION

Satellite cells differentiate and begin fusing appreciably to muscle fibers between days 4 and 5 of mechanical overload in adult mice (25) and ∼5 days after transplantation into muscle of mice (74). In our 72-h overload experiments, the contributions of miR-16 from satellite cells versus other cell types in muscle tissue (e.g., muscle fibers or immune cells) (75) cannot be discerned. The influence of hypertrophy per se versus a degeneration/regeneration response that can occur with synergist ablation (19, 76) is also unclear. Nevertheless, our observation of lower miR-16 in murine muscle tissue after 72 h of MOV corresponds with reduced muscle miR-16 levels during early in vivo regeneration after injury (6) and during recovery from a bout of resistance exercise (37). To model miR-16 regulation of myogenic cell behavior, we repressed it in proliferating myoblasts and performed transcriptomic and proteomic profiling. These data provide a framework for understanding miR-16’s role in myogenic cell fate.

miR-16-5p can induce p53 signaling in myogenic cells (34). Repression of p53-related gene expression with miR-16 knockdown dovetails with this observation. In C2C12s in vitro, the expression of ribosomal proteins is upregulated during early differentiation (77, 78). Enrichment of ribosomal protein genes is associated with satellite cell fusion into the myofiber syncytium during hypertrophy (79). The induction of ribosomal protein genes, specifically Rps26, by miR-16 inhibition may characterize myogenic cells that are primed for fusion (79). The lncRNA H19 was elevated with miR-16 knockdown, whereas Malat1 and Xist were lower. lncRNAs are not typically targeted for decay by miRNAs since translation is seemingly required for RNA destabilization (80). H19 induction (81–84) and Malat1 reduction (85) are strongly implicated in myogenic cell differentiation, but how miR-16 could affect lncRNA levels in myogenic cells deserves further investigation. In concert with evidence suggesting that miR-16 can target Myomaker, the gatekeeper of myogenic cell fusion that gradually increases during differentiation (86, 87), our RNA-seq data collectively point to declining miR-16 levels facilitating lineage progression toward differentiation.

At the protein level, repressing miR-16 in proliferating myoblasts leads to several alterations that are indicative of miR-16’s roles in the regulation of differentiation. The enrichment of myosin light chains, noted in our data, is a sign of myogenic cell maturation (88, 89). Upregulation of smooth muscle actin (Acta2) is also strongly associated with myoblast differentiation (90–93). Satellite cell differentiation is controlled by a metabolic shift and progression from glycolysis to the TCA cycle (94). Elevated TCA cycle proteins with miR-16 repression therefore seems intuitive from a metabolic perspective. Induction and phosphorylation of EEF1A2 is linked to myogenic cell differentiation (65–67). EEF1A2 is muscle enriched (95) and highly regulated during muscle development (96), protects myotubes from cell death (65), and controls Utrophin levels in skeletal muscle (97, 98). EEF1A2 is also a core component of cardiomyocyte differentiation (99). The causal functions of EEF1A2 control by miR-16 in myoblast differentiation deserve further study. Higher PTEN and OPA1 with miR-16 inhibition is noteworthy since both are implicated in reinforcing satellite cell quiescence (68, 70, 71). PTEN can repress the satellite cell identity gene Pax7 (72, 100) but may also facilitate a return to quiescence (70, 71) that could ensue if differentiation and fusion does not progress. Whereas OPA1 supports quiescence by maintaining mitochondrial integrity (68), OPA1 induction and mitophagy is an essential component of successful C2C12 myoblast differentiation (69).

The inhibition of miR-16 in proliferating myoblasts, which occurs naturally during myogenic differentiation, reveals its contributions to this process. Specifically, we uncover miR-16’s regulation of ribosomal, p53-related, and lncRNA gene expression as well as metabolic- and muscle maturation-related protein abundance. A more detailed understanding of miR-16 dynamics, specifically in satellite cells in vivo during different stages of myogenesis, regeneration, and hypertrophy, will inform how miR-16 controls muscle mass in varying circumstances. The present study is limited to a single time point with an immortalized cell line. We also do not report on myogenic cell behavior after miR-16 inhibition, although this has been documented in detail elsewhere (6, 34, 40). Limitations aside, we provide insights from two -omic layers to expand our understanding of miR-16 regulation in myogenic cells.

DATA AVAILABILITY

Processed data are provided in Supplemental Tables S1–S4 and were deposited in GEO (GSE229134).

SUPPLEMENTAL DATA

Supplemental Tables S1–S4: https://doi.org/10.6084/m9.figshare.22151624.

GRANTS

This work was supported by NIH Grant R00 AG063994 to K.A.M. Support for proteomics was provided by an Arkansas Biosciences Institute grant to N.P.G.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

N.P.G. and K.A.M. conceived and designed research; S.L., D.L., F.M., P.J.K., and F.v. performed experiments; S.L., F.M., P.J.K., I.J.V., and K.A.M. analyzed data; S.L., F.M., I.J.V., and K.A.M. interpreted results of experiments; S.L., I.J.V., and K.A.M. prepared figures; S.L., D.L., and K.A.M. drafted manuscript; S.L., D.L., F.M., P.J.K., I.J.V., F.v., N.P.G., and K.A.M. edited and revised manuscript; S.L., D.L., F.M., P.J.K., I.J.V., F.v., N.P.G., and K.A.M. approved final version of manuscript.