Search
Close this search box.

Neurofibromin 1 controls metabolic balance and Notch-dependent quiescence of murine juvenile myogenic progenitors – Nature Communications

Premature cell cycle exit and impaired myogenic differentiation of Nf1Myf5 MPs reduces myonuclear accrual and MuSC numbers

Mice with conditional inactivation of Nf1 targeted to myoblasts by using Myf5Cre (Myf5Cre;Nf1flox/flox, or “Nf1Myf5”) reduce mTORC1-dependent anabolic myofiber growth during postnatal development in a variety of fore- and hind limb muscles, already visible during the first 3 weeks of postnatal life31. During juvenile development, however, muscle growth occurs by a combination of myonuclear cell accrual and metabolic growth3,7,8, so we analyzed Pax7+ juvenile MP behavior in Nf1Myf5 mice. Reverse transcription–quantitative polymerase chain reaction (RT-qPCR) and RNA-sequencing (RNA-Seq) confirmed efficient decrease in Nf1 messenger RNA (mRNA) in fluorescence-activated cell sorting (FACS)-isolated MPs at postnatal day 7 (p7) (Supplementary Fig. 1a). Nf1 expression was unaltered in p14 MPs of Nf1-haploinsufficient Myf5Cre;Nf1flox/+ mice (Supplementary Fig. 1b), which were used as controls for all further experiments; male and female animals were mixed in both groups. Myf5Cre mice are haploinsufficient for Myf5; however, Myf5 expression was not affected in p7 MPs of Myf5Cre/+ mice (Supplementary Fig. 1c). Both observations overlap our previous findings from Nf1Myf5 muscle tissue31, suggesting that common compensatory mechanisms are in place.

Proportions of proliferating of Pax7+ cells assessed by Ki67 immunolabeling of tissue sections showed a slight reduction in Nf1Myf5 muscle at p7, with a high decrease at p14 (Fig. 1a). At p21, Pax7+ cells appeared mostly non-proliferative in Nf1Myf5 muscles (Fig. 1a, b). A low fraction of adult MuSCs is in the cell cycle33, which was also seen at p84 in controls but not in Nf1Myf5 Pax7+ MuSCs (Fig. 1a). Freshly FACS-isolated p14 MPs after cytospin confirmed a reduced fraction of proliferating cells (Fig. 1c). In addition, p14 Nf1Myf5 MPs showed increased Pax7 protein abundance, indicated by fluorescence intensity measurement (Fig. 1d).

Fig. 1: Premature cell cycle exit and impaired differentiation of Nf1Myf5 MPs reduces myonuclear accrual and MuSC numbers.
figure 1

a Ki67+/Pax7+ cells quantification relative to all Pax7+ cells in TA muscles of control and Nf1Myf5 mice at indicated time points. p, postnatal day (n = 3 animals per genotype; p-values shown). b Representative immunolabeling images of Pax7 (red), Ki67 (green), and DAPI (nuclei, blue) of p21 muscle sections of control or Nf1Myf5 mice. Arrows indicate Pax7+/Ki67+ cells (n = 3 animals per genotype). c Cytospin of FACS-isolated p14 MPs from control or Nf1Myf5 mice labeled for Pax7 (green), Ki67 (red), and DAPI (nuclei, blue). Quantification of Ki67+/Pax7+ cells relative to all Pax7+ cells shown right (n = 3 animals per genotype; p-value shown). d Quantification of anti-Pax7 relative fluorescence intensity (RFI) on images as in (c). Data range is shown as violin plot with median and interquartile range, means of biological replicates are shown as dots (n = 3 animals per genotype; p-value shown). e Cytospin of FACS-isolated p14 MPs from control or Nf1Myf5 mice labeled for Pax7 (green), MyoD (red), and DAPI (nuclei, blue). Quantification of MyoD+/Pax7+ cells (right) (n = 3 animals per genotype; p-value shown). f In vitro differentiation of FACS-isolated p14 MPs from control or Nf1Myf5 mice after 2 d differentiation stained for Myosin (Mf20, green), MyoD (red) and DAPI (nuclei, blue). Quantification of MyoD+ nuclei within Mf20+ myotubes relative to all MyoD+ nuclei (right) (n = 3 animals per genotype; p-value shown). g Left: Representative images of single fibers isolated from 15-week EDL muscles stained for MyHC-2B (red) and DAPI (nuclei, blue). Boxed region shown as magnification. Right: Quantification of nuclei per myofiber and myonuclear domain (cell volume/number of nuclei); pL picoliter, (n = 3 animals per genotype; p-values shown). h Pax7+ cell quantification on sections of TA muscles of control or Nf1Myf5 mice at indicated time points (n = 3 animals per genotype; p-values shown). Data are mean ± SEM; P-value calculated by two-sided unpaired t-test. Source data are provided as a Source Data file.

Cytospun p14 MPs showed a relative decrease in MyoD+/Pax7+ cell numbers (Fig. 1e). Freshly isolated Nf1Myf5 p14 MPs plated in high density and immediately subjected to differentiation conditions showed a strong decrease in myotube formation compared to control MPs (Fig. 1f). No alterations in proliferation rate or Pax7/MyoD ratio was observed in haploinsufficient Myf5Cre;Nf1flox/+ p14 MPs (Supplementary Fig. 1d). In summary, Nf1Myf5 MPs were less proliferative than controls, and following activation in vitro showed impaired differentiation.

Decreased proliferation and differentiation led to reduced myonuclear accrual in Nf1Myf5 mice, as shown by reduced myonuclear numbers in single fast (MyHC-2B+) fibers from extensor digitorum longus (EDL) muscles of adult (15-week-old) mice (Fig. 1g). In addition, we found a reduced myonuclear domain (the amount of cytoplasm allocated to one myonucleus) in Nf1Myf5 fibers (Fig. 1g), suggesting no compensatory myonuclear domain growth34 occurred in this model, consistent with the metabolic growth defect of Nf1Myf5 muscle31. Pax7+ cell numbers of Nf1Myf5 muscles were normal at p7 but reduced in the following 2 weeks of postnatal life, and at p21, Pax7+ cell numbers in Nf1Myf5 muscles reduced to ~50% of control levels (Fig. 1h) and remained constant thereafter, as found at p84 (Fig. 1h), indicating a lasting decrease in MuSC numbers. We did not detect aberrant apoptosis in Nf1Myf5 muscle (Supplementary Fig. 1e).

We conclude that precocious postnatal cell cycle withdrawal of Pax7+ MPs and a differentiation blockade explain the decrease in MP numbers and myonuclear accretion, as well as the diminished adult MuSC pool in Nf1Myf5 mutants.

Nf1 is dispensable in muscle fibers

Myf5Cre targets myogenic progenitors (myoblasts), thus leading to an early recombination in the majority of the myogenic lineage35. We first analyzed Nf1 expression during myogenic differentiation. Nf1 gene expression decreased during myogenic differentiation of primary mouse myoblasts (Fig. 2a). In addition, in freshly isolated p7 MPs, Nf1 mRNA was less abundant compared to p21 MPs (Fig. 2b). Nf1 mRNA was present in p21 whole muscle tissue (Fig. 2b); however, adherent fibroblastic populations appeared as the major source of this expression (Fig. 2b). To disentangle the function of Nf1 within myofibers uncoupled from an earlier function in MPs, we inactivated Nf1 using Acta1Cre, which targets myofibers but not myoblasts36 via expression of Cre from a transgene driven by the human skeletal actin promoter. Acta1Cre specificity in limb muscle fibers, but not Pax7+ myogenic progenitors (MPs), was confirmed in Rosa26mTmG reporter mice37 (Supplementary Fig. 2a, b). RT-qPCR confirmed the efficiency of Nf1 deletion in p21 muscle tissue (Supplementary Fig. 2c). Surprisingly, Acta1Cre;Nf1flox/flox mice (Nf1Acta1) showed normal growth and were indistinguishable from littermates (Fig. 2c). The whole muscle cross-sectional area and fiber diameters of TA, EDL and Triceps muscles of Nf1Acta1 mice were equal to controls (Fig. 2d, e, Supplementary Fig. 2d). While Nf1Myf5 muscles showed a fiber type shift and altered metabolic gene expression profiles31, no changes in relative numbers of Type 1, Type 2A or Type 2B fibers were found in Nf1Acta1 TA or EDL muscles (Fig. 2f, g), and no alteration in a selection of metabolic genes that were deregulated in Nf1Myf5 muscle31 was found (Supplementary Fig. 2e).

Fig. 2: Nf1 is dispensable in myofibers.
figure 2

a Quantitative real-time PCR for Nf1 (left) and Myog (right) on primary mouse myoblasts cultured in proliferation medium or upon myogenic induction for indicated time (n = 3 animals per genotype; each dot represents the mean of three technical replicates from one biological replicate; p-values shown). b Quantitative real-time PCR for Nf1 in p7 or p21 MPs, p21 whole muscle tissue or p21 muscle-derived fast-adhering fibroblastic cells (FBs) (n = 3 animals per genotype; each dot represents the mean of three technical replicates from one biological replicate; p-values shown). c Whole-body appearance of control and Nf1Acta1 mice at p21. d Cross sections of lower hind limbs of control and Nf1Acta1 mice at p21 immunolabeled for Laminin (green), TA Tibialis anterior, EDL Extensor digitorum longus. Magnifications of indicated areas in TA muscles shown right. e Quantification of cross-sectional area (CSA; left) and myofiber Feret’s minimum diameter (right) of control and Nf1Acta1 TA and EDL muscles (n = 3 animals per genotype; p-values shown). f immunolabeling for Laminin (gray), MyHC-1 (red), MyHC-2A (purple), and MyHC-2B (green) on cross sections of TA (left) and EDL (right) muscles of control and Nf1Acta1 mice at p21. g Quantification of fiber types in p21 control and Nf1Acta1 TA and EDL muscles (n = 3 animals per genotype; p-values shown). Data are mean ± SEM; P-value calculated by two-sided unpaired t-test. Source data are provided as a Source Data file.

We conclude that Nf1 is downregulated during myogenic differentiation and is not required in mature muscle fibers, suggesting that myopathy of Nf1Myf5 animals was caused by aberrant progenitor programming.

Nf1-deficient MPs are shifted to premature quiescence

Cell cycle exit with a lack of differentiation and increased Pax7 expression indicates a shift of MPs to a quiescent phenotype. To further address this at the phenotype onset, we analyzed freshly FACS-isolated Nf1Myf5 and control p7 MPs by RNA-Seq. Two biological replicates, each consisting of cells pooled from 2 mice, were used for each genotype (Supplementary Fig. 3a, b, Supplementary Data 1). Gene set enrichment analysis (GSEA) showed an enrichment for NRAS Signaling in Nf1Myf5 MPs (Fig. 3a) in line with upregulated RAS pathway activity, and p-Erk immunolabeling intensity was increased in cytospun Nf1Myf5 MPs (Fig. 3b). Consistent with the reduced differentiation potential of Nf1Myf5 MPs, myogenesis-related GSEA terms were enriched in controls (Fig. 3c, Supplementary Fig. 3c). In contrast, GSEA terms associated with the ECM and basal lamina, both essential for MuSC quiescence11,38, were overrepresented in mutants (Supplementary Fig. 3d).

Fig. 3: Premature shift of Nf1Myf5 MPs to quiescence.
figure 3

a GSEA of control and Nf1Myf5 p7 MP RNA-Seq data for “NRAS Signaling”. b Labeling of cytospun control and Nf1Myf5 p7 MPs for Pax7 (green), phosphor-ERK1/2 (pErk1/2, red) and DAPI (nuclei, blue). Quantification of relative fluorescence intensity (RFI) for anti-pErk1/2 shown right. Data range is shown as violin plot with median and interquartile range, means of biological replicates are shown as dots (n = 3 animals per genotype; p-value shown). c GSEA of RNA-Seq data from control or Nf1Myf5 p7 MPs shows “MyoD targets” enriched in controls. d Volcano plot of transcriptome data from freshly FACS-isolated control or Nf1Myf5 p7 MPs. Individual transcripts deregulated in Nf1Myf5 MPs are indicated (blue: down; red: up). DE genes were identified by a log2 fold change over 2 or below 0.5 and a Benjamini-Hochberg adjusted p-value (padj) <0.01. Only genes with RPKM above 2 were considered. e Heatmap shows reduced MuSC activation–related gene expression in p7 Nf1Myf5 MPs. f Heatmap shows increased MuSC quiescence-related gene expression in p7 Nf1Myf5 MPs. g Heatmap shows increased expression of imprinted gene network genes in p7 Nf1Myf5 MPs. h RT-qPCR confirmation of differential expression of indicated genes in Nf1Myf5 p7 MPs (n = 3 animals per genotype; each dot represents the mean of three technical replicates from one biological replicate; p-values shown). i Reduced cell diameter in p7 Nf1Myf5 freshly sorted MPs. Representative images (left); quantification (right) (n = 3 animals per genotype; p-value shown). j Western blot shows reduced p70s6 kinase phosphorylation at Thr-389 in p7 Nf1Myf5 MPs (n = 3 animals per genotype; p-value shown). k Labeling of cytospun control and Nf1Myf5 p7 MPs for Pax7 (green), phosphor-Serine-235/236 S6 ribosomal protein (p-S6, red) and DAPI (nuclei, blue). Quantification of relative fluorescence intensity (RFI) for anti-p-S6 shown right. Data range is shown as violin plot with median and interquartile range, means of biological replicates are shown as dots (n = 3 animals per genotype; p-values shown). Data are mean ± SEM; P-value calculated by two-sided unpaired t-test. Source data are provided as a Source Data file.

DESeq2 analysis confirmed downregulation of myogenic differentiation genes and upregulation of Pax7 and other quiescence-related genes in Nf1Myf5 MPs (Fig. 3d). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of differentially regulated genes confirmed “ECM–receptor interaction” and “Focal adhesion” among the highest-enriched terms in genes upregulated in Nf1Myf5 MPs, while genes downregulated in Nf1Myf5 MPs showed enrichment for numerous terms related to cellular metabolism (Supplementary Fig. 3e).

We compiled MuSC activation and quiescence signatures, for which published transcriptome datasets11,39,40,41,42,43 were mined for genes commonly down- or upregulated. This yielded 142 activation-associated, and 136 quiescence-associated transcripts (gene lists shown in Supplementary Data 2). Only transcripts that showed an RPKM above 2 in our RNA-Seq data were further considered (117 for activation and 124 for quiescence). We filtered this list for genes showing a p(adj) value ≤ 0.05 in the DeSeq2 analysis. Of 117 activation-associated transcripts, 11 were up- and 47 were downregulated in Nf1Myf5 MPs, including Myog (Supplementary data 2). Of 124 quiescence-associated transcripts, 6 were down- and 78 were upregulated in Nf1Myf5 MPs (Supplementary data 2). This included Cdkn1b and Cdkn1c encoding cell cycle inhibitors p27 and p57 consistent with cell cycle exit, or Collagen type 5 subunit genes known to be involved in MuSC quiescence11. A selection of differentially expressed genes is shown in (Fig. 3e, f). Furthermore, the so-called imprinted gene network, known to be highly expressed in quiescent stem cells44,45, was upregulated in Nf1Myf5 MPs (Fig. 3g). We finally compared our RNA-Seq data to the dataset from Ryall et al39., comprising 2-month-old MuSCs freshly isolated comparable to our protocol, and MuSCs that were kept for 2 days in culture to reflect activated cells. Comparison of normalized read counts confirmed a shift of Nf1Myf5 p7 MPs transcriptome toward the signature of quiescent MuSCs (Supplementary Fig. 3f).

RT-qPCR confirmed upregulation of quiescence-related genes Pax7 and Spry1, and downregulation of activation–related genes Myod, Myog, and Myh3 and ATP2a1 (Fig. 3h). Neither Pax7, Calcr, Myog or Myh3 were deregulated in p14 MPs of Nf1-haploinsufficient Myf5Cre;Nf1flox/+ mutants (Supplementary Fig. 3g). Freshly isolated Nf1Myf5 MPs were smaller than control cells (Fig. 3i), a known feature of quiescent MuSCs46. In addition, phosphorylation of p70S6 kinase and of S6 ribosomal protein serine-235/236 as mammalian target of rapamycin complex 1 (mTORC1) signaling readouts, which is known to be induced upon MuSC activation46, was reduced in Nf1Myf5 p7 MPs (Fig. 3j, k; full blots for Fig. 3j shown in the source data file).

Results indicated that Nf1Myf5 p7 MPs show a phenotypic shift toward MuSC quiescence, including a transcriptomic signature, cell cycle exit, and mTORC1 activity.

Quiescence shift of Nf1Myf5 MPs reflects an altered epigenetic landscape

We next assessed three major chromatin marks: histone 3 lysine 4 trimethylation (H3K4me3), generally associated with active promoters; H3K27me3, associated with transcription inhibition;47 and DNA methylation (5’cytosine), globally associated with transcription inhibition48. Chromatin immunoprecipitation sequencing (ChIP-Seq) analysis of freshly isolated p7 MPs suggested slightly decreased H3K4me3 levels around the transcriptional start site (TSS) averaged across all genes between controls and Nf1Myf5 p7 MPs (Fig. 4a), and a global reduction of H3K27me3 levels in Nf1Myf5 p7 MPs (Fig. 4b). We, however, note that our analysis did not employ IgG or input analysis, or e.g. chromatin spike-in, thus quantitative assumptions at individual loci should be taken with caution. We confirmed globally reduced H3K27me3 levels by immunolabeling (Fig. 4c). GO analysis (biological process) of all 1157 genes with significantly decreased H3K27me3 levels in Nf1Myf5 p7 MPs identified by DiffBind (Gene list in Supplementary data 3) showed enrichment of terms associated to transcriptional regulation, cellular differentiation, but also “Notch signaling” (Fig. 4d). Intersecting genes with reduced H3K27me3 levels with genes upregulated in Nf1Myf5 p7 MPs showed only a low overlap of 248 genes (Fig. 4e; gene list in Supplementary data 4), suggesting that reduced H3K27me3 alone cannot explain gene deregulation in Nf1Myf5 MPs. GO analysis of the 248 genes mostly yielded general terms as “multicellular organism development” or “cell differentiation”, but also “transmembrane receptor protein tyrosine kinase signaling pathway” and “positive regulation of kinase activity” in line with increased RAS/MAPK signaling (Supplementary Fig. 4a). However, among these genes were quiescence-associated transcripts Pax7 and Pax3, and Notch pathway components Jag1 and Dll4 (Fig. 4e). ChIP-Seq tracks show decreased apparent H3K27me3 decoration at the Pax7 locus (Fig. 4f).

Fig. 4: Epigenetic changes associated with quiescence shift of Nf1Myf5 MPs.
figure 4

a, b Averaged normalized coverage in given region surrounding the transcriptional start site (TSS) across all genes for H3K4me3 and H3K27me3, derived from ChIP-Seq performed on control and Nf1Myf5 FACS-isolated p7 MPs. c Immunolabeling for Pax7 (green) and H3K27me3 (red) on cytospun control and Nf1Myf5 p7 MPs. Quantification of anti-H3K27me3 relative fluorescence intensity (RFI) is shown right. Data range is shown as violin plot with median and interquartile range, means of biological replicates are shown as dots (n = 3 animals per genotype; p-values shown). d GO analysis of all genes with significantly reduced H3M27me3. e Intersection of genes with reduced H3K27me3 and genes upregulated in p7 Nf1Myf5 MPs. f ChIP-Seq tracks for H3K4me3 and H3K27me3 at the Pax7 locus in control and Nf1Myf5 p7 MPs. g Heatmap depiction of DNA methylation–related gene expression in control and Nf1Myf5 p7 MPs. h RT-qPCR of Dnmt1 and Dnmt3a in control and Nf1Myf5 p7 MPs (n = 3 animals per genotype; p-values shown). i Enrichment analysis of Regions with increased or decreased methylation levels in Nf1Myf5 vs. control p7 MPs for different regions of interest (ROIs; promoter defined TSS as ±500 bases). Bar height corresponds to the odds ratio of DMIs in specific ROIs (ratios of numbers of DMRs in specific ROIs relative to the ratios of numbers of DMRs found in all regions analyzed). j GO analysis of regions with increased or decreased methylation levels in Nf1Myf5 vs. control p7 MPs. k MeDIP-Seq tracks from control and Nf1Myf5 p7 MPs at the Myl1 locus. l Log2(RPKM) values for Myl1 in p7 MPs transcriptome data (n = 2 animals per genotype; mean values and Padj.-value shown). m RT-qPCR of Myl1 expression in control and Nf1Myf5 p7 MPs (n = 3 animals per genotype; p-value shown). n MeDIP-Seq tracks from control and Nf1Myf5 p7 MPs at the Pfkfb1 locus. o RPKM values for Pfkfb1 in p7 MP transcriptome data (n = 2 animals per genotype; mean values and Padj.-value shown). p RT-qPCR of Pfkfb1 expression in control and Nf1Myf5 p7 MPs (n = 3 animals per genotype; p-value shown). Data are mean ± SEM; P-value calculated by two-sided unpaired t-test. Source data are provided as a Source Data file.

RNA-Seq analysis showed upregulation of DNA-demethylases Tet1-3 and all three relevant DNA methyltransferases, Dnmt1, Dnmt3a, and Dnmt3b (Fig. 4g). Dnmt1 and Dnmt3a were highly expressed in p7 MPs, but Dnmt3b had low expression levels based on RPKM values (Supplementary Data 1). RT-qPCR confirmed Dnmt1 and Dnmt3a upregulation in Nf1Myf5 MPs (Fig. 4h). Methylated DNA immunoprecipitation sequencing (MeDIP-Seq) analysis of freshly isolated p7 MPs showed differential methylation between controls and Nf1Myf5 predominantly at CpG islands (Fig. 4i). Enrichment analysis of differentially methylated regions (DMRs) showed that predominantly CpG islands gained methylation in Nf1Myf5 MPs (Fig. 4i).

Gene Ontology (GO) overrepresentation analysis of genes in proximity to CpG islands with increased methylation in Nf1Myf5 MPs showed terms associated with RNA synthesis and transcription, and cellular metabolism (Fig. 4j). Mining the proximity of differentially methylated regions for myogenesis-related genes showed a CpG island with gain of methylation in Nf1Myf5 MPs 3.5 kilobases (kb) upstream of Myl1 (Fig. 4k), which is part of the activation signature and is downregulated in Nf1Myf5 MPs in transcriptome data (Fig. 4l). RT-qPCR confirmed Myl1 downregulation (Fig. 4m). Analysis of metabolism-related genes showed gain of methylation of a CpG island 1.5 kb upstream of Pfkfb1 in Nf1Myf5 MPs, encoding phosphofructo-kinase-fructose-bisphosphatase 1 (Fig. 4n). Pfkfb1 was the highest expressed isoform of all Pfkfbs in juvenile MPs (Supplementary Data 1), and Pfkfb1 mRNA expression was strongly downregulated in Nf1Myf5 MPs in transcriptome data (Fig. 4o) and RT-qPCR (Fig. 4p). In addition, gain of methylation in Nf1Myf5 MPs overlapped the promoter of Ndufb11, encoding a subunit of mitochondrial complex I, and Idh3g, encoding a subunit of mitochondrial isocitrate dehydrogenase, which catalyzes the rate-limiting step of the tricarboxic acid cycle (TCA) (Supplementary Fig. 4b).

Therefore, epigenetic alterations in Nf1Myf5 MPs, at the genes we subjected to validation, are consistent with a shift toward quiescence. These alterations could contribute to impaired myogenic differentiation, and gain of methylation and transcriptional downregulation of metabolic genes indicate changes in cellular energy metabolism in Nf1Myf5 MPs.

Metabolic reprogramming of Nf1Myf5 juvenile MPs that is conserved in myofibers

Consistent with possible metabolic alterations, overrepresentation analysis with KEGG database pathways of the p7 MP transcriptome showed enrichment of “metabolic pathways,” “carbon metabolism,” “biosynthesis of amino acids,” and “glycolysis/gluconeogenesis” in genes downregulated in Nf1Myf5 MPs, and “protein digestion and absorption” in upregulated genes (Supplementary Fig. 3e). GSEA showed enrichment of “glycolysis/gluconeogenesis” and “oxidative phosphorylation” in controls (Fig. 5a). We identified global downregulation of genes of the glycolytic pathway and the pyruvate dehydrogenase complex, the citrate cycle, and the mitochondrial electron transport chain (ETC) in Nf1Myf5 p7 MPs (Fig. 5b–d). Expression of glycolytic genes Hk2 and Pfkfb1 was unaltered in p14 MPs of Nf1-haploinsufficient Myf5Cre;Nf1flox/+ mutants (Supplementary Fig. 5a).

Fig. 5: Metabolic reprogramming of Nf1Myf5 MPs.
figure 5

a GSEA of control and Nf1Myf5 p7 MP RNA-Seq data for “glycolysis – gluconeogenesis” and “oxidative phosphorylation.” bd Heatmaps show significant DEGs in Nf1Myf5 versus control MPs related to glycolysis and pyruvate dehydrogenase complex (b), TCA cycle components (c) and electron transport chain components (d). e SeahorseXF flux analysis of control and Nf1Myf5 p7 MPs; quantification of ECAR (n = 3 independent biological replicates from 3 animals per genotype; each dot represents the mean of five technical replicates from one biological replicate; p-values shown). f SeahorseXF flux analysis of control and Nf1Myf5 p7 MPs; quantification of OCR (n = 3 independent biological replicates from 3 animals per genotype; each dot represents the mean of five technical replicates from one biological replicate; p-values shown). g Venn diagram showing 130 commonly downregulated genes between Nf1Myf5 p7 MPs and Nf1Myf5 p21 muscle. GO analysis of commonly downregulated genes shown below. h Averaged normalized coverage for H4K16ac derived from ChIP-Seq on control and Nf1Myf5 p7 FACS-isolated MPs. TSS, transcription start site. i Immunolabeling for Pax7 (green) and H4K16ac (red) on FACS-isolated cytospun MPs from p7 control and Nf1Myf5 animals. Quantification of anti-H4K16ac relative fluorescence intensity (RFI) is shown right. Data range is shown as violin plot with median and interquartile range, means of biological replicates are shown as dots (n = 3 animals per genotype; p-value shown). j ChIP-Seq tracks for H4K16ac from control and Nf1Myf5 p7 MPs at the Myh3 locus. Data are mean ± SEM; P-value calculated by two-sided unpaired t-test. Source data are provided as a Source Data file.

Analyzing transcriptomic data by kinetic metabolic modeling49,50 confirmed a widespread metabolic shutdown in Nf1Myf5 MPs decreasing uptake and utilization of glucose, fatty acids, and branched-chain amino acids accompanied by decreased capacity for ATP production and oxygen consumption (Supplementary Fig. 5b).

Seahorse real-time metabolic flux analysis with freshly isolated p7 MPs to assess the metabolic consequence of this deregulation showed a strong reduction of extracellular acidification rate (ECAR) in Nf1Myf5 MPs, indicating severe glycolytic flux inhibition (Fig. 5e). In contrast, Nf1Myf5 p7 MPs showed only a moderate decrease in the basal oxygen consumption rate (OCR) below statistical significance in (Fig. 5f), indicating that oxidative phosphorylation capacity is still sufficient to maintain resting energy demand in MPs. Therefore, Nf1Myf5 MPs mainly use low-level oxidative metabolism as an energy source, consistent with a quiescent phenotype51.

Differentiated muscle in Nf1Myf5 animals showed fast fiber atrophy, shift of glycolytic to oxidative fiber types, and increased oxygen consumption31. This was confirmed by kinetic metabolic modeling analysis of p21 Nf1Myf5 muscle tissue transcriptome data31 suggesting impaired capacity for glucose utilization, but increased capacity for fatty acid utilization and unchanged capacity for utilization of branched-chain amino acids, concomitant with increased capacity for ATP production and increased oxygen consumption (Supplementary Fig. 5c). Direct comparison of Nf1Myf5 p7 MP transcriptome data to Nf1Myf5 p21 muscle data showed common downregulation of only 130 genes (Fig. 5j). GO overrepresentation analysis of this gene set showed enrichment of terms related to glucose/carbon metabolism and amino acid synthesis (Fig. 5g). GO analysis of 799 genes downregulated in p21 Nf1Myf5 muscle, but not in p7 MPs, did not yield any metabolism-related terms (Supplementary Fig. 5d), but rather terms as “Z-disc” possibly reflecting fiber atrophy31. Thus, transcriptome analysis and metabolic flux modeling of Nf1-deficient progenitors and differentiated muscle indicate a continuous deregulation of specifically carbohydrate metabolism. As Nf1Acta1 mice showed no significant defect in muscle size, fiber types, and expression of a panel of metabolic genes, this suggests that perturbed muscle fiber metabolism in Nf1Myf5 animals31 can be traced back to defects in juvenile MPs, indicating that that metabolic reprogramming in Nf1Myf5 juvenile MPs is transmitted to myofibers.

A metabolic switch from slow oxidative to forced glycolytic metabolism occurs during adult MuSC exit from quiescence to activation39. Concomitant NAD+ depletion inhibits Sirt1 function, which acts as a histone deacetylase mainly targeting H4K16. Increased H4K16 acetylation upon muscle-specific Sirt1 deletion induces expression of MuSC activation and myogenic differentiation-related genes39. H4K16ac ChIP-Seq showed a decrease in global levels of H4K16ac in p7 Nf1Myf5 MPs (Fig. 5h). Analysis of global H4K16ac levels in freshly isolated p7 MPs (Fig. 5i) and in p7 and 12-week-old muscle sections by immunofluorescence (Supplementary Fig. 6a, b) confirmed long-term decreased H4K16ac levels in Nf1Myf5 Pax7+ cells compared to controls.

Myh3, Bgn, Fst and Mylk2, which are upregulated in MuSC-specific Sirt1 conditional mice39, were downregulated in Nf1Myf5 MPs (Supplementary Fig. 6c) and showed reduced H4K16ac decoration at their gene bodies (Fig. 5j and Supplementary Fig. 6d). Thus, Nf1Myf5 MPs are driven toward quiescence and show metabolic reprogramming with severely inhibited glycolytic metabolism, and decreased H4K14ac and expression of myogenic differentiation-related genes.

Increased Notch signaling induced by a Mek/Erk/NOS cascade drives Nf1Myf5 juvenile MP quiescence shift

To analyze the mechanism of Nf1Myf5 MPs quiescence shift and reprogramming, we performed in vitro culture of juvenile primary myoblasts. RT-qPCR confirmed effective Nf1 knockdown (Supplementary Fig. 7a). Surprisingly, in vitro, Nf1Myf5 myoblasts did not reproduce the in vivo proliferative behavior, but showed enhanced proliferation (Supplementary Fig. 7b). Switching cells to a differentiation medium after 2 d of culture showed effective block of myogenic differentiation in Nf1Myf5 primary MPs (Supplementary Fig. 7b), as observed before for FACS-isolated MPs. Both increased proliferation and blocked differentiation fully depended on Mek/Erk signaling, as shown by inhibition with UO126 (Supplementary Fig. 7b). The discrepancy between in vivo and in vitro proliferation behavior of Nf1Myf5 MPs indicated that in vivo non-cell-autonomous microenvironmental factors override or divert Mek/Erk signaling in juvenile MPs, inhibiting cell proliferation.

Transcriptome analysis indicated upregulation of Delta/Notch signaling pathway components Dll1 and Notch1/3, and upregulation of Notch pathway targets Hes1, Hey1, Heyl, Calcr, Col5a1 and Col5a3 (Fig. 3f). GSEA showed enrichment of “Notch targets” in Nf1Myf5 MPs (Supplementary Fig. 7c). Upregulation of Notch1, Notch3, Hes1, and Hey1 in p7 Nf1Myf5 MPs was confirmed by RT-qPCR (Fig. 6a). Hes1, Hey1 and Notch1 expression was unaltered in p14 MPs of Nf1-haploinsufficient Myf5Cre;Nf1flox/+ mutants (Supplementary Fig. 7d).

Fig. 6: Increased Notch signaling drives Nf1Myf5 MPs to quiescence.
figure 6

a RT-qPCR of Notch pathway component and target genes in RNA extracted from freshly FACS-isolated control and Nf1Myf5 p7 MPs (n = 3 animals per genotype; each dot represents the mean of three technical replicates from one biological replicate; p-values shown). b RT-qPCR for Notch targets on FACS-isolated control or Nf1Myf5 p14 MPs cultured on Matrigel without coating or Jagged-1 coating for 48 h (n = 3 independent experiments from 3 animals per genotype; p-values shown). c FACS-isolated control or Nf1Myf5 p14 MPs cultured on Matrigel without coating or Jagged-1 coating for 48 h stained for Pax7 (green) and Ki67 (red). d Ki67+ cell quantification among Pax7+ cells on image data as in (c) (n = 3 animals per genotype; p-values shown). e Anti-Pax7 relative fluorescence intensity (RFI) quantification on image data as in (c). Data range is shown as violin plot with median and interquartile range, means of biological replicates are shown as dots (n = 3 animals per genotype; p-values shown). f Immunolabeling for MyoD (red) on FACS-isolated control or Nf1Myf5 p14 MPs cultured on Matrigel w/o coating or Jagged-1 coating for 48 h. g Quantification of MyoD+ cells / total cells on image data as in (f) (n = 3 animals per genotype; p-values shown). h Quantification of anti-MyoD relative fluorescence intensity (RFI) on image data as in (f). Data range is shown as violin plot with median and interquartile range, means of biological replicates are shown as dots (n = 3 animals per genotype; p-values shown). i GSEA on RNA-Seq data from control and Nf1Myf5 p7 MPs for “nitric oxide stimulates guanylate cyclase”. j RT-qPCR for Notch targets Pax7, Hes1 and Hey1 on FACS-isolated control or Nf1Myf5 p14 MPs. MPs were cultured on Matrigel without coating or with Jagged-1 coating for 48 h, with or without Mek inhibitor UO126 or pan-NOS inhibitor L-NAME. Bars show fold-changes of Nf1Myf5 MPs relative to control MPs, control MPs were set as 1 (n = 3 independent experiments from 3 animals per genotype; p-values shown). Data are mean ± SEM; P-value calculated by two-sided unpaired t-test. Source data are provided as a Source Data file.

We therefore tested whether Notch signaling represents the in vivo niche factor lacking in vitro by culturing MPs on cell culture plates coated with recombinant Jagged-1, which activates Notch signaling in myogenic cells52,53,54. To calibrate the system, we first cultured wild type primary myoblasts for 2 d in proliferation medium on uncoated control dishes, or dishes coated with different concentrations of Jagged-1 ligand. This showed induction of Hes1 and Hey1 expression already at 2.5 ng/μl, with maximal induction reached at 5 ng/μl (Supplementary Fig. 7e). Compared to control cells, Nf1Myf5 primary myoblasts showed increased induction of Notch target gene expression on recombinant Jagged-1–coated (5 ng/μl) dishes (Fig. 6b), indicating that Nf1Myf5 MPs are hypersensitive to Notch pathway activation upon external ligand stimulation.

Placement of Nf1Myf5 p14 FACS-isolated MPs cultured for 2 d in proliferation medium on uncoated dishes showed increased proliferation (Fig. 6c, d), as seen for primary myoblasts before (Supplementary Fig. 7b). We thus assessed, whether Jagged-1 treatment could reduce MP proliferation. Cultivation on Jagged-1-coated dishes (5 ng/μl) led to a reduced proliferation rate already in control p14 MPs, which was exacerbated in Nf1Myf5 MPs (Fig. 6c, d). This was confirmed by a dose-response titration using primary myoblasts showing that 2.5 ng/μl Jagged-1 reduced Nf1Myf5 myoblast proliferation rates to control levels, while higher concentrations reduced Nf1Myf5 myoblast proliferation rates below control levels (Supplementary Fig. 7f).

Both control and Nf1Myf5 MPs had low Pax7 expression after 2 d cultivation on control dishes (Fig. 6c, e) consistent with previous observations55. Jagged-1 maintained Pax7 expression in control and Nf1Myf5 MPs (Fig. 6c, e), with a relative increase in Pax7 abundance in Nf1Myf5 MPs (Fig. 6e). Conversely, Jagged-1 reduced the relative numbers of MyoD+ cells (Fig. 6f, g) and MyoD abundance (Fig. 6f, h) in control MPs, which was both exacerbated in Nf1Myf5 MPs (Fig. 6f–h), indicating that Jagged-1 induces a shift toward quiescence in juvenile control MPs, which is intensified in Nf1Myf5 MPs.

In Nf1-deficient oligodendrocytes, a Mek/Erk/nitric oxide synthase (NOS)/cyclic guanosine monophosphate (cGMP)/protein kinase G (PKG) pathway drives Notch pathway activation56. GSEA showed NO-cGMP signaling enriched in Nf1Myf5 MPs (Fig. 6i). Consistent with this, Nf1Myf5 MPs placed on Jagged-1-coated dishes and treated with Mek inhibitor UO126 or pan-NOS inhibitor L-NAME canceled the hyper-responsiveness to Jagged-1 (Fig. 6j), although we cannot formally exclude an effect of the inhibitors independent of Jagged-1 treatment.

We conclude that in juvenile MPs, a Ras/Mek/Erk/NOS pathway funnels into activation of the Notch pathway, inducing quiescence, which is exacerbated by lack on Nf1.

Inhibition of Notch signaling prevents quiescence shift of Nf1Myf5 juvenile MPs and ameliorates the whole-body phenotype of Nf1Myf5 mice

Notch signaling regulates cell metabolism in several systems57,58,59. We thus first analyzed whether Notch signaling is upstream of metabolic gene expression in juvenile MPs. RT-qPCR of selected glycolysis and mitochondrial gene expression levels in WT p14 MPs cultured on control dishes or in the presence of Jagged-1 indicated that the Notch pathway can inhibit energy metabolism-related gene expression in juvenile MPs (Fig. 7a). Jagged-1 stimulation especially affected glycolytic genes as Pfkfb1, Pfkfb3, Pfkm, Eno3, Ldha and Hk2, and it mildly affected mtCO1 and Ndufv1 as representatives of the TCA cycle and ETC (Fig. 7a), overlapping transcriptome data of Nf1Myf5 p7 MPs. This suggests that in juvenile MPs, activation of the Notch pathway contributes to metabolic reprogramming by inhibiting glycolytic gene expression.

Fig. 7: Rescue of Pax7 cell depletion, cell cycle exit, and metabolic reprogramming by Notch pathway inhibition.
figure 7

a RT-qPCR of selected glycolysis, TCA, and OXPHOS genes on FACS-isolated WT p14 MPs cultured on Matrigel without coating or Jagged-1 coating for 48 h. Pax7, Hes1, and Hey1 tested as internal controls (n = 4 animals per condition; each dot represents the mean of three technical replicates from one biological sample; p-values shown). b Schematic depiction of DAPT treatment of Nf1Myf5 animals. c Representative images of TA muscle sections of postnatal Nf1Myf5 mice treated with placebo or DAPT, stained for Pax7 (red), Ki67 (green), collagen IV (gray), and DAPI (blue; nuclei). d p21 Pax7+ cell quantification in Nf1Myf5 mice treated with placebo or DAPT (n = 4 animals per condition; p-values shown). e p21 Ki67+/Pax7+ cell quantification relative to Pax7+ cells in Nf1Myf5 mice treated with placebo or DAPT (n = 4 animals per condition; p-value shown). f RT-qPCR for glycolysis, TCA, and OXPHOS genes on muscle tissue from Nf1Myf5 mice treated with placebo or DAPT (n = 3 animals per condition; each dot represents the mean of three technical replicates from one biological sample; p-values shown). g Representative images of Laminin (green) immunolabeling on sections of Nf1Myf5 mice treated with placebo or DAPT shown left. Right: distribution of myofiber diameter in Nf1Myf5 mice treated with placebo or DAPT (n = 4 animals per condition). h Body weight of Nf1Myf5 mice treated with placebo or DAPT (n = 4 animals per condition; p-value shown). i Posterior subcutaneous white adipose tissue weight in Nf1Myf5 mice treated with placebo or DAPT (n = 4 animals per condition; p-value shown). Data are mean ± SEM; P-value calculated by two-sided unpaired t-test. Source data are provided as a Source Data file.

To test whether Notch signaling is needed for premature quiescence induction and long-term metabolic reprogramming in vivo, we treated Nf1Myf5 pups with 5 doses of 30 mg/kg of the Notch pathway inhibitor DAPT (or placebo control) from p6 to p18 (Fig. 7b). DAPT is an inhibitor of γ-Secretase, preventing Notch cleavage and thus signal transduction to the nucleus. To monitor Notch pathway inhibition in vivo, we measured the expression of two Notch pathway components and four Notch pathway targets using RT-qPCR and found reduced expression of all genes analyzed (Supplementary Fig. 8). In vivo DAPT treatment increased Pax7+ cell numbers (Fig. 7c, d) and Pax7+ cell proliferation (Fig. 7c, e) compared to placebo-treated Nf1Myf5 mice. Thus, premature quiescence induction was prevented by in vivo DAPT treatment. In vivo DAPT treatment increased expression of glycolytic genes in p21 Nf1Myf5 muscle, while Ndufv1 and mtCO1 expression stayed the same (Fig. 7f).

Nf1Myf5 mice show muscle atrophy and a whole-body catabolic phenotype with attrition of white adipose tissue because of increased muscular consumption of fatty acids31. In vivo DAPT treatment increased the apparent myofiber size (Fig. 7g) and the body weight (Fig. 7h) of Nf1Myf5 mice, indicating partial rescue of the muscle growth phenotype. In addition, in vivo DAPT treatment of Nf1Myf5 mice increased the white adipose tissue depot weight (Fig. 7i), indicating rescue of aberrant myofiber lipid metabolism.

We conclude that the Notch pathway is needed in vivo to induce premature quiescence in Nf1Myf5 MPs, and inhibition of the Notch pathway ameliorates metabolic reprogramming and improves the whole-body catabolic phenotype of Nf1Myf5 mice.