p38α MAPK Regulates Adult Muscle Stem Cell Fate by Restricting Progenitor Proliferation During Postnatal Growth and Repair

Authors


  • Author contributions: P.B.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; D.P.: collection and/or assembly of data, data analysis and interpretation, and manuscript writing; S.W.: conception and design, data analysis and interpretation, and manuscript writing; D.O.: collection and/or assembly of data; J.M.P.: conception and design, financial support, and manuscript writing.

Abstract

Stem cell function is essential for the maintenance of adult tissue homeostasis. Controlling the balance between self-renewal and differentiation is crucial to maintain a receptive satellite cell pool capable of responding to growth and regeneration cues. The mitogen-activated protein kinase p38α has been implicated in the regulation of these processes but its influence in adult muscle remains unknown. Using conditional satellite cell p38α knockout mice we have demonstrated that p38α restricts excess proliferation in the postnatal growth phase while promoting timely myoblast differentiation. Differentiation was still able to occur in the p38α-null satellite cells, however, but was delayed. An absence of p38α resulted in a postnatal growth defect along with the persistence of an increased reservoir of satellite cells into adulthood. This population was still capable of responding to cardiotoxin-induced injury, resulting in complete, albeit delayed, regeneration, with further enhancement of the satellite cell population. Increased p38γ phosphorylation accompanied the absence of p38α, and inhibition of p38γ ex vivo substantially decreased the myogenic defect. We have used genome-wide transcriptome analysis to characterize the changes in expression that occur between resting and regenerating muscle, and the influence p38α has on these expression profiles. This study provides novel evidence for the fundamental role of p38α in adult muscle homeostasis in vivo. STEM Cells 2013;31:1597–1610

Introduction

Adult tissue homeostasis relies on the timely response of resident stem cell populations. Acute injury and routine tissue maintenance stimulate quiescent stem cells to activate and differentiate, restoring tissue capability, while self-renewing to maintain a receptive stem cell pool. In this study, we use muscle as a paradigm for assessing the regenerative response of adult stem cells to growth and repair processes.

The development and repair of skeletal muscle is a tightly regulated process, requiring the maintenance of a reservoir of resident muscle stem cells capable of responding to growth and regeneration cues. Satellite cells are the principal muscle stem cell [1–4], responsible for the majority of postnatal growth and repair [5], and are capable of remarkable regenerative capability [6]. In response to extracellular stimuli, they activate and undergo a crucial proliferative expansion, before either returning to quiescence or differentiating and fusing to existing myofibers. Many muscle pathologies, such as age-related atrophy and muscular dystrophies, involve insufficient or exhausted satellite cell proliferation, impeding the maintenance of muscle homeostasis and crucial repair mechanisms [7, 8].

Among a number of signaling pathways associated with the fate “decisions” of satellite cells, the p38 mitogen-activated protein kinase (MAPK) pathway plays an integral role [9–11]. There are four vertebrate isoforms of p38 (p38α, p38β, p38γ, and p38δ), which are activated via homologous TGY dual phosphorylation by the MAPK kinases MKK3 and MKK6 [12, 13]. Once activated, they phosphorylate serine/threonine residues on downstream substrates, including a range of transcription factors and protein kinases. The α and γ isoforms are most highly expressed in muscle [14], while studies using the pyridinyl imidazole compound SB203580 have suggested that p38α/β are the crucial isoforms during myogenic differentiation [15–18].

The function of p38 during myogenesis is complex, with activation of p38 during the myogenic process recruiting the SWI-SNF chromatin remodeling complexes to specific myogenic loci [19] as well as modifying gene expression post-transcriptionally via p38-mediated regulation of myogenic mRNA stability [20]. p38 increases myogenic regulatory factor (MRF) activity via phosphorylation of MEF2 [21] and E proteins, promoting heterodimerization of MRFs such as MyoD with E47. This enhances RNA Pol II recruitment to myogenic loci, initiating the differentiation program [22–24]. In parallel, p38 antagonizes the proliferation-promoting JNK pathway through upregulation of the JNK phosphatase MKP-1 [25]. More recently, p38 has been linked to the epigenetic regulation of myogenesis through both repressive [26, 27] and activating [28] histone modifications. p38γ acts in opposition to p38α, blocking premature differentiation through induction of a repressive MyoD transcriptional complex [26]. The p38 isoforms therefore regulate satellite cell fate at multiple levels, their activity representing a key “gateway” in the regulation of myogenesis.

Conditional tissue-specific p38α-null mice have revealed the isoform's importance in inflammation [29–31], proliferation [32], and differentiation [31, 33]. Satellite cells cultured ex vivo from prenatal mice deficient in nonplacentally expressed p38α show significant deficiencies in differentiation [34], but due to the neonatal lethality of p38α-null mice [35–37] the contribution of p38α in adult muscle in vivo remains unexplored. Using novel conditional satellite cell p38α-null mice, we have investigated the role of p38α in muscle growth and repair, revealing a distinct role for p38α in the postnatal regulation of satellite cell expansion. We have demonstrated that p38α acts to restrict excessive proliferation during the postnatal hyperproliferative phase as well as directing myoblasts toward a differentiated state in response to both growth and repair stimuli. In addition, we have demonstrated a significant upregulation of active p38γ in p38α-null satellite cells, with inhibition of p38γ activity reducing the ex vivo proliferative phenotype.

Materials and Methods

Mice, Regeneration Studies, and Satellite Cell Isolation

All experiments were performed in accordance with Babraham Institute animal ethics committee regulations under the terms of the UK Home Office. To conditionally delete p38α (MAPK14), mice containing floxed alleles of MAPK14 [32] (purchased from MRC Harwell) were crossed with mice expressing Cre recombinase under the Pax7 promoter [38] (provided by Mario Capecchi—University of Utah). All mice were homologous for the MAPK14 floxed alleles, with or without one copy of the Pax7-Cre transgene. Unless stated, 7–9-week-old age- and sex-matched mice were used.

To induce activation of satellite cells in postnatal muscle, 100 μl of 10 μM cardiotoxin (Sigma-Aldrich, Gillingham, UK, http://www.sigmaaldrich.com) was injected into the tibialis anterior/extensor digitorum longus muscle group (TA/EDL) under isoflurane anesthesia and the muscles harvested after 5, 10, 25, and 50 days. For proliferation quantification, the Click-iT EdU Alexa Fluor 488 Imaging Kit (Invitrogen) was used as instructed. Briefly, 0.5 mg/ml EdU in phosphate buffered saline was injected intraperitoneally at 0.1 mg per 20 g b.wt. 48 hours and 24 hours before the muscles were harvested. The click chemical reaction was performed after immunostaining with primary and secondary antibodies (see below) and before DAPI staining. Fluorescence-activated cell sorting (FACS) isolation and satellite cell culture was performed as previously described [39].

Immunofluorescence and Histology

Immunofluorescence on cultured satellite cells and ex vivo muscle was performed as previously described [39, 40].

Bright-field images were acquired using a Qimaging micropublisher 3.3RTV camera on an Olympus BX41 microscope. Immunofluorescence images were acquired on an Olympus FV1000 confocal microscope. For fiber size analysis, laminin (α-2 chain) immunofluorescence was used to denote myofiber boundaries and cross-sectional area was calculated using ImageJ software (NIH). Total fiber number was quantified from tiled images of TA/EDL cross-sections.

Antibodies

Primary antibodies: Pax7 (Developmental Studies Hybridoma Bank, Iowa City, IA, http://dshb.biology.uiowa.edu), MyoD (Santa Cruz Biotechnology, Heidelburg, Germany, http://www.scbt.com), laminin α-2 chain (Enzo Life Sciences, Exeter, UK, http://www.enzolifesciences.com), myogenin (Santa Cruz Biotechnology), skeletal fast myosin (Sigma-Aldrich), skeletal slow myosin (Sigma-Aldrich), myosin heavy chain (Developmental Studies Hybridoma Bank), pH3 Ser10 (Cell Signaling Technology, Hitchin, UK, http://www.cellsignal.com), Ki67 (Leica, Milton Keynes, UK, http://www.leicabiosystems.com), and M-cadherin (Santa Cruz Biotechnology). Secondary antibodies: donkey anti-rabbit AF488, goat anti-mouse (IgG1 specific) AF568, and goat anti-rat AF633 (Life Technologies, Paisley, UK, http://www.invitrogen.com).

Western Blotting

Western blotting was performed as previously described [40], with the following antibodies (in addition to those above): p38 (Cell Signaling Technology), phospho-p38 (Cell Signaling Technology), p38α (R&D Systems, Abingdon, UK, http://www.rndsy stems.com), and GAPDH (Abcam, Cambridge, UK, http://www.abcam.com).

Coimmunoprecipitation

Protein A dynabeads (Invitrogen) were washed with RIPA buffer (50 mM Tris-HCl pH 7.4, 1% Nonidet P40, 0.25% sodium deoxycholate, 150 mM NaCl, 1 mM EDTA, supplemented with 1 mM phenylmethanesulfonyl fluoride, and protease inhibitor mix) and resuspended in 100 μl RIPA. Five microliters of pan phospho-p38 antibody (Promega, Southhampton, UK, http://www.pro mega.co.uk) was added, and rotated at 4°C overnight. Two hundred micrograms of muscle lysate was added and rotated for 1 hour. Immunoprecipitates were extensively washed, separated on polyacrylamide gels, and transferred to nitrocellulose membranes. Precipitated proteins were revealed by Western blot with p38γ (R&D Systems) and p38α antibodies.

Mass Spectrometry Analysis

Muscle lysates were separated by SDS-PAGE gel and stained with Coomassie. Regions of gel lanes, corresponding to the migration position of p38, were excised, then destained, reduced, carbamidomethylated, and digested overnight with trypsin (Promega sequencing grade 10 ng/ml in 25 mM ammonium bicarbonate at 30°C). Digests were analyzed by nanoLC-MS/MS on a LTQ Orbitrap Velos mass spectrometer (Thermo Scientific, Loughborough, UK, http://www.thermoscientific.com) fitted with a nanoelectrospray ion-source. Peptides were separated on a reversed-phase column (0.075 × 100 mm2; ReproSil-Pur 120 C18-AQ 3 μm) using an acetonitrile gradient (0%–35% in 60 minutes at 300 nl/minute) containing 0.1% formic acid. The mass spectrometer was set up for targeted analysis of the phosphorylation state of the peptides HTDDEMTGYVATR (p38 alpha) and QADSEMTGYVVTR (p38 gamma) using an inclusion list comprising the calculated m/z values of the monoisotopic doubly charged pseudomolecular ions of the nonphosphorylated, monophosphorylated, and diphosphorylated forms of both peptides, in the unoxidized and methionine-oxidized states. Full scan data were acquired over the m/z range 350–1,600 at a nominal resolution setting of 30,000, followed by up to 10 HCD spectra of ions from the inclusion list at 35% relative collision energy. The p38 peptide spectra were identified by searching the mass spectral data against mouse entries in the Uniprot database using Mascot software (Matrix Science). All p38 peptide MS/MS spectra were manually validated. Apparent stoichiometries of phosphorylation were calculated from the peak areas of the extracted ion chromatograms of the phospho- and corresponding nonphosphopeptides, assumed equal sensitivity to phosphorylated and nonphosphorylated peptides.

Statistical Analysis

For all quantitative analyses presented, a minimum of three replicates were performed. Data are presented as means ± SEs. ANOVA with post hoc tests and Student's t tests were performed as appropriate (indicated in figure legends), using GraphPad Prism.

RT-qPCR, mRNA Seq, and Analysis

RT-qPCR was performed as described previously [39]. To isolate activated, proliferating satellite cells from regenerating muscle, the right and left gastrocnemius of knockout and control mice (four mice per genotype) were injected with 50 μl of 10 μM cardiotoxin. Twenty-four hours after injury, the muscles were harvested, activated satellite cells FACS isolated, and flash frozen in TRIzol reagent (Life Technologies). Satellite cell RNA was also isolated from uninjured muscle as described above (eight mice per genotype). Total RNA was isolated using standard phenol chloroform extraction methods. RNA quality was checked using a Pico chip on Bioanalyzer (Agilent, Wokingham, UK, http://www.agilent.com), and sequencing libraries prepared using Illumina TruSeq RNA sample prep kit v2 (Illumina, Saffron Walden, UK, http://www.illumina.com) according to the manufacturer's instructions. The libraries were run on a HiSeq 1000 (Illumina).

Reads were aligned to NCBI build 37 mouse genome, and corrected log 2 transformed Reads Per Million values obtained using SeqMonk software (Babraham Institute). Differentially expressed genes were identified using an intensity difference analysis (p < .05) (supporting information Table S1 for detailed explanation) and selected genes validated using qPCR. Gene ontology (GO) analysis was performed using DAVID [41].

Results

An Absence of p38α in Satellite Cells Impairs Postnatal Skeletal Muscle Growth

p38α is the crucial isoform in skeletal muscle [14, 34, 42], therefore we wanted to elucidate its role in skeletal muscle in vivo. As germline p38α-null mice are embryonic lethal we crossed p38α floxed mice with the Pax7-Cre line, producing satellite cell p38α knockout mice (Pax7 is transiently expressed in some embryonic neural cells [43]). Using this strategy, p38α would therefore be deleted in the whole Pax7 lineage, from satellite cells through to the myofiber. Skeletal muscle from the p38α conditional null mice (subsequently referred to as “p38αKO”) exhibited an absence of p38α protein (Fig. 1A). PCR analysis showed the 411 bp band resulting from deletion of exons 2 and 3 present in the p38αKO mouse (supporting information Fig. S1A). p38αKO mice developed a modest but significant growth defect at around 3 weeks of age (Fig. 1B), apparent in both sexes (supporting information Fig. S1B) and continuing into adulthood. This was due to reduced skeletal muscle mass (Fig. 1C), while other organ weights were unaffected (Fig. 1D). To determine the cause of this reduced muscle mass, myofiber cross-sectional area was calculated using muscle sections immunostained for laminin α-2 chain, denoting the myofiber boundary (Fig. 1E). The p38αKO mice had significantly smaller myofibers (Fig. 1F), but total fiber number remained unchanged (supporting information Fig. S1C). These data support the hypothesis that the decreased muscle mass is due to a lack of postnatal hypertrophic growth, not embryonic myofiber development (as fiber number is defined at birth [44]).

Figure 1.

Conditional knockout of p38α impairs skeletal muscle growth. (A): Western blot for p38α (with GAPDH loading control) showing lack of p38α protein in p38αKO mice. Two representative mice of each genotype shown. (B): Whole mouse body weights (female): means ± SEM; n ≥ 15 for each time point. **, p < .01, significantly different from p38αfl, assessed by ANOVA. (C): Tibialis anterior/extensor digitorum longus (TA/EDL) muscle weight (relative to total body weight) scatter plot: n = 10. *, p < .05, assessed by Student's t test. (D): Heart and kidney weights (relative to total body weight) scatter plot: n = 5 and 10, respectively. No significant differences, assessed by Student's t test. (E): Immunofluorescent staining on cross-sections of TA/EDL for laminin α-2, denoting myofiber boundaries, costained with DAPI. Scale bars = 100 μm. (F): Quantitation of the proportion of myofibers per cross-sectional area (average cross-sectional area also shown). Approximately 700 myofibers assessed per n, n = 3. ***, p < .005, assessed by ANOVA. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; GAPDH, Glyceraldehyde 3-phosphate dehydrogenase.

Conditional Satellite Cell Knockout of p38α Increases Satellite Cell Number Postnatally Through an Increased Proliferative Expansion

We next wanted to examine the consequences of p38α knockout within the satellite cell population. At birth, and consistent with the gross muscle morphology described above, there was no difference in satellite cell number (as shown by Pax7+ nuclei) between p38αKO and p38α floxed control (subsequently called “p38αfl”) muscle (Fig. 2A, 2B), supporting the observed mouse weights, with the p38αKO weight defect only developing postnatally. To understand the role of p38α in postnatal muscle function, satellite cell number quantification was performed at P10 and P21, during the postnatal hyperproliferative phase (P1–P21), revealing a significant increase in satellite cell number between p38αKO and p38αfl muscle during this phase (Fig. 2A, 2B). By P49 there are 50% ± 8% more satellite cells in the p38αKO muscle than the p38αfl muscle (Fig. 2A, 2B). We hypothesized that this increase in satellite cell number was due to increased progenitor proliferation postnatally. To confirm this, we performed Ki67 immunostaining on the sections, observing a significant increase in Ki67+/Pax7+ cells—proliferating satellite cells—at P10 (Fig. 2C). With an increased pool of proliferating progenitors, we might predict increased subsequent differentiation. In the p38αKO sections, however, this was not observed, as an equivalent number of myogenin+ cells was observed in p38αfl and p38αKO sections throughout the postnatal hyperproliferative phase (Fig. 2D). These data support a shift of balance from differentiation toward self-renewal.

Figure 2.

Conditional knockout of p38α increases satellite cell number in vivo through an increased proliferative expansion. (A): Immunofluorescence time course during postnatal muscle growth. Sections from newborn (P0), 10-day old (P10), and 21-day old (P21) mice immunostained for Pax7 and DAPI; sections from 7-week-old (P49) mice immunostained for Pax7, laminin α-2, and DAPI. Arrowheads indicate Pax7+ cells. Scale bars = 100 μm. (B): Quantitation of Pax7+ cell number per field of view (×40) during postnatal growth, log transformed: means ± SEM. *, p < .05; **, p < .01; ***, p < .005, assessed by Student's t tests. (C): Proliferating satellite cells (Pax7/Ki67 dual positive cells) per field of view (×40) during postnatal hyperproliferative growth phase: means + SEM. *, p < .05, assessed by Student's t test. (D): Terminally differentiating satellite cells (Myogenin+ cells) per field of view (×40) during postnatal hyperproliferative phase: means + SEM. No significant differences, assessed by Student's t test. (B–D): 15 images assessed per n, n = 3. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; FOV, field of view.

p38α-Null Satellite Cells Exhibit a Proliferative Expansion Ex Vivo at the Expense of Differentiation

To demonstrate that this myogenic defect was inherent within the satellite cell population, and not due to signaling from the p38α-null myofiber, quiescent satellite cells were isolated from mouse skeletal muscle. The isolation process activates the satellite cells, which can then be monitored throughout their myogenic progression ex vivo. More p38αKO satellite cells maintained Pax7 expression at 120 hours after isolation (Fig. 3A, 3B), demonstrating an increase in Ki67+ proliferative progenitors (Fig. 3C), with fewer committed to terminal differentiation as indicated by myogenin expression (Fig. 3A, 3D). In concert with this effect, fewer multinucleated myotubes were present in the p38α-null cultures at 120 hours and 168 hours postisolation (Fig. 3A, 3E), although mononuclear MHC+ cells were still observed. Interestingly, by 168 hours postisolation the defects in proliferation and commitment to differentiation were no longer apparent, suggesting that there is a temporal aspect to p38α's actions which results in a delay in the myogenic process instead of a total block in differentiation. This hypothesis is reinforced by the presence of multinucleated myotubes in the p38α-null cultures, demonstrating that a proportion of myoblasts are still undergoing terminal differentiation. Protein analysis confirmed this shift of balance away from differentiation, with increased Pax7 and decreased MHC observed in the p38αKO cultures at 120 hours (Fig. 3F). This proliferative increase is therefore inherent within the satellite cell pool, as isolation and culture ex vivo support the postnatal growth defect described.

Figure 3.

Isolated satellite cells exhibit a proliferative expansion ex vivo at the expense of differentiation. (A): Immunofluorescent staining for Pax7 and Myogenin 120 hours post-fluorescence-activated cell sorting (FACS) isolation of satellite cells (left panels) and MHC and Myogenin 168 hours post-FACS isolation, costained with DAPI. Scale bars = 100 μm. (B): Pax7+ cells per field of view (×40) as a percentage of total nuclei, 120 hours and 168 hours postisolation: means + SEM. **, p < .01, assessed by ANOVA. (C): Ki67+ cells per field of view (×40) as a percentage of total nuclei, 120 hours and 168 hours postisolation: means + SEM. *, p < .05, assessed by ANOVA. (D): Myogenin+ cells per field of view (×40) as a percentage of total nuclei, 120 hours and 168 hours postisolation: means + SEM. ***, p < .005, assessed by ANOVA. (E): Number of multinucleated myotubes (MHC-expressing myotubes with ≥3 nuclei) per field of view (×40), 120 hours and 168 hours postisolation: means + SEM. (B–D): 15 images quantified per n, n = 3. (F): Western blot of satellite cells 120 hours postisolation, with antibodies against MHC, Pax7, total p38, and GAPDH loading control. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; GAPDH, Glyceraldehyde 3-phosphate dehydrogenase.

The absence of p38α in satellite cells ex vivo demonstrates a similar myogenic shift of balance observed when wild-type satellite cells are isolated and treated with SB203580 (SB). FACS-isolated satellite cells were cultured in the presence of 5 μM SB or DMSO control for 120 hours. The SB-treated cultures showed an increased cell expansion ex vivo (supporting information Fig. S2A), retaining a twofold higher number of Pax7+ myogenic progenitor cells than the control cultures (supporting information Fig. S2B). These progenitors demonstrated increased proliferation with SB treatment (supporting information Fig. S2C). In accordance with this, fewer differentiating, myogenin+ cells were present (supporting information Fig. S2A). Some multinucleated MHC+ myotubes were present in SB-treated cultures (supporting information Fig. S2D)—interestingly a smaller decrease than was observed in p38α-null satellite cells (Fig. 3E). Protein analysis of the SB-treated cultures confirmed these differences (supporting information Fig. S2E). Inhibiting p38α/β's action in wild-type activated satellite cells therefore causes a similar, but not identical, ex vivo phenotype to that observed in satellite cells isolated from p38αKO muscle.

An Absence of p38α Results in Dysregulation of the Transcriptome in Satellite Cells from Resting and Regenerating Muscle

In order to investigate global transcriptome changes involved in the described myogenic variations, mRNA-seq analysis was performed on freshly isolated satellite cells from p38αfl and p38αKO mice (resting muscle). In addition, to examine the effects of p38α-knockout in regenerating muscle (i.e., in active, proliferating satellite cells), satellite cells were isolated from cardiotoxin-treated muscle, 24 hours after injury. Loss of p38α resulted in a number of upregulated and downregulated transcripts in both resting and regenerating muscle samples (Fig. 4A, 4B). The majority of changing transcripts were downregulated in p38αKO satellite cells, with only two upregulated transcripts identified from regenerating muscle (Fig. 4C, 4D). Interestingly, all differentially expressed transcripts were unique to either the resting or regenerating muscle, suggesting little overlap between p38α's targets in resting muscle with those upon activation in response to injury (supporting information Tables S1, S2, respectively). GO analysis of the downregulated transcripts from resting muscle p38α-null satellite cells revealed high enrichment for signaling-related genes (Fig. 4E). In addition, a significant proportion were associated with the Wnt signaling pathway (p = .0024), known to be associated with p38 signaling in other lineages [45, 46]. Similarly, GO analysis of downregulated transcripts from regenerating muscle identified a large proportion of signaling-related genes (Fig. 4F). In addition, and extending previous observations, other downregulated GO terms included genes associated with TNF receptors, upstream activators of p38 signaling [27], implying the presence of a feedback loop.

Figure 4.

Transcriptome changes in p38α-null satellite cells from resting and regenerating muscle. (A, B): Scatter plots comparing global gene expression profiles between p38αfl and p38αKO satellite cell RNA isolated from resting muscle (A) and regenerating muscle (B). R-Value represents coefficient of correlation. Differentially expressed genes (assessed by intensity difference, p < .05) are circled. (C): Representation of differentially expressed genes in p38α-null satellite cells isolated from resting muscle (top) and regenerating muscle (bottom). Each bar represents one gene, a bar under the axis implies downregulation and above the axis implies upregulation. (D): Heat maps of differentially expressed genes from resting (left) and regenerating (right) muscle. Each row represents one gene, with red representing high expression through green to blue representing low expression. Upregulated and downregulated genes are separated by a black line. (E): Gene ontology analysis of genes downregulated in p38αKO satellite cells isolated from resting muscle. Fold enrichment and gene count under each term shown. (F): Gene ontology analysis of genes downregulated in p38αKO satellite cells isolated from regenerating muscle. Fold enrichment and gene count under each term shown. Abbreviations: EGF, epidermal growth factor; TNFR, tumor necrosis factor receptor.

p38α-Null Satellite Cells Exhibit an Altered Transcriptional Response to Cardiotoxin-Induced Injury

To characterize the changes in the transcriptome that occur upon injury, p38αfl transcript profiles were compared between resting and regenerating muscle. While satellite cell transcriptome analysis has been performed using microarray analysis of regenerating mdx mice [47] and barium chloride-injured mice immediately following injury [48], the profiles of satellite cells from acutely injured regenerating muscle have not been investigated until now. RNA-seq has the advantage of a larger dynamic range in terms of expression pattern, allowing greater insight into the changes in gene expression occurring in different satellite cell contexts. The expression profiles were less highly aligned than for either of the comparisons with p38αKO (R = 0.840), supporting a large change in global gene expression in response to cardiotoxin-induced injury (Fig. 5A). The differentially expressed genes consisted of a number of both upregulated and downregulated transcripts (Fig. 5B, 5C), which included known p38-associated regeneration factors such as MyoD [19] (supporting information Table S3). GO analysis of the upregulated transcripts included cell cycle-related genes, as predicted, while GO analysis of the downregulated transcripts included genes associated with signaling and cell adhesion.

Figure 5.

p38α-null satellite cells exhibit a different transcriptional response to cardiotoxin-induced injury than control satellite cells. (A): Scatter plot comparing global gene expression profiles between p38αfl satellite cells from resting muscle (y-axis) and regenerating muscle (x-axis). R-Value represents coefficient of correlation. Differentially expressed genes (assessed by intensity difference, p < .05) are circled. (B): Heat map of differentially expressed genes identified between p38αfl satellite cells from resting (Rest.) and regenerating (Regen.) muscle. Each row represents one gene, with red representing high expression through green to blue representing low expression. Upregulated and downregulated genes are separated by a black line. (C): Representation of differentially expressed genes between p38αfl satellite cells from resting and regenerating muscle. Each bar represents one gene. (D): Scatter plot comparing global gene expression profiles between p38αKO satellite cells from resting muscle (y-axis) and regenerating muscle (x-axis). R-Value represents coefficient of correlation. Differentially expressed genes (assessed by intensity difference, p < .05) are circled. (E, F): Venn diagram of genes upregulated (E) and downregulated (F) upon regeneration from p38αfl (red circle) and p38αKO (green circle) satellite cells. Numbers in the circles represent the number of unique genes to that area of the diagram. Numbers outside the circles represent the total number of genes in each population. Gene ontology analysis shown for the nonoverlapping genes in each population.

We next investigated the differences in expression that occur upon regeneration in the absence of p38α signaling in the satellite cells. In total, fewer differentially expressed genes were identified, with a higher coefficient of correlation (R = 0.901) suggesting a more subdued expression change (Fig. 5D). As expected, a proportion of the p38αKO upregulated and downregulated genes overlapped with the changes observed in the control satellite cells (Fig. 5E, 5F). Interestingly, however, this overlap was restricted to six genes in the upregulated dataset and three genes in the downregulated dataset. There were therefore 10 genes no longer upregulated in the absence of p38α, and 28 genes uniquely upregulated only in the absence of p38α. GO analysis of these populations revealed genes associated with cell cycle progression and DNA replication were upregulated only in the absence of p38α (Fig. 5E). This strongly supports the proliferative increase observed throughout the study, as an absence of p38α results in the upregulation of proliferation-associated genes upon satellite cell activation. These data demonstrate that global gene expression changes are associated with the absence of p38α, and that the satellite cell regenerative response is altered on a transcriptome level, upregulating proliferation-associated genes which results in the increase in satellite cell number observed postnatally.

Acute Injury in p38αKO Mice Results in a Prolonged Satellite Cell Response and an Increased Stem Cell Pool

Postnatal muscle development and adult regeneration appear to involve distinct mechanisms [49], and our data suggest p38α has different targets in satellite cells isolated from resting muscle from satellite cells isolated from regenerating muscle. To investigate whether the upregulation of proliferation-associated genes in response to injury results in a parallel satellite cell number increase to that observed postnatally (i.e., whether or not the phenotype was restricted to the postnatal growth phase), cardiotoxin-induced regeneration was examined, stimulating the satellite cell population to respond to an acute muscle injury. This would also determine whether any proliferative increase is at the expense of self-renewal (examining the ability of satellite cells to maintain a stem cell pool after regeneration) or differentiation (examining the ability of p38αKO to fully regenerate after injury). Both p38αfl muscle and p38αKO muscle had similar gross muscle morphology during and after regeneration, with centrally nucleated fibers (indicative of myofiber regeneration) abundant by 25 days postinjury (DPI) (Fig. 6A). During regeneration, the satellite cell pool expands in response to the injury, before differentiating to repair the damaged area and returning to a baseline satellite cell number, maintaining the stem cell population [39, 50]. While the p38αfl muscle responded to the muscle injury as expected by displaying a 10-fold satellite cell number increase and returning to baseline levels by 25 DPI, the p38αKO muscle displayed a heightened satellite cell expansion (Fig. 6B and representative five DPI images supporting information Fig. S3A). This corresponds with the upregulation of proliferation-associated genes identified by RNA-seq analysis (Fig. 5E). Importantly, by 25 DPI satellite cell numbers had not reduced to preinjury numbers, despite the apparent complete regeneration of muscle. This corresponded with an increased proliferative response as measured by EdU incorporation, with more Pax7+/EdU+ cells observed across the regeneration time course (Fig. 6C). Interestingly, while all satellite cells in the p38αfl muscle had returned to quiescence by 25 DPI, some proliferating, EdU+ satellite cells were still present in the 25 DPI p38αKO muscle sections. The increased satellite cell population was enhanced in p38αKO muscle as regeneration progressed, until by 50 DPI there were 150% more satellite cells in the p38α-null muscle than the p38αfl muscle (Fig. 6D). By this time none of the satellite cells were EdU+, suggesting that the proliferative satellite cell population had returned to quiescence, resulting in an increased pool of stem cells to respond to future injury.

Figure 6.

Cardiotoxin treatment in p38αKO mice results in impaired early regeneration and a prolonged satellite cell response. (A): Representative tibialis anterior/extensor digitorum longus (TA/EDL) sections during the regeneration time course 5, 10, and 25 DPI, stained with hematoxylin and eosin (5 days, 10 days, and 25 days left panels), and immunostained for laminin α-2 and DAPI (25 days right panels). Scale bars = 100 μm. (B): Satellite cell response to cardiotoxin injury, quantitated as Pax7+ cell number per field of view (×40) in undamaged muscle and 5, 10, and 25 DPI: means ± SEM. ***, p < .005 assessed by Student's t tests. (C): Proliferating satellite cells (Pax7+ cells having incorporated EdU after two 24 hour pulses) quantitated per field of view (×40) in undamaged muscle and 5, 10, and 25 DPI: means ± SEM. *, p < .05; ***, p < .005, assessed by Student's t tests. (D): p38αKO muscle Pax7+ cell number in response to cardiotoxin injury represented as a percentage increase over p38αfl muscle: means + SEM. (E): Percentage of centrally nucleated myofibers per field of view (×40) assessed from TA/EDL sections immunostained for laminin α-2 and DAPI (see (A) right panels): means + SEM. **, p < .01 assessed by Student's t test. (B–E): 15 images quantified per n, n = 3. (F): Average myofiber cross-sectional area in untreated muscle and 25 days postcardiotoxin injury (CTX treated): means + SEM. Abbreviations: DPI, days postinjury; FOV, field of view.

We next wanted to ascertain whether or not this proliferative increase was at the expense of differentiation (as documented ex vivo), therefore potentially resulting in a regeneration defect in p38αKO muscle. Accordingly, at 10 DPI, the p38α-null muscle had significantly fewer centrally nucleated myofibers than the floxed control (Fig. 6E). By 25 DPI, however, this defect was no longer apparent, mirroring the delay in differentiation observed in the satellite cells ex vivo (supporting information Fig. S3B). Indeed, by 25 DPI the average p38αKO fiber size was similar to that of uninjured p38αKO mice, suggesting that regeneration had occurred to an equivalent extent to the control mice (Fig. 6F). As observed in the resting muscle, the proportion of small fibers was significantly greater in p38αKO muscle (supporting information Fig. S3C). Collectively, these data suggest that a lack of p38α causes a proliferative increase in regenerating muscle similar to that observed in the postnatal growth phase. This results in an increased satellite cell pool during regeneration which is still present in the fully regenerated muscle. The delay in differentiation causes a temporary regenerative defect in early regeneration, but by 25 DPI p38αfl muscle and p38αKO muscle demonstrate equivalent regeneration, with fiber sizes returning to preinjury levels.

p38α-Null Satellite Cells Upregulate Active p38γ, with Inhibition of p38γ Restoring Ex Vivo Satellite Cell Function

We next wanted to determine whether the other p38 isoforms had a functional role in the p38αKO muscle phenotype. p38γ is also highly expressed in muscle [14], while p38β and p38δ show low and undetectable expression, respectively (supporting information Fig. S4A). We therefore investigated p38γ activity in p38αKO muscle lysates, revealing a considerable upregulation of phospho-p38γ in parallel with the reduction in phospho-p38α (Fig. 7A), confirmed by coimmunoprecipitations pulling down p38γ with the pan-pp38 antibody (Fig. 7B). Single replicate mass spectrometry allowed quantification of this increase in p38γ activity, demonstrating that while in p38αfl muscle only 2% of the p38γ is phosphorylated, in p38αKO muscle this increases to 24%, in parallel with a decrease of a similar percentage in p38α (Fig. 7C). p38γ protein levels remained unchanged between the two samples. This is the first evidence of compensatory p38 isoform activity in muscle, and we therefore wanted to determine whether the upregulation of p38γ phosphorylation was contributing to the muscle phenotype observed. Satellite cells were isolated and treated with 1 μM BIRB796 (Merck Millipore, Nottingham, UK, http://www.merckmillipore.co.uk), a selective small molecule inhibitor of all isoforms of p38 [51], therefore allowing us to investigate the effects of p38γ inhibition on the already p38α-null p38αKO satellite cells. The cultures were analyzed at 120 hours postisolation, the time point at which the myogenic defect was observed ex vivo (Fig. 3). Treatment with BIRB796 resulted in an increase in myogenin+ cells to p38αfl levels (Fig. 7D, 7E), while the number of multinucleated myotubes returned to that observed in p38αfl cultures (Fig. 7D, 7F). The proliferative (Ki67+) population was reduced concurrently (supporting information Fig. S4B). This suggests that the ex vivo myogenic defect observed was not solely due to an absence of p38α as would be expected, but additionally a substantial upregulation of p38γ activity in the satellite cells. p38γ acts antagonistically to the role of p38α, positively regulating proliferation of the myogenic precursors [26], hence inhibiting this activity in p38αKO satellite cells allows for resumed myogenic progression.

Figure 7.

Treatment with BIRB796 rescues ex vivo proliferation defect. (A): Western blot of muscle lysates, with antibodies against pan phospho-p38 (pp38) showing pp38α and pp38γ, total p38γ, and GAPDH loading control. (B): Immunoprecipitation of muscle lysates with pp38 antibody; bound proteins were revealed by Western blot with antibodies against p38γ and p38α (control). Arrowhead denotes heavy chain band. (C): Mass spectrometry analysis of muscle lysates showing the apparent stoichiometry of phosphorylation of p38 isoforms. No p38α present in p38αKO lysate, while total p38γ levels remained unchanged between lysates. (D): Immunofluorescence for Pax7 and Myogenin (top panels) and MHC (bottom panels) in satellite cells 120 hours post-fluorescence-activated cell sorting isolation, treated with 1 μM BIRB796 (BIRB) or DMSO control (Ctrl). Scale bars = 100 μm. (E): Myogenin+ cells per field of view (×40) as a percentage of total nuclei, in BIRB-treated and control cultures: means + SEM. **, p < .01, assessed by ANOVA. (F): Number of multinucleated myotubes (MHC expressing myotubes with ≥3 nuclei) per field of view (×40), in BIRB-treated and control cultures: means + SEM. **, p < .01, assessed by ANOVA. Abbreviations: BIRB, BIRB796; DAPI, 4′,6-diamidino-2-phenylindole; GAPDH, Glyceraldehyde 3-phosphate dehydrogenase.

Discussion

The ability of muscle to grow postnatally and regenerate in response to injury relies on the continued activity of satellite cells. Tight control over the balance between self-renewal and differentiation ensures timely muscle repair while still maintaining a functioning progenitor pool. The p38 isoforms are key regulators of proliferation and differentiation in numerous lineages [31–33], and while p38α's antiproliferative role in adult cardiac muscle has been described [52], until now adult skeletal muscle has not been investigated. We have shown that conditional knockout of p38α in the Pax7-expressing muscle lineage results in a mouse growth defect occurring in the satellite cell postnatal hyperproliferative phase. Reduced myofiber size in the p38αKO muscle and unchanged total fiber number confirm that postnatal hypertrophic growth is diminished by lack of p38α while embryonic development remains unaffected. This may be due to the postnatal reliance on Pax7+ progenitors [53–55], while embryonically Pax3+ as well as Pax7+ populations are required for growth [56]. Compensation by a Pax3+ population may therefore ameliorate any defect present in the Pax7+ population being detected during embryonic growth.

The postnatal p38αKO satellite cell population displayed increased proliferation both in vivo and ex vivo, establishing that the myogenic defect is inherent within the satellite cells and not due to signaling from p38α-null myofibers. We have demonstrated that an absence of p38α results in an increased satellite cell pool, still capable of proliferation and, crucially, differentiation. Multinucleated MHC+ myotubes were also present in SB-treated cultures—interestingly displaying a smaller decrease over controls than was observed in p38α-null satellite cells. This could result from the prolonged effects of p38α knockout in p38αKO satellite cells, compared with short-term inhibition ex vivo. The accumulation of mononuclear MHC+ cells suggests an additional defect in cell fusion, an observation supported by the reduced size of p38αKO fibers in vivo. This mirrors the fusion defect described in myoblasts lacking the p38 target gene CD53, which displayed a reduced fusion index in vitro and reduced fiber size in vivo [57].

The fact that the proliferative increases ex vivo were only observed at 120 hours and not at 168 hours suggests a crucial temporal aspect in p38's influence. This is reinforced by the observation that regeneration in vivo was delayed but by 25 DPI fiber size and centrally nucleated fiber number were equivalent in p38αfl and p38αKO muscle. Interestingly, after regeneration in p38αKO muscle the reservoir of satellite cells remained increased, while satellite cell number in control muscle quickly returned to preinjury levels. This could suggest an over-promotion of satellite cell self-renewal, while seemingly not at the expense of regeneration.

Through the use of genome-wide transcriptome analysis, we have characterized the early satellite cell response to acute injury. Previous work has demonstrated transcriptome changes in Pax3+ satellite cells population (from abdominal muscles) in response to injury [47], and in conjunction with this data, our analyses could be used to define the different responses of muscle to chronic (e.g., in dystrophic muscle) or acute (cardiotoxin-treated) injury as well as the responses of different satellite cell populations. The presence of known regeneration factors such as the DNA synthesis-associated gene Rrm2 [58] in the upregulated gene list suggests the usefulness of our data to identify novel regeneration markers and regulators of satellite cell function. Regeneration in the absence of p38α revealed markedly different gene expression compared with that of the p38αfl satellite cells. The presence of cell cycle and DNA replication genes upregulated uniquely in the p38α-null satellite cells strongly supports the satellite cell response to injury observed in vivo, with a greater satellite cell expansion due to an upregulation of proliferation-enhancing genes. These results confirm previous descriptions of the role of p38 in cell cycle exit. Positive cell cycle regulators such as cyclin D1 are upregulated upon induction of differentiation in p38α-null myoblasts [34], and we have demonstrated other cell cycle regulators such as Cdk1 to be upregulated in regenerating p38αKO muscle. Myogenin expression mediates cell cycle exit, and knockdown of p38α inhibits myogenin expression [57]. In addition, p38γ represses myogenin activity through induction of a MyoD repressive complex [26], so an absence of p38α and increased activity of p38γ would accentuate this cell cycle-promoting effect.

As described in Figure 7, p38α knockout is accompanied by a substantial increase in phosphorylated p38γ. This is an important finding and suggests that the antagonistic actions of the two isoforms may account for the phenotype observed. p38 isoforms have homologous phosphorylation sites, so it is likely that upstream kinases such as MKK3/MKK6 phosphorylate p38γ in the absence of p38α. Crosstalk between p38 isoforms has been suggested in p38γ-null fibroblasts [59], but this the first evidence of coregulatory p38 functions in adult stem cells. Previous studies support the observations in p38αKO muscle being due to the absence of p38α [15–18, 34] but the described upregulation of p38γ activity appears to have a combinatorial role in these myogenic effects, hence inhibition of p38γ results in a substantially decreased myogenic defect ex vivo. The balance between p38α and p38γ is therefore essential in maintaining adult muscle homeostasis, with the differentiation-promoting activity of p38α balanced by p38γ's role in maintaining the proliferative state.

Conclusion

This study has demonstrated crucial roles for p38α and p38γ in adult satellite cell homeostasis, balancing appropriate proliferation with timely differentiation. Importantly, we have demonstrated that a lack of p38α, in combination with increased p38γ activity, results in an increased satellite cell pool, still capable of responding to muscle injury. Indeed, after injury the pool of quiescent satellite cells is further enhanced, suggesting that further research into p38 could aid in potential therapeutic treatments for muscular dystrophies and age-related disorders.

Acknowledgements

We thank Mario Capecchi for the Pax7-Cre mouse strain. Illumina sequencing was performed by Kristina Tabbada and sequencing analysis aided by Simon Andrews. We thank BBSRC for competitive strategic funding to J.M.P. via The Babraham Institute, studentship funding for P.B. and S.W., and Responsive-mode funding (BB-H019243-1) to D.P. and J.M.P. D.P. is currently affiliated with the Department of Pharmacology, University of Cambridge, Tennis Court Road, Cambridge CB2 1PD, U.K. and S.W. is currently affiliated with the Centre for Stem Cells and Regenerative Medicine, Kings College London, London SE1 9RT, U.K.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.

Ancillary