Fasting induces a highly resilient deep quiescent state in muscle stem cells via ketone body signaling

The effects of BHB are due to its role as an HDAC inhibitor and are mediated by p53

Short-term fasting is beneficial for the regeneration of multiple tissue types. However, the effects of fasting on muscle regeneration are largely unknown. Here, we report that fasting slows muscle repair both immediately after the conclusion of fasting as well as after multiple days of refeeding. We show that ketosis, either endogenously produced during fasting or a ketogenic diet or exogenously administered, promotes a deep quiescent state in muscle stem cells (MuSCs). Although deep quiescent MuSCs are less poised to activate, slowing muscle regeneration, they have markedly improved survival when facing sources of cellular stress. Furthermore, we show that ketone bodies, specifically β-hydroxybutyrate, directly promote MuSC deep quiescence via a nonmetabolic mechanism. We show that β-hydroxybutyrate functions as an HDAC inhibitor within MuSCs, leading to acetylation and activation of an HDAC1 target protein p53. Finally, we demonstrate that p53 activation contributes to the deep quiescence and enhanced resilience observed during fasting.

In this study, we show that fasting causes MuSCs to enter a state of deep quiescence (DQ) that is characterized by delayed activation and enhanced resilience to nutrient, cytotoxic, and proliferative stress. In this state, MuSCs are functionally and transcriptionally less committed to a myogenic program and more stem-like as assessed by muscle regeneration and transplantation assays as well as RNA sequencing (RNA-seq) analysis. Furthermore, increased MuSC resilience and delayed activation also results from feeding mice a ketogenic diet or injecting them with exogenous ketone bodies. Mechanistically, this deep quiescent state results from the inhibition of HDAC1 activity by increased levels of the primary circulating ketone body, β-hydroxybutyrate (BHB). We show, using pharmacological and genetic tools, that p53 activation downstream of HDAC1 inhibition is both necessary and sufficient to drive ketosis-induced DQ. Altogether, this report highlights the novel finding that a metabolite produced endogenously during DR elicits a protective state in a stem cell population.

Adult stem cells are critically important for maintaining tissue integrity and for regenerating damaged tissue after injury. MuSCs are indispensable for muscle tissue repair (). Long-term regenerative capacity of muscle is dependent on the ability of MuSCs to remain quiescent in the absence of injury and inability to maintain quiescence results in defective muscle repair (). In response to muscle injury, MuSCs enter the cell cycle, proliferate as myoblasts, and either self-renew to replenish the stem cell pool or fuse into nascent fibers to repair the injury (). In the absence of injury, quiescence is actively maintained and tightly regulated by a combination of signaling factors from the surrounding microenvironment and cell-intrinsic gene regulation, namely expression of cell cycle inhibitors and repression of cyclins, cyclin-dependent kinases, and checkpoint kinases (). For example, Notch signaling from the MuSC niche is known to be important in both maintenance of the quiescent state as well as self-renewal and reestablishment of quiescence following MuSC activation (). Quiescent MuSCs, like other quiescent stem cell populations, are characterized by stress resistance and low metabolic activity (). The effect of fasting on maintenance of this quiescent state, and on MuSC function in general, is unknown. The goal of this study was to comprehensively characterize the effect of fasting on muscle tissue repair and MuSC function.

Fasting affects a diverse population of stem and progenitor cells, frequently ameliorating various stem cell aging phenotypes (). For example, fasting increases the ability of intestinal stem cells (ISCs) from young and aged mice to form intestinal organoids (). The extent of this functional improvement positively correlates with the duration of the fast and is dependent on ISC fatty acid oxidation. Similarly, the age-associated decrease in expression of the muscle stem cell (MuSC)-specific transcription factor Pax7 can be restored to youthful levels after multiple rounds of a short-term calorically restrictive diet designed to mimic fasting (). This same fasting-mimicking diet (FMD) was also shown to rescue the decline in mesenchymal stem and progenitor cell number within bone marrow that is observed with aging (). Likewise, fasting or FMD promotes self-renewal of hematopoietic stem cells (HSCs), increases HSC number, and protects HSCs against cytotoxic agents (). In some instances, the affected stem cells may even rejuvenate their associated tissue in response to DR. For example, periodic fasting corrects the myeloid bias observed in aged hematopoiesis ().

Dietary restriction (DR) robustly improves healthspan and lifespan in multiple model systems (). Beneficial effects of DR have been observed at the organismal level, from simple eukaryotes to humans, and also at the single-cell level in adult stem cell populations (). These improvements are associated with cell and tissue functional maintenance and a delay or even prevention of age-related pathologies such as cognitive decline, inflammation, and cancer (). Commonly studied paradigms of DR are fasting and caloric restriction (CR). Fasting can lead cells and tissues to enter a protected state in which they become highly resistant to environmental stresses and toxicity (). For example, preoperative fasting lessens hepatic damage resulting from ischemia/reperfusion injury (). In some contexts, fasting or CR has been shown to accelerate tissue regeneration. For example, intestinal epithelial and hepatic tissue regeneration after acute damage have been reported to improve after intense CR ().

The effect of retarded growth upon the length of life span and upon the ultimate body size. 1935.

The effect of retarded growth upon the length of life span and upon the ultimate body size.

To test for the sufficiency of p53 activity to promote DQ, we injected mice intraperitoneally with vehicle or with nutlin-3, a p53 activator (). We found that activating p53 in vivo could indeed recapitulate the DQ phenotypes of MuSCs ( Figures 6 H–6J and S7 G–S7H) that we had seen previously with both ketosis as well as HDAC inhibition. Additionally, consistent with what we found with BHB and givinostat treatments, treating isolated MuSCs with nutlin-3 for 48 h delayed cell quiescence exit and promoted resilience in cultured MuSCs ( Figures S7 I and S7J). Activity of p53 can be regulated through post-translational acetylation of various lysine residues in the C terminus of the protein, including lysine 379 (382 in humans) (). Recent studies have found that the stabilizing effects of acetylation can be mimicked by mutating various C-terminal lysine residues to glutamine (p53) (). To directly demonstrate the ability of p53 acetylation to mimic the effects of ketosis, we tested whether MuSCs isolated from these p53mutant mice exhibit the hallmarks of DQ. Indeed, p53MuSCs present evidence of elevated p53 activity as well as reduced cell size and mitochondrial content at time of isolation, delayed S phase entry, and enhanced survival ( Figures 7 A–7E ). Collectively, our results highlight both the necessity and sufficiency of p53 activity in promoting MuSC DQ.

(B–E) Quantification of (B) relative cell size (forward scatter), (C) MitoTracker signal, (D) percent EdU incorporation, and (E) percent survival of MuSCs from WT and p53mutant mice (n = 5). Transgenic p53mice are heterozygous for acetylation mimicking mutations ().

(A) Quantification of p21 transcript levels by qPCR of mRNA extracted from MuSCs of WT and p53 KQ mutant mice. p21 transcript levels were normalized to GAPDH transcript levels (n = 4).

To test for a potential role of p53 in mediating any of the phenotypes of KIDQ, we used a Pax7CreER driver to specifically ablate p53 in MuSCs as previously described (). Two weeks after the completion of tamoxifen dosing, MuSC-specific p53 KO mice and wild-type (WT) controls were fasted or ad libitum fed for 60 h, at which point their MuSCs were isolated for analysis. We found that key hallmarks of DQ, including reduced cell size, decreased mitochondrial content, and delayed S phase entry, were partially abrogated in response to the ablation of p53 in MuSCs Figures S7 B–S7F). In order to determine if p53 were mediating the direct effects of BHB on MuSCs, we isolated p53-ablated MuSCs and WT controls and we treated the cells ex vivo with BHB. We found that that p53 KO MuSCs were largely insensitive to BHB-induced changes in cell cycle entry and survival ( Figures 6 F and 6G)

We next wanted to determine exactly which of the HDAC1 targets, when hyperacetylated in response to HDAC inhibition, might be driving the DQ phenotype in MuSCs. To address this question, we performed a pan acetyl-lysine western blot to assess for any pattern of increased protein acetylation in response to fasting or ketone body treatment. Intriguingly given that p53 is a canonical HDAC1 target (), we found a protein of approximately 53 kD that was most prominently hyperacetylated in response to ketosis ( Figure 6 A ). Mouse p53 can be acetylated at various residues, including K379 (K382 in human) (), and K379 acetylation is known to promote DNA binding activity of p53, induce apoptosis in cancer cells, and promote survival in certain noncancer cells (). Using an antibody specific to acetyl-p53 K379, we confirmed hyperacetylation of p53 by both western blot of FACS-purified MuSCs and immunofluorescence analysis of MuSCs on isolated myofibers in response to fasting, the ketogenic diet, and ketone body injections ( Figures 6 B–6D and S7 A). These findings are consistent with recent work suggesting that liver p53 is also hyperacetylated in response to both short-term fasting and ketogenic diet treatment (). In addition, we identified a robust p53 transcriptional signature in MuSCs by gene set enrichment analysis in response to ketosis ( Figure 6 E). Recent work from our lab has highlighted the unique and unexpected role of p53 in promoting the survival of MuSCs (). We therefore surmised that acetylation and activation of p53 via HDAC1 inhibition might be one of the pathways contributing to the DQ state of enhanced MuSC resilience in response to ketosis and givinostat treatment.

(H–J) Quantification of (H) relative cell size (based on forward scatter from FACS) in freshly isolated MuSCs, (I) relative MitoTracker intensity in freshly isolated MuSCs, and (J) percent EdU incorporation at 48 h in MuSCs derived from mice that were fasted for 60 h or injected with givinostat (10 mg/kg daily) or with the p53 activator nutlin-3 (20 mg/kg daily) for 1 week. Values from each treatment condition were normalized to respective ad libitum or vehicle controls (n = 4).

(G) Quantification of cell survival in MuSCs isolated from p53 cKO or WT mice treated with BHB in culture. Cells were treated with either 10 mM BHB or vehicle for 96 h and subsequently stained for annexin V and PI. Percent cell survival was quantified as the percentage of cells that stained negative for both annexin V and PI.

(F) Quantification of EdU incorporation in MuSCs derived from p53 cKO or WT mice treated with BHB in culture. Cells were treated with either 10 mM BHB or vehicle for 48 h in the continuous presence of EdU (n = 4).

(D) Western blot analysis (top) and quantification (bottom) of p53 acetylation (K379) in MuSCs from mice injected with vehicle or BHB. Total p53 was used as a loading control for Ac-p53.

(C) Immunofluorescence staining (left) of Pax7 and acetyl-p53 on single isolated myofibers isolated from fasted and ad libitum-fed control mice. Nuclei were stained with DAPI. Arrows indicate MuSCs. Quantification (right) of percent of Pax7 positive MuSCs that stained positive for acetyl-p53.

(B) Representative western blot (top) showing acetyl-p53 levels in MuSCs isolated from fasted or ad libitum-fed mice. Total H3 protein was used as a loading control. Quantification (bottom) of relative acetyl-p53 levels normalized first to total H3 and then to ad libitum-fed controls (n = 3).

(A) Representative western blot showing pan acetyl-lysine levels in freshly isolated MuSCs from mice treated with BHB or vehicle for 1 day. Bands at 53, 17, and 15 kD were identified as exhibiting increased acetylation upon in vivo BHB treatment. H3 was used as a control for protein loading.

In order to narrow down exactly which class of HDACs, when inhibited, is sufficient to promote MuSC DQ, we performed an HDAC inhibitor screen using a chemical library of well-characterized HDAC inhibitors. Our screen revealed that HDAC class I (and very likely HDAC1 within class I) was highly enriched among inhibitor targets that were able to promote DQ Figures S6 A–S6D). To confirm the findings of our screen in vivo, we injected mice with inhibitors from the screen that spanned the different classes of HDAC inhibitors. Consistent with the findings from our screen, we found that only HDAC class I inhibitors, when injected intraperitoneally, were able to elicit the phenotypic hallmarks of DQ in MuSCs ( Figures S6 E and S6F).

The best characterized nonmetabolic function of BHB is as an endogenous inhibitor of histone deacetylases (HDACs) (). To probe whether or not BHB might be acting as an HDAC inhibitor in MuSCs, we treated isolated MuSCs with BHB and measured acetylation of histone H3 (a well-characterized HDAC substrate) at lysine 9. We found that BHB administration resulted in significant elevations in H3K9 acetylation, consistent with its role as a potential HDAC inhibitor ( Figure 5 E). Similarly, fasting-induced ketosis resulted in elevated H3K9 as well as H3K14 acetylation ( Figures S4 G and S4H). If the effects of BHB on promoting MuSC DQ were indeed the result of HDAC inhibition, we surmised that HDAC inhibition might phenocopy the effects of fasting, the ketogenic diet, and exogenous ketone body administration on the promotion of DQ in MuSCs. To address this, we injected mice intraperitoneally with an HDAC inhibitor in clinical use, givinostat, or with a vehicle control for 1 week. We found that MuSCs isolated from mice injected with givinostat displayed all of the DQ hallmarks seen in MuSCs from ketotic mice ( Figures 5 F, 5G, S5 A, and S5B). Givinostat treatment had no effect on the body weight or ketone levels of the mice ( Figures S5 C and S5D). Consistent with the hallmarks of the DQ phenotype, MuSCs isolated from givinostat-injected mice exhibited a remarkable ability to survive in culture and to resist the stress of acute nutrient deprivation compared with control MuSCs ( Figures 5 H, S5 E, and S5F). Notably, as we observed with fasting, the effects of givinostat on MuSCs perdured multiple days after treatment was discontinued ( Figures S5 G–S5J). To examine whether this phenotype was the result of a direct effect of givinostat on the MuSCs themselves, we isolated MuSCs and treated them in culture with givinostat for 48 h. Similar to what we observed with ex vivo BHB treatment, we found that treatment of isolated MuSCs with givinostat ex vivo resulted in a dramatic enhancement of MuSC survival along with a substantial delay in the entry into S phase ( Figure 5 I).

Since nonmetabolic signaling functions have been attributed to both AcAc and BHB (), we next wanted to distinguish which of the two metabolites is the primary driver of DQ during ketosis. Addressing this question by simply treating MuSCs with BHB and AcAc separately is confounded by the fact that the two molecules readily interconvert through the enzymatic activity of β-hydroxybutyrate dehydrogenase (BDH1). However, BDH1 is known to specifically convert the endogenously produced R enantiomer of BHB into AcAc but does not metabolize the S-BHB (). We took advantage of this limitation in the enzymatic activity of BDH1 by treating MuSCs with the nonmetabolizable S-enantiomer to determine whether BHB could promote DQ absent its conversion into AcAc. Indeed, we found that treatment of MuSCs with S-BHB was sufficient to delay quiescence exit, as measured by EdU incorporation ( Figure 5 C), and to increase survival rate of cultured MuSCs by >20% ( Figure 5 D). These data support our previous finding that ketosis promotes MuSC DQ independent of ketone body metabolism by OXCT1 and establish that BHB alone is sufficient to promote MuSC DQ, in absence of its conversion to AcAc.

We hypothesized that the effect of ketone bodies in MuSCs was due to changes in the metabolic activity of these quiescent cells. Previous transcriptomic analyses from our lab and others have revealed a host of unique metabolic changes that define the quiescent state (). To probe the metabolism of quiescence at the protein level, we employed an untargeted shotgun proteomics analysis of quiescent MuSCs, comparing the data with those from activated MuSCs. Screening for quiescence-enriched metabolic enzymes, we found that OXCT1, the rate-limiting enzyme in ketone body metabolism (), was highly expressed in the quiescent state ( Figures 5 A and S4 A). This suggested to us that ketone bodies may be promoting DQ in MuSCs via their conversion to acetyl CoA. To address this, we generated a conditional OXCT1 knockout (KO) mouse (OXCT1;Pax7;ROSA26) to specifically ablate OXCT1 in MuSCs following tamoxifen injection ( Figures S4 B and S4C) (). Two weeks after the completion of tamoxifen dosing, we injected the control and OXCT1mice with ketone bodies 3 times per day for 7 days. To our surprise, we found that there was no significant difference in the induction of DQ in MuSCs in response to exogenous ketone body treatment between the two strains ( Figures 5 B and S4 D–S4F). This finding suggested to us that ketone bodies are likely acting nonmetabolically to elicit DQ in these cells.

(I) Representative immunofluorescence images (left) of MuSCs treated with givinostat or vehicle in culture for 48 h. Cells were grown in the continuous presence of EdU, and after 48 h, they were stained for EdU (488 channel) and TUNEL (594 channel). Nuclei were stained with DAPI. Quantification of EdU incorporation (middle) and cell death (right).

(H) Quantification of cell death in MuSCs from givinostat- or vehicle-treated mice. Cells were grown in culture for 48 h and subsequently stained for annexin V and propidium iodide. Cell death was quantified by the percentage of cells that were positive for annexin V and propidium iodide.

(F) Quantification of EdU incorporation in MuSCs isolated from mice treated with givinostat (10 mg/kg i.p. daily) or vehicle for 1 week. Following isolation, MuSCs were maintained in culture for 48 h in the continuous presence of EdU (n = 4).

(E) Representative western blot analysis (left) of histone 3 acetylation at lysine 9 in MuSCs in response to 10 mM BHB treatment for 24 h in culture. Total histone H3 is used to control for protein loading (n = 3 per group). Quantification (right) of relative histone H3 acetylation at lysine 9 normalized first to total H3 and then to vehicle control.

(D) Survival rates of MuSCs after 96 h in culture with growth media that contained vehicle or 5 mM of the indicated stereoisomer of BHB.

(C) Percent EdU incorporation of MuSCs after 48 h in culture with growth media that contained vehicle or 2.5 mM of the indicated stereoisomer of BHB.

(B) Quantification of EdU incorporation in MuSCs derived from OXCT1 cKO or WT mice that had been treated with either ketone bodies or vehicle for 1 week. Cells were maintained in culture for 48 h in the continuous presence of EdU (n = 4).

(A) Shotgun proteomics volcano plot (left) comparing relative spectral count values between quiescent and activated MuSCs. The y axis represents the p value and the x axis represents the ratio of protein spectral counts of quiescent divided by activated MuSCs. Statistically significant protein changes reside above the red line demarcating a p value of 0.05. OXCT1 is identified and is shown in a heatmap (right) alongside other quiescence-enriched metabolic enzymes. Quiescence-enriched proteins in the heatmap are shown as relative fold change normalized to activated MuSCs (n = 2; 20 pooled mice per replicate).

We next wanted to know if KIDQ was a result of the direct effect of the ketone bodies on the MuSCs themselves. Indeed, treatment of isolated MuSCs with BHB (the primary ketone body produced during fasting) produced a striking improvement in cell survival and a concomitant decrease in the rate of S phase entry in a dosage-dependent manner, consistent with the DQ phenotype observed in MuSCs isolated from ketotic mice ( Figure 4 C). Similarly, directly treating isolated human MuSCs with BHB also delayed quiescence exit and reduced cell death in culture ( Figures S3 L and S3M). To determine the functional relevance of this BHB-driven cell-intrinsic survival enhancement to in vivo potency of murine MuSCs, we performed a stem cell transplantation assay. RFP-labeled MuSCs were purified and maintained for 60 h in medium supplemented with BHB or control medium and then transplanted into injured, irradiated muscles of recipient mice. Fourteen days later, we examined host muscles for the presence of RFP-labeled muscle fibers. Compared with control MuSCs, BHB-treated MuSCs contributed much more robustly to the formation of new muscle fibers ( Figure 4 D). Our finding that ketosis causes reduced expression of genes involved in proliferation and myogenic differentiation ( Figures 3 B–3D and S3 C) is consistent with the interpretation that increased myofiber formation results from improved MuSC survival during and after transplant, rather than increased proliferation of transplanted cells. To directly test whether BHB treatment leads to enhanced MuSC resilience, we assessed whether ketosis enhances stem cell protection in vivo by exposing vehicle- or BHB-treated mice to genotoxic irradiation. MuSCs from irradiated muscles were then isolated, cultured, and assayed for viability. MuSCs from animals treated with BHB showed significantly reduced death following in vivo X-ray exposure ( Figure 4 E). Taken together, these data suggest that the ketone body BHB produces a highly resilient DQ state in MuSCs.

We have recently shown that old MuSCs have an impaired ability to survive the transition between quiescence and activation (). Given the resistance of DQ MuSCs to activation-induced death, we wanted to know if we could rescue the known survival impairments of old MuSCs by promoting KIDQ. In order to address this, we injected a cohort of 24-month-old mice with either vehicle or ketone bodies for 1 week, at which point we sacrificed the mice and isolated MuSCs from the hindlimbs. After plating the cells for 48 h in culture, we again measured cell death. Quite strikingly, we found that ketone body treatment could indeed rescue the survival defects that are characteristic of old MuSC activation ( Figure 4 B).

In order to test for the propensity of DQ MuSCs to exit quiescence and enter the cell cycle, we plated MuSCs from ketotic and control mice in culture with the intention of studying in more detail their kinetics of activation. However, surprisingly, we observed that a greater number of MuSCs from ketotic mice were present in culture after 48 h of activation compared with their control counterparts ( Figure S3 G). Having already observed that DQ MuSCs take longer to break quiescence and enter the cell cycle, we surmised that the most likely explanation for this increase in cell number was not due to faster expansion but rather an enhanced ability to survive the stress of the activation process (). Consistent with this hypothesis, we found that DQ MuSCs exhibited a dramatic decrease in cell death in culture ( Figures S3 H–S3J), suggesting that the state of DQ is one of a marked increase in resilience. We next asked whether these DQ MuSCs could also resist other forms of stress as well. Indeed, we found that MuSCs isolated from ketone body-injected mice were much more resistant to oxidative stress as well as that of acute nutrient deprivation ( Figures 4 A and S3 K). Together, these data highlight another key property of DQ cells and that is one of increased resilience and general stress resistance.

(E) Workflow (left) and quantification (right) of in vivo MuSC resilience in which mice were treated with BHB and then exposed to Xray irradiation to the hindlimbs. MuSCs from irradiated muscles were isolated and analyzed for viability in culture (n = 6–8 per group).

(D) Representative fluorescence images (left) of RFP-expressing MuSCs treated in culture with 10 mM BHB or vehicle for 60 h and subsequently transplanted into preinjured and preirradiated recipient TA muscle (n = 6–7 per group). Cryosections harvested 2 weeks after transplantation were analyzed for RFP expression, and the number of RFP positive fibers was quantified (right).

(C) Representative brightfield images (top) of MuSCs following 96 h of growth in the presence of vehicle control or 10 mM BHB. Quantification of percent EdU incorporation in MuSCs after 48 h (bottom left; n = 4 or 5) and percent MuSC survival after 96 h (bottom right; n = 5) in growth media containing racemic (R/S) BHB at different doses.

(B) Representative FACS plots (left) of MuSCs isolated from young and old mice following 1 week of in vivo treatment with ketone bodies or vehicle. After isolation, cells were grown for 48 h in culture and subsequently stained with annexin V to measure cell death (n = 4). Quantification (right) of cell death by measuring the percentage of cells that are annexin V positive.

(A) Representative FACS plots (left) of freshly isolated MuSCs from ketone body- or vehicle-treated mice 2 h following H 2 O 2 challenge. MuSCs were stained with annexin V (x axis) and DAPI (y axis) to measure cell death (n = 4). Quantification (right) of cell death as the percentage of cells that are both propidium iodide and annexin V positive.

Previous work has shown that the Pax7subpopulation of MuSCs has a higher capacity to seed the MuSC niche than the Pax7population (). To assess the engraftment potential of DQ MuSCs compared with control MuSCs, we tested whether DQ cells could outcompete their control counterparts in a competitive transplantation assay. In this paradigm, we cotransplanted distinctly labeled DQ MuSCs and control MuSCs into the same injured recipient muscle. After 28 days, we isolated by FACS those transplanted DQ and control MuSCs that had engrafted. We found that DQ MuSCs consistently outcompeted their control counterparts and were present in higher numbers in the recipient muscle ( Figure S3 F). These results suggest that the state of MuSC DQ is also characterized by an enhanced capacity for long-term engraftment.

In addition to cell-cycle changes, our transcriptomic analysis revealed that freshly isolated MuSCs from ketone-injected mice exhibited lower expression of myogenic genes such as MYF5 and MYOD1 ( Figures 3 D and S3 C). They also displayed increased expression of CD34, a MuSC stemness marker () ( Figure 3 D). Previous work from our lab and others has shown that quiescent MuSCs contain MYOD1 transcripts (), perhaps allowing quiescent MuSCs to be poised for myogenesis upon activation. Because DQ MuSCs contain less MYOD1 and higher levels of CD34, this may be indicative of MuSCs that are more stem-like and less poised for myogenic commitment. Although total Pax7 transcript levels were not significantly altered in the freshly isolated MuSCs from ketone-treated mice ( Figure S3 D), we found that freshly isolated DQ MuSCs had a larger percentage of cells with Pax7promoter activity ( Figure 3 E). Our work is consistent with previous findings that have found a strong correlation between Pax7 promoter activity and delayed exit from quiescence (). Additionally, we found that Pax7 activity persisted much longer during the course of activation of MuSCs from ketone-treated mice ( Figures 3 F and S3 E), consistent with a state of DQ.

To probe the molecular changes underlying this ketone-induced DQ (KIDQ), we sequenced the transcriptomes of freshly isolated MuSCs from mice that were injected with ketone bodies or vehicle control ( Figure 3 A ). Consistent with our evidence that MuSCs from ketone body-treated mice exist in a deeper state of quiescence, gene ontology analysis revealed significant and robust changes in cell cycle genes, including downregulation of well-established proliferation genes such as the E2F family members, CDK2, and multiple members of the cyclin family of proteins ( Figures 3 B and 3C). We also observed an upregulation of genes involved in the inhibition of cell proliferation and growth, including CDKN1A (p21) and CDKN2D (p19) ( Figure 3 C). This signature is consistent with the transcriptional signature of DQ recently described in fibroblasts in response to altered lysosomal activity in vitro (). Notably, we found a significantly increased ratio of p21 to cyclin D1 in MuSCs from ketone body-injected mice ( Figure S3 A), which is consistent with a higher E2F switching threshold that has previously been shown to be a critical driver of quiescence depth (). No change was observed in the senescence marker p16 in response to ketone bodies, consistent with the absence of any features of senescence in MuSCs in response to fasting, ketogenic diet, or ketone body injections ( Figure S3 B).

(F) Quantification of the percentage of Pax7 hi promoter activity in MuSCs isolated from Tg(Pax7-ZsGreen) mice injected with ketone bodies or vehicle. Cells were plated for either 0 or 48 h, at which point Pax7 promoter activity was assessed by FACS.

(E) Representative FACS plots of freshly isolated MuSCs isolated from Tg(Pax7-ZsGreen) mice injected intraperitoneally with ketone bodies or vehicle for 1 week (n = 3). The y axis (FITC) shows Pax7 promoter activity and the x axis (FSC) shows cell size. Quantification (right) of the percentage of Pax7 hi versus Pax7 lo promoter activity in freshly isolated MuSCs from Tg(Pax7-ZsGreen) mice injected with ketone bodies or vehicle (n = 3).

(A) RNA-seq volcano plot comparing differential expression of genes between freshly isolated MuSCs from mice treated with ketone bodies or vehicle. The y axis is the FDR-normalized p value, and the x axis is the relative fold change of MuSC genes from ketone-treated divided by vehicle-treated mice. Statistically significant changes reside above the red line demarcating an FDR corrected p value of 0.05 (n = 4).

Recent work has suggested that many of the beneficial effects of fasting (especially longer term fasting and CR) may be the result of the ketosis that accompanies the period of fasting (). In order to assess whether ketosis, absence of the stark energy imbalance seen with fasting, might be responsible for promoting this state of DQ, we fed mice an ad libitum ketogenic diet for 3 weeks to induce endogenous ketosis ( Figure S2 A) and asked whether that might also induce a similar state of DQ in the MuSCs. We found that the MuSCs isolated from mice on the ketogenic diet exhibited the same hallmarks of the DQ state that we observed with fasting ( Figures S2 B–S2E). Similarly, we detected a significant delay in muscle regeneration in mice that were fed a ketogenic diet ( Figures S2 F and S2G). Given that the ketogenic diet and fasting both involve drastic hormonal changes that accompany carbohydrate restriction, we next asked whether supplementation of an ad libitum chow diet with exogenous ketone bodies could also promote MuSC DQ. Mice were injected intraperitoneally with the two primary ketone bodies (BHB and acetoacetate) three times per day for 7 days to mimic the endogenous ketosis brought about by fasting and the ketogenic diet ( Figures S2 H–S2K). Unlike fasting or the ketogenic diet, exogenous ketone body supplementation resulted in no change in blood glucose levels or body weight ( Figures S2 L and S2M). Similar to fasting and the ketogenic diet, exogenous ketone bodies also induced all of the hallmarks of DQ in MuSCs, including smaller size, less mitochondrial content, less oxygen consumption, less RNA content, and a slower rate of S phase entry ( Figures 2 D–2H and S2 N). We also found that the MuSCs from ketone-treated mice had a decreased propensity to break quiescence and enter S phase under homeostatic conditions in vivo ( Figure 2 I). Collectively, these data suggest that ketosis itself, and ketone bodies in particular, might be a primary driver of MuSC DQ.

Because MuSCs are critically important for muscle regeneration after injury (), we next wanted to examine how short-term fasting might be affecting the function of the MuSCs themselves. After fasting a cohort of mice, we isolated quiescent MuSCs from the hindlimbs by fluorescence-activated cell sorting (FACS) purification as previously described (). We found that MuSCs from fasted mice were smaller than their control counterparts at the time of isolation ( Figure 2 A ). Additionally, MuSCs from fasted mice exhibited a significant reduction in mitochondrial content, RNA content, and basal oxygen consumption compared with control MuSCs ( Figures 2 B, S1 A, and S1B). These cells were also significantly delayed in their time to enter S phase both after ex vivo activation as well as in response to in vivo injury ( Figures 2 C and S1 C). Therefore, MuSCs from fasted mice exhibit properties that are characteristic of cells in a state of DQ (). In agreement with our muscle regeneration data, we found that this state of DQ persists up to 2 days after refeeding ( Figures S1 D–S1F), despite the return to baseline of body weight ( Figure 1 G). These data suggest that DQ represents a stem cell state that can perdure for multiple days after the conclusion of fasting and that this perdurance may explain the delay in muscle regeneration that we observed after fasting, even following 72 h of refeeding.

(I) Cell-cycle entry based on EdU incorporation of MuSCs from mice treated with ketone bodies or vehicle. Mice were injected with 50 mg/kg EdU once daily for 2 weeks. Freshly isolated MuSCs were immediately fixed and stained for EdU.

(H) Quantification of EdU incorporation in MuSCs isolated from ketone body- or vehicle-treated mice and maintained in culture for 48 h in the continuous presence of EdU (n = 4).

(E) Representative confocal microscopy images of freshly isolated MuSCs from ketone body- or vehicle-treated mice showing total mitochondrial staining with MitoTracker Deep Red as well as mitochondrially produced reactive oxygen species with MitoSox Green. Nuclei were stained with DAPI.

(D) Quantification of the cell size (based on forward scatter in FACS plots) of freshly isolated MuSCs from ketone body- or vehicle-treated mice. Mice were injected intraperitoneally with ketone bodies (200 mg/kg BHB and 200 mg/kg acetoacetate) or vehicle three times per day for 1 week (n = 4).

(C) Quantification of EdU incorporation in MuSCs isolated from fasted (60 h) and ad libitum-fed control mice after 48 h in culture (n = 4).

(A) Representative brightfield images (left) and representative FACS plots (middle) of MuSCs harvested from fasted (60 h) and ad libitum-fed mice. Right panel shows quantification of cell size based on forward scatter in FACS plots (n = 4).

To comprehensively characterize the effects of fasting on muscle regeneration, we injured the tibialis anterior (TA) muscle of the lower hindlimb in mice fasted for 0, 1, 2, or 2.5 days ( Figure 1 A ). After 7 days of recovery, we isolated the muscles and examined regeneration histologically. We found that recovery, as measured by regenerating myofiber cross-sectional area (CSA), progressively declined as a function of duration of fasting prior to injury ( Figures 1 B and 1C). To explore the reversibility of this regenerative delay, we measured the extent of muscle repair in mice that had been fasted for 2.5 days and subsequently refed for 1, 2, 3, or 7 days prior to injury ( Figure 1 D). Intriguingly, we found that an impairment of regeneration persisted up to 3 days after refeeding, despite a complete return of body weight ( Figures 1 E–1G). Importantly, 1 week of refeeding was able to restore muscle regeneration back to baseline ( Figure 1 F). These findings indicate that fasting causes a transient state of impaired regenerative activity that persists days after refeeding.

(G) Percent change in body mass (left) and change in blood BHB concentration (right) of mice during fasting and after refeeding.

(F) Quantification of regenerating muscle fiber cross-sectional area from mice fasted for 2.5 days then refed for 1, 2, 3, or 7 days before injury as described in (D).

(E) Representative histology of regenerating muscle from mice that were fasted for 2.5 days and subsequently refed for 3 or 7 days before injury. TA muscles were harvested 7 days after injury, sectioned, and stained for laminin. Nuclei were stained with DAPI.

(D) Diagram of time course that depicts fasting and refeeding durations before BaCl 2 injury to the TA muscle, followed by 7 days of recovery.

(C) Quantification of regenerating muscle fiber cross-sectional area from mice fed ad libitum or fasted for 1, 2, or 2.5 days before injury as described in (A).

(B) Representative immunostaining of regenerating muscle from mice fed ad libitum or fasted for 2.5 days before injury. TA muscles were harvested 7 days after injury, sectioned, and stained for laminin. Nuclei were stained with DAPI.

(A) Diagram of time course that depicts fasting durations before BaCl 2 injury to the TA muscle, followed by refeeding and 7 days of recovery.

Discussion

Han et al., 2018 Han Y.-M.

Bedarida T.

Ding Y.

Somba B.K.

Lu Q.

Wang Q.

Song P.

Zou M.-H. β-Hydroxybutyrate prevents vascular senescence through hnRNP A1-mediated upregulation of Oct4. Han et al., 2018 Han Y.-M.

Bedarida T.

Ding Y.

Somba B.K.

Lu Q.

Wang Q.

Song P.

Zou M.-H. β-Hydroxybutyrate prevents vascular senescence through hnRNP A1-mediated upregulation of Oct4. Cheng et al., 2019 Cheng C.-W.

Biton M.

Haber A.L.

Gunduz N.

Eng G.

Gaynor L.T.

Tripathi S.

Calibasi-Kocal G.

Rickelt S.

Butty V.L.

et al. Ketone body signaling mediates intestinal stem cell homeostasis and adaptation to diet. Huang et al., 2018 Huang C.

Wang P.

Xu X.

Zhang Y.

Gong Y.

Hu W.

Gao M.

Wu Y.

Ling Y.

Zhao X.

et al. The ketone body metabolite beta-hydroxybutyrate induces an antidepression-associated ramification of microglia via HDACs inhibition-triggered Akt-small RhoGTPase activation. Veech et al., 2017 Veech R.L.

Bradshaw P.C.

Clarke K.

Curtis W.

Pawlosky R.

King M.T. Ketone bodies mimic the life span extending properties of caloric restriction. Huang et al., 2018 Huang C.

Wang P.

Xu X.

Zhang Y.

Gong Y.

Hu W.

Gao M.

Wu Y.

Ling Y.

Zhao X.

et al. The ketone body metabolite beta-hydroxybutyrate induces an antidepression-associated ramification of microglia via HDACs inhibition-triggered Akt-small RhoGTPase activation. Zou et al., 2016 Zou X.

Meng J.

Li L.

Han W.

Li C.

Zhong R.

Miao X.

Cai J.

Zhang Y.

Zhu D. Acetoacetate accelerates muscle regeneration aqqqnd ameliorates muscular dystrophy in mice. In addition to the hormonal and metabolic changes that accompany nutrient deprivation, fasting is characterized by a dramatic elevation in blood ketone levels (). Here, we show that the administration of ketone bodies themselves, in the absence of any other feature of nutrient deprivation, is sufficient to elicit DQ of MuSCs. Evolutionarily, elevated ketone bodies in the blood may be a rapid way to signal to cell types throughout the body to allocate resources to survival mechanisms, like increased resilience, during a time of nutrient deprivation. Consistent with our findings in MuSCs, this role of exogenous ketone bodies in promoting a quiescent state was recently elucidated in vascular cells (). The authors showed that BHB was sufficient to promote quiescence while simultaneously preventing senescence of these cells. Other studies have similarly shown that many of the beneficial effects of nutrient deprivation in other cell and tissue types can also be attributed, in large part, to the accompanying ketosis (). For example, a recent study showed that the neuroprotective and lifespan-extending properties of CR are in part mediated by the ketone bodies themselves (). Interestingly, however, one recent study has reported that a localized intramuscular injection of acetoacetate during muscle regeneration can promote MuSC proliferation and accelerate muscle regeneration through activation of the MEK1-ERK1/2-cyclin D1 pathway (). Notably, this study focused only on the effects of the acetoacetate immediately following injury, when MuSCs are proliferating, without exploring its role during quiescence. Additionally, this study did not look at the effects of systemic ketosis or its predominant ketone body, BHB, on MuSCs. As such, this study does not address the effects of ketosis on quiescent stem cell function.

Cheng et al., 2019 Cheng C.-W.

Biton M.

Haber A.L.

Gunduz N.

Eng G.

Gaynor L.T.

Tripathi S.

Calibasi-Kocal G.

Rickelt S.

Butty V.L.

et al. Ketone body signaling mediates intestinal stem cell homeostasis and adaptation to diet. Newman, and Verdin, 2014 Newman J.C.

Verdin E. Ketone bodies as signaling metabolites. Shimazu et al., 2013 Shimazu T.

Hirschey M.D.

Newman J.

He W.

Shirakawa K.

Le Moan N.

Grueter C.A.

Lim H.

Saunders L.R.

Stevens R.D.

et al. Suppression of oxidative stress by beta-hydroxybutyrate, an endogenous histone deacetylase inhibitor. Yang et al., 2014 Yang J.

Tang Y.

Liu H.

Guo F.

Ni J.

Le W. Suppression of histone deacetylation promotes the differentiation of human pluripotent stem cells towards neural progenitor cells. Cheng et al., 2019 Cheng C.-W.

Biton M.

Haber A.L.

Gunduz N.

Eng G.

Gaynor L.T.

Tripathi S.

Calibasi-Kocal G.

Rickelt S.

Butty V.L.

et al. Ketone body signaling mediates intestinal stem cell homeostasis and adaptation to diet. Leibowitz et al., 2018 Leibowitz B.J.

Yang L.

Wei L.

Buchanan M.E.

Rachid M.

Parise R.A.

Beumer J.H.

Eiseman J.L.

Schoen R.E.

Zhang L.

Yu J. Targeting p53-dependent stem cell loss for intestinal chemoprotection. Although the canonical function of ketone bodies is as a metabolic fuel source and precursor for the production of mitochondrial acetyl CoA, our work indicates that the role of ketone bodies in promoting DQ within MuSCs is not a result of their metabolism to acetyl CoA. Indeed, consistent with findings in other cell and tissue types (), our data suggest that ketone bodies, specifically the ketone BHB, may act via inhibiting HDAC activity. Specifically, our data indicate that class I HDAC inhibition, specifically HDAC1 within class I, may be driving the DQ of these cells. A number of previous studies have demonstrated the role of pharmacological HDAC inhibition in controlling stem cell fate and function (). Here, we present evidence in support of the notion that reduction of HDAC activity by an endogenous inhibitor, namely circulating BHB, may underlie the effects of fasting on a stem cell population.

Liu et al., 2019 Liu K.

Li F.

Sun Q.

Lin N.

Han H.

You K.

Tian F.

Mao Z.

Li T.

Tong T.

et al. p53 β-hydroxybutyrylation attenuates p53 activity. Brochier et al., 2013 Brochier C.

Dennis G.

Rivieccio M.A.

McLaughlin K.

Coppola G.

Ratan R.R.

Langley B. Specific acetylation of p53 by HDAC inhibition prevents DNA damage-induced apoptosis in neurons. Brochier et al., 2013 Brochier C.

Dennis G.

Rivieccio M.A.

McLaughlin K.

Coppola G.

Ratan R.R.

Langley B. Specific acetylation of p53 by HDAC inhibition prevents DNA damage-induced apoptosis in neurons. Liu, and Kulesz-Martin, 2000 Liu Y.

Kulesz-Martin M. P53 regulation and function in normal cells and tumors. Liu et al., 2019 Liu K.

Li F.

Sun Q.

Lin N.

Han H.

You K.

Tian F.

Mao Z.

Li T.

Tong T.

et al. p53 β-hydroxybutyrylation attenuates p53 activity. Roberts et al., 2017 Roberts M.N.

Wallace M.A.

Tomilov A.A.

Zhou Z.

Marcotte G.R.

Tran D.

Perez G.

Gutierrez-Casado E.

Koike S.

Knotts T.A.

et al. A ketogenic diet extends longevity and healthspan in adult mice. Recent work has demonstrated that p53 can be post-translationally regulated through β-hydroxybutyrylation (BHBylation), which attenuates acetylation and activation of p53 (). In that study, global p53 acetylation levels were decreased after BHB treatment, but p53 acetylation at individual residues was not determined. This is a notable caveat, as p53 K379 acetylation has a primary role in promoting the antiapoptotic functions of p53, despite the acetylation status of other residues (). In our hands, p53 becomes hyperacetylated at K379 during ketosis, indicating that ketosis could conceivably promote acetylation at certain prominent residues to drive p53 activity, whereas global p53 acetylation is downregulated through BHBylation. Consistently, our data show a robust transcriptional signature of upregulated p53 activity after ketone body treatment, suggesting that site-specific acetylation of p53 is perhaps more consequential for regulating p53 transcriptional activity in MuSCs than BHBylation. Furthermore, important differences have been observed in the regulation and activity of p53 between healthy cells and tumor cells (). This is another notable caveat, as regulation of p53 through BHBylation was studied in large part using U2OS, HCT116, and H1299 cancer cell lines (). Although the referenced study did find p53 BHBylation and target gene (PUMA and p21) downregulation in mouse thymus after fasting, p53 hyperacetylation and elevated target gene (DDIT4) expression has been previously observed in liver of mice that consumed a ketogenic diet (), suggesting that p53 regulation during ketosis is tissue dependent. Future studies could elucidate the influence of BHBylation on site-specific p53 acetylation, how BHBylation and acetylation at different sites might be integrated to modulate p53 transcriptional activity, and how p53 acetylation and BHBylation are differentially regulated in various tissues during ketosis.