Aged Stem Cells Reprogram Their Daily Rhythmic Functions to Adapt to Stress

Article

Graphical Abstract Authors Guiomar Solanas, Francisca Oliveira Peixoto, Eusebio Perdiguero, ..., EpidermalE and Paolo Sassone-Corsi, Pura Mun˜ oz-Ca´ noves, MuscleM SCs Salvador Aznar Benitah

Correspondence ADULT AGED [email protected] (P.M.-C.), salvador.aznar-benitah@irbbarcelona. Rhythmic org (S.A.B.) Reprograming In Brief Rhythmic functionstions RhythmicRhyt functions The daily rhythmic transcriptome is extensively reprogrammed in aged stem Caloric cells, switching from genes involved in Restriction homeostasis to those involved in tissue- Homeostasis Stress adaptation speciﬁc stresses.

Highlights d Tissue stem cells retain a rhythmic circadian machinery during aging d Daily rhythms are reprogrammed in aged SCs to cope with tissue-speciﬁc stress d Rewiring of daily rhythms in aged SCs is prevented by caloric restriction d Deletion of core clock genes does not recapitulate age- related reprogramming

Solanas et al., 2017, Cell 170, 678–692 August 10, 2017 ª 2017 Elsevier Inc. http://dx.doi.org/10.1016/j.cell.2017.07.035 Article

Aged Stem Cells Reprogram Their Daily Rhythmic Functions to Adapt to Stress

Guiomar Solanas,1,7 Francisca Oliveira Peixoto,1,7 Eusebio Perdiguero,3 Merce` Jardı´,3 Vanessa Ruiz-Bonilla,3 Debayan Datta,1 Aikaterini Symeonidi,1 Andre´ s Castellanos,1 Patrick-Simon Welz,1 Juan Martı´n Caballero,5 Paolo Sassone-Corsi,6 Pura Mun˜ oz-Ca´ noves,2,3,4,* and Salvador Aznar Benitah1,2,8,* 1Institute for Research in Biomedicine (IRB Barcelona), The Barcelona Institute of Science and Technology, 08028 Barcelona, Spain 2ICREA, Catalan Institution for Research and Advanced Studies, 08010 Barcelona, Spain 3Universitat Pompeu Fabra (UPF), Department of Experimental and Health Sciences (DCEXS) and CIBER on Neurodegenerative Diseases (CIBERNED), 08003 Barcelona, Spain 4Spanish National Center for Cardiovascular Research (CNIC), 28029 Madrid, Spain 5PCB-PRBB Animal Facilities, 08028 Barcelona, Spain 6Center for Epigenetics and Metabolism, University of California, Irvine, CA 92607, USA 7These authors contributed equally 8Lead Contact *Correspondence: [email protected] (P.M.-C.), [email protected] (S.A.B.) http://dx.doi.org/10.1016/j.cell.2017.07.035

SUMMARY segregating DNA replication from the light phase of the day, when maximal oxidative phosphorylation occurs (Gaddameedhi Normal homeostatic functions of adult stem cells et al., 2011; Geyfman et al., 2012; Stringari et al., 2015). The have rhythmic daily oscillations that are believed to day-night rhythmic oscillation of genes that are important for tis- become arrhythmic during aging. Unexpectedly, we sue homeostasis occurs in numerous tissues (Janich et al., 2014). find that aged mice remain behaviorally circadian The core circadian machinery is predominantly responsible for and that their epidermal and muscle stem cells retain establishing most daily rhythms and is essential for tissue a robustly rhythmic core circadian machinery. How- homeostasis. Thus, deletion of the core clock transcription factor Bmal1 in mice results in numerous aging-like pathologies ever, the oscillating transcriptome is extensively that severely reduce their lifespan; these include low body reprogrammed in aged stem cells, switching from weight, arthritis, brittle bones, corneal degeneration, diabetes, genes involved in homeostasis to those involved in intestinal permeability and inflammation, skin aging, and neuro- tissue-specific stresses, such as DNA damage or degeneration (Janich et al., 2014). Further, evidence suggests inefficient autophagy. Importantly, deletion of circa- that the expression of the core clock machinery in physiologically dian clock components did not reproduce the hall- aged mice is dampened in their suprachiasmatic nuclei (SCN) marks of this reprogramming, underscoring that (Bonaconsa et al., 2014; Chang and Guarente, 2013). Likewise, rewiring, rather than arrhythmia, is associated with humans show a decline in the robustness of their sleep and physiological aging. While age-associated rewiring wakefulness cycles during aging (Roenneberg et al., 2007). of the oscillatory diurnal transcriptome is not recapit- However, several important questions remain unanswered, ulated by a high-fat diet in young adult mice, it is particularly when considering the proven functional decline of most stem cell functions with aging (Goodell and Rando, significantly prevented by long-term caloric restric- 2015). For instance, is the rhythmic oscillation of adult SC function in aged mice. Thus, stem cells rewire their diurnal tions in peripheral tissues affected during aging? And if so, is it timed functions to adapt to metabolic cues and to altered in the same manner in different tissues? In addition, tissue-specific age-related traits. can these functional rhythms be normalized by physiological (nutritional) interventions previously shown to ameliorate health INTRODUCTION and lifespan during aging? To address these questions, we analyzed the effects of aging on the functional diurnal rhythms Tissue function is subject to daily fluctuations, and the activity of of SCs in two tissues with very different cellular turnover rates: adult stem cells (SCs) and their progenitors is consequently under the epidermis and the skeletal muscle (Tetteh et al., 2015). robust circadian control (Janich et al., 2014). For instance, in skin, the expression of activation or dormancy pathways oscillates in a RESULTS rhythmic diurnal manner to regulate the activation and timing of proliferation of hair follicle SCs (Janich et al., 2011; Plikus et al., Aged Adult SCs Retain a Rhythmic Core Clock 2013). Diurnal rhythms also temporally determine the optimal Machinery but Reprogram Their Rhythmic timing of proliferation and differentiation of epidermal SCs (epSCs) Transcriptome (Janich et al., 2013). Importantly, the core clock machinery pro- Despite sharing a similar functional decay with aging, epSCs and tects epSCs from genotoxic stress by dictating and temporally skeletal muscle SCs (muSCs) represent two extreme stem cell

678 Cell 170, 678–692, August 10, 2017 ª 2017 Elsevier Inc. 679 rmine 1 0.5 0 −0.5 −1 P-value 3.02E-08 1.81E-03 2.80E-03 1.44E-06 2.99E-04 Aged Adult

mean_CT20

mean_CT16 60

mean_CT12 5

mean_CT8 g 1 1560

mean_CT4 AgedAged

mean_CT0 DNA repair 9 1 0.5 0 −0.5 −1 74 749 mitotic M phase Amplitude rhythmic genes

mean_CT20 t (legend continued on next page) mean_CT16

mean_CT12 cell cycle DNA replication double strand break repair , 678–692, August 10, 2017 mean_CT8 395 Adult

mean_CT4 2395 2

Common mean_CT0

0 4 8 12 16 204 0 8 12 16 20 170

Amplitude of oscillations Rhythmic genes in EpSC Rhythmic

y t i s n e d

e n e G circadian regulation of gene expression Cell D enes g 1 0.5 0 −0.5 −1 P-value 4.50E-12 5.10E-06 1.28E-05 4.18E-04 8.72E-03 1.12E-04 3.00E-04 (p-val<0.05)

mean_CT20

mean_CT16 Adult Aged Apply JTK algorithm

mean_CT12 o identify rhythmic mean_CT8 to identify rhythmic genest

mean_CT4

mean_CT0 DNA repair A. 1 0.5 0 −0.5 −1 mismatch repair rhythmic genes

mean_CT20

recombinational repair ZT20-24 mean_CT16 cellular response to IL-6

mean_CT12 Figure S1 replication fork processing

double strand break repair

mean_CT8 Adult Aged Adult Aged ZT16-20

mean_CT4

mean_CT0

0 4 8 12 16 204 0 8 12 16 20

mitotic DNA damage checkpoint ZT12-16 Aged only ZT8-12

D:D1 ZT4-8 ZT0-4

0.4 0.3 0.2 0.1 0.0 Proportion of Visits of Proportion 1 0.5 0 −0.5 −1 P-value 7.80E-04 8.72E-06 3.89E-05 1.06E-03 1.45E-03 5.89E-08 mean_CT20

mean_CT16

mean_CT12

mean_CT8

ZT20

mean_CT4 ZT20-24

mean_CT0

secretion

ZT16

autophagy ZT16-20 0.05). The sorting strategy is described in 1 0.5 0 −0.5 −1

wound healing

ZT12-16 ZT12 ZT8-12

rhythmic genes

mean_CT20 L:D ZT8 ZT4-8 mean_CT16 epidermis development

mean_CT12

response to oxygen levels

keratinocyte differentiation ZT4 mean_CT8 ZT0-4 Adult Aged Rhythmic oscillations

Sorting of EpSCs at mean_CT4

6 time points of the day mean_CT0 0 4 8 12 16 204 0 8 12 16 20

0.4 0.3 0.2 0.1 0.0 ZT0 Adult only Proportion of Visits of Proportion C EFG A B Figure 1. Reprogramming of(A) the Experimental Rhythmic setup Transcriptome and in number Aged(lights of on), epSCs rhythmic ZT4, ZT8, genes ZT12, in ZT16, and adultdaily ZT20 (3,144 (with rhythmic genes) ZT genes given and (p in aged hours). Extracted (2,309 RNA genes) was epSCs. processed for epSCs microarrays, were and the sorted JTK every algorithm 4 was used hr to at dete time points ZT0 types: most epSCs proliferate almost continuously, while muSCs Aged epSCs Lose Their Rhythmic Expression of usually remain quiescent unless faced with muscle fiber damage Homeostasis Genes but Establish a De Novo Rhythmic (Clayton et al., 2007; Mascre´ et al., 2012; Sada et al., 2016; Transcriptome Predominantly Associated with DNA Sousa-Victor et al., 2015). Thus, we chose these two SC types Damage to study if the diurnal functional rhythms of SCs are altered To determine the functional consequences of this rhythmic during aging, and if so, whether this alteration is influenced reprogramming in aged SCs, we first analyzed gene ontology by how SCs behave in their steady state. We sorted integrin (GO) data from epSCs, which have a high turnover rate (Solanas a6bright/CD34– (for epSCs) or integrin a7bright/CD34+ (for muSCs) and Benitah, 2013). Transcripts that were oscillatory only in adult from (young) adult mice and aged mice (R 18 months) at six time epSCs predominantly regulated homeostatic functions, such as points over a day, using the zeitgeber times (ZT; time in 12 hr keratinocyte differentiation, as we have previously described light-12 hr dark cycles) of ZT0 (defined as lights on), ZT4, ZT8, (Janich et al., 2011; 2013)(Figure 1E; Table S1). In contrast, ZT12, ZT16, and ZT20. We then analyzed the whole transcrip- the rhythmic transcripts in aged epSCs were no longer related tome with gene expression arrays (Figures 1A, S1A, and S1B) to homeostatic functions but rather to stress conditions, and and used the Jonckheere-Terpstra-Kendall (JTK) algorithm to mainly to inflammation (cellular response to IL6) and DNA dam- identify transcripts that are rhythmically expressed in a diurnal age (mismatch repair, mitotic DNA damage checkpoint, double manner (i.e., with a period of approximately 24 hr; Table S1). strand break repair, and replication fork processing) (Figure 1F; We detected robust rhythmic expression in 3,144 transcripts Table S1). The cohort of genes that remained rhythmic upon from adult epSCs and 1,979 transcripts from adult muSCs, of aging belonged to three main categories: control of circadian which 76% and 72%, respectively, ceased to be oscillatory rhythms (further underscoring that aged SCs do not lose their in aged mice (Figures 1A and 2A; Table S1). Although both ability to remain rhythmic), DNA replication, and mitosis (Fig- types of aged SCs retained a similar percentage (23%–28%) ure 1G; Table S1). of transcripts that remained rhythmic, they each expressed a To better dissect which functions epSCs perform when mice new cohort of oscillatory genes that was not present in their are active (at night) and at rest (during the day) and to determine adult counterparts (Figures 1A and 2A). Loss or gain of rhythmi- if these functions change during aging, we performed GO cally expressed genes in aged SCs was not due to down- or analyses of the ‘‘diurnal’’ genes (peaking during ZT0–ZT6) and upregulation of their expression, as their average expression rhythmic ‘‘nocturnal’’ genes (peaking during ZT12–ZT18) in adult levels remained the same over time (Figure S1C). Thus, aged and aged epSCs (Figure S3A). We have previously shown that SCs do not become arrhythmic; rather, their oscillatory tran- adult epSCs are more predisposed to differentiate at night scriptome is drastically reprogrammed during physiolog- (Janich et al., 2011; 2013). Indeed, among the ‘‘night’’ gene ical aging. set, we observed a significant enrichment for genes involved in In aged mice, the amplitude of daily electrical rhythms is GO categories such as response to TGFbeta and keratinocyte reduced in the SCN, the central pacemaker of the brain that differentiation (Figure 3A; Table S1). On the other hand, aged transmits changes in light signals to the rest of the body (Farajnia epSCs no longer expressed ‘‘night’’ genes related to epidermal et al., 2012; Nakamura et al., 2011). However, we found that differentiation in a rhythmic manner (Figure 3A). Previous reports although aged mice were significantly less active than adult have shown that epSCs replicate their DNA at night, thus avoid- mice throughout the 24 hr day (Figure S2A), their physiological ing the interval of maximum oxidative phosphorylation that activity remained robustly circadian, with comparable ampli- occurs during the day and minimizing genotoxic stress (Geyfman tude, phase, and p values, both in light-dark conditions and after et al., 2012; Stringari et al., 2015). Accordingly, most genes of the two weeks of constant darkness (Figures 1B and S2B–SBD). DNA replication machinery showed rhythmic expression in adult Similarly, all core clock genes oscillated with equal amplitude SCs, with peaks at night, including those encoding the MCM and period in both aged and adult SCs, including those that are proteins, Cdc45, Rrm1, Rrm2, Gins1, Gins2, Lig1, and E2f1 transcriptionally regulated directly by Bmal1/Clock, or by the (many of which are essential for the initiation of DNA replication) secondary circadian loop ROR/Rev-Erb (Figures 1C, 2E, and (Figures 3B, S3D, and S3E). Interestingly, all DNA replication S1D). In addition, the average amplitude of oscillations of the genes were also oscillatory in aged epSCs (Figures 3A, 3B, daily rhythmic transcriptome was identical in adult and aged and S3D). However, in aged epSCs, but not adult epSCs, the SCs (Figures 1D and 2A). Thus, aged mice are not behaviorally oscillation of DNA replication genes was accompanied by the arrhythmic, and the core clock machinery remains robustly oscil- rhythmic expression of many genes involved in different types latory in their epSCs and muSCs. of DNA damage and repair (such as NER, BER, and Mismatch

(B) Plots showing adult and aged mice activity during the day in light-dark (L:D) or in constant darkness (D:D). Results are shown as number of visits in every interval normalized by the total number of visits during that week for each age group. Error bars represent ± SEM. (C) Expression levels over time of core clock circadian genes (Arntl, Npas2, Per2, Per3, Nr1d1, Nr1d2, Cry1, and Cry2) in adult and aged epSCs. Error bars represent SD. ZT24 = ZT0. (D) Distribution of oscillation amplitude of the adult and aged sets of rhythmic genes, normalized for their total number of rhythmic genes. Results are presented as gene density. (E and F) GO analysis of genes exclusively rhythmic in adult epSCs (E) or aged epSCs (F). (G) GO analysis of common rhythmic genes; heatmaps show their expression levels in adult epSCs (left) and aged epSCs (right). See also Figures S1 and S2 and Table S1.

680 Cell 170, 678–692, August 10, 2017 A Rhythmic genes in MuSC Aged JTK p-val < 0.005 Adult

Adult Agedg s

1421 558 16631663 G

Amplitude B C D Adult only rhythmic genes P-value Aged only rhythmic genes P-value Common rhythmic genes P-value cell proliferation 4.11E-13 inflammatory response 6.70E-03 circadian rhythm 1.94E-10 skeletal muscle tissue development 1.95E-05 mitochondrial DNA repair 2.40E-03 cell-substrate adhesion 5.47E-05 myotube differentiation 7.50E-04 cytokine production 1.30E-03 actin cytoskeleton organization 5.21E-05 cellular response to DNA damage stimulus 4.74E-03 response to redox state 5.39E-04 response to insulin 1.45E-03 Adult Aged Adult Aged Adult Aged

1 1 1 1 1 1

0.5 0.5 0.5 0.5 0.5 0.5

0 0 0 0 0 0

−0.5 −0.5 −0.5 −0.5 −0.5 −0.5

−1 −1 −1 −1 −1 −1 mean_CT0 mean_CT4 mean_CT8 mean_CT12 mean_CT16 mean_CT0 mean_CT4 mean_CT8 mean_CT16 mean_CT20 mean_CT20 mean_CT12 mean_CT20 mean_CT0 mean_CT4 mean_CT8 mean_CT12 mean_CT16 mean_CT20 mean_CT0 mean_CT4 mean_CT8 mean_CT12 mean_CT16 mean_CT0 mean_CT4 mean_CT8 mean_CT12 mean_CT16 mean_CT20 mean_CT0 mean_CT4 mean_CT8 mean_CT12 mean_CT16 mean_CT20 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20

E F

LAMP1 and LC3 Colocalization * ** (%) 60

N.S. ) * LC3 40 and

20 Colocalization

(LAMP1 0

Adut Adult Aged Aged 4 4 6 6 T T 1 1 Z Z T T Z Z

Figure 2. Reprogramming of the Rhythmic Transcriptome in Aged muSCs (A) Number of rhythmic genes and the amplitude of oscillation in adult (1,979 genes) and aged (2,221 genes) muSCs, sorted every 4 hr from ZT0. RNA was extracted and processed for microarrays. Rhythmic genes in muSCs were determined based on the JTK algorithm (p % 0.005). The sorting strategy is described in Figure S1B. The distribution of oscillation amplitude of the adult and aged sets of rhythmic genes was normalized for their total number of rhythmic genes. Results are presented as gene density. (B and C) GO analysis of genes exclusively rhythmic in adult muSCs (B) or aged muSCs (C). (D) GO analysis of common oscillatory genes; heatmaps show expression levels in adult muSCs (left) and aged muSCs (right). (E) Expression levels over time of core clock circadian genes (Arntl, Npas2, Per2, Per3, Nr1d1, Nr1d2, Cry1, and Cry2) in adult and aged muSCs. Error bars represent SD. ZT24 = ZT0. (F) Autophagy levels. Percentage of colocalization of LAMP1 and LC3 puncta in adult and aged muSCs at ZT4 and ZT16. Error bars represent ± SEM. *p % 0.05; **p % 0.01. See also Figures S1, S4, and S5 and Table S1.

Repair), including those related to replicative stress, such as coinciding with the time of maximum DNA replication in the Brca2 or Tipin (replication fork processing) (Figures 3A and S3E). epidermis (Figures 3B–3D and S3D). We hypothesized that this In adult epSCs, the oscillatory expression of almost every might reﬂect a regulatory loop that prevents the DNA replication gene involved in DNA replication and replicative fork processing machinery from unnecessarily re-initiating DNA replication dur- had a similar pattern, with a statistically signiﬁcant drop at ZT16, ing the night. Intriguingly, this drop was no longer present in

Cell 170, 678–692, August 10, 2017 681 aged epSCs, suggesting that DNA replication might be delayed (Figures S3A–S3C). Thus, aged epidermal SCs are susceptible to or stalled, indicative of replicative stress (Figures 3B, 3D, and accumulating oxidative DNA damage, as their timing of DNA S3D).We therefore next measured whether DNA replication replication is not correctly segregated from that of maximal changed between adult and aged epSCs. In adult epSCs, cellular oxidation. Importantly, however, aged epidermal SCs maximal DNA replication occurred at ZT16 (night) and dropped retain their drive to divide rhythmically, supporting daily tissue to low levels by ZT4 (day), as previously shown (Figure 3C) (Geyf- homeostasis. man et al., 2012; Stringari et al., 2015; Janich et al., 2013). Accordingly, the total mean intensities of foci marked with phos- Oscillatory Rewiring in Aged Quiescent muSCs pho-RPA32/RPA2 (single-stranded DNA) or gH2Ax (replicating, Associates with Loss of Basal Autophagy, but Not DNA as well as damaged, DNA) were slightly elevated at ZT16 as Damage compared to ZT4, indicating maximal DNA replication at Z16 Consistent with the normally quiescent state of muSCs, genes (Figures 3C, 3E, and 3F). Strikingly, however, 4-fold fewer aged involved in DNA replication were not expressed in a rhythmic epSCs were in S-phase at ZT16; yet, interestingly, about manner in either adult or aged muSCs; accordingly, aged muSCs 8–12 hr later (around ZT4), the percentage of aged epSCs in did not show any signs of replicative stress (Figures 2B–2D and S-phase was similar to that of adult epSCs at ZT16 (Figure 3C). S4D; Table S1). Nonetheless, adult quiescent muSCs expressed Accordingly, this extended the peak of pRPA/gH2Ax (indicative many transcripts that are required for their homeostasis in a of DNA replication) to ZT4 in aged epSCs (Figures 3E and 3F). rhythmic manner, including myotube differentiation and cell pro- Longer genes are more likely to accumulate transcription-stalling liferation (note that both of these GO categories encompass lesions (Vermeij et al., 2016). In agreement with this, we observed genes in the TGFb/Bmp and Fgf pathways, which regulate that, in aged epSCs, the longest genes were expressed at lower muSC quiescence maintenance and readiness for activation, levels during the night hours, when the replicative stress levels rather than cell division or myogenic differentiation genes per might be higher (Figure 3H). se) (Table S1)(Bernet et al., 2014; Chakkalakal et al., 2012; Chak- Thus, aged epSCs were less likely to enter S-phase at the kalakal and Brack, 2012; McCroskery et al., 2003). Adult muSCs physiological time at which DNA replication occurs in adult also expressed several transcripts in a daily oscillatory manner epSCs (ZT16), and the peak of their DNA replication was delayed that may be involved in double-strand break repair (cellular and extended well within day hours (i.e., ZT4). Importantly, this response to DNA damage stimulus), such as Rad23a, Ercc4, (mis)timing likely exposes unwound DNA to genotoxic stress, and Xpa, in agreement with previous observations that quiescent as it occurs when oxidative phosphorylation is maximal. In muSCs are more predisposed to repairing this type of damage fact, the rhythmic GO analysis confirmed that the expression of than differentiated muscle cells (Figure S4E) (Vahidi Ferdousi genes involved in oxidation-reduction process, mitochondrion et al., 2014). Interestingly, the expression of the majority of these organization, and response to oxidative stress peaked during genes peaked during the night, suggesting that this activity is the day (ZT0–ZT6) in adult epSCs (Figures S3A–S3C). Impor- strongest in muSCs when mice are actively using their muscles. tantly, the oscillatory profile (i.e., phase and amplitude) of these Similar to aged epSCs, aged muSCs mostly lost the regulated genes was maintained in aged epSCs (Figure S4A). Interestingly, timing of their homeostatic functions and gained a new rhythmic aged epSCs presented high levels of DNA oxidation, as transcriptional program—in this case, for genes involved measured by 8oxodG, which persisted throughout the day (Fig- in inflammation, cytokine production, and mitochondrial DNA ure 3G). Thus, the concomitant peaks of DNA replication and repair (Figures 2C and S5B; Table S1). oxidative stress might contribute to the accumulation of oxidized As mentioned above, approximately 25% of the diurnal DNA in aged epSCs; future experiments will be necessary to rhythmic genes in adult epSCs and muSCs remained oscillatory determine this causal link. in their aged counterparts. In aged muSCs, the core clock As aged epSCs extended their DNA replication into day (ZT4), machinery remained robustly circadian, and both the average we tested whether mitosis was also delayed. Although about amplitude and period of oscillation of the genes commonly rhyth- 50% fewer aged epSCs entered M-phase than adult epSCs, mic in adult and aged muSCs were the same (Figures 2D and 2E; strikingly, maximal M-phase occurred at ZT20 in both adult Table S1). Interestingly, the majority of the rhythmic genes that and aged epidermis, which is 4 hr after the S-phase peak in adult were common to adult and aged muSCs were predominantly epSCs (Figure S4B). No additional mitosis peak was observed involved in cell-substrate adhesion and cytoskeletal organization after ZT4 in aged epSCs (Figure S4B). Thus, aged epSCs dealing (Figure 2D; Table S1) and included genes encoding paxillin, with replicative stress at ZT4 (Figures 3E and 3F) might either be vinculin, calponin, and CSF1r, which are associated with muSC stalled until ZT20, as occurs in yeast with replicative stress interactions with the local environment (Goetsch et al., 2014; Ishii undergoing mitosis (Amaral et al., 2016), or be cleared away by and Lo, 2001; Segawa et al., 2008; Shibukawa et al., 2013). This a yet-unidentified mechanism, independent of apoptosis (Fig- is consistent with the well-accepted notion that the resting ure S4C). The observation that these aged epSCs, burdened muSC is ready to respond with the appropriate receptors and with damaged DNA, still underwent rhythmic mitosis at the cor- signaling pathways but protects its quiescence by mechanisms rect time correlates well with the large set of genes involved in that include immobilization of ligands by extracellular matrix mitotic M-phase that remained rhythmic in aged epSCs (Fig- (ECM) components and synthesis of inhibitors for intracellular ure 1G; Table S1). Accordingly, both adult and aged epSCs ex- signaling pathways (Montarras et al., 2013). Thus, rhythmically pressed numerous genes involved in the last step of cell division expressed genes allowing muSCs to interact with the ECM and (with GOs such as cytokinesis) in the early day hours (ZT0–ZT6) surrounding niche components may be required to preserve

682 Cell 170, 678–692, August 10, 2017 A Adult “night” rhythmic genes P-value Aged “night” rhythmic genes P-value B DNA replication 1.41E-12 DNA replication 9.00E-40 DNA repair 6.82E-07 DNA repair 7.79E-30 Adult * ER unfolded protein response 2.16E-06 recombinational repair 4.08E-12 Aged rhythmic process 4.22E-06 DNA damage checkpoint 2.10E-09 response to hypoxia 3.13E-05 replication fork processing 1.52E-07 epidermal cell differentiation 2.40E-05 nucleotide-excision repair 2.78E-04 keratinocyte differentiation 7.49E-05 base-excision repair 2.61E-04 keratinization 2.11E-04 rhythmic process 4.63E-04 regulation of translation 2.81E-04 DNA replication checkpoint 6.24E-04 response to transforming growth factor beta 7.13E-03 TOR signaling 6.50E-04 non-recombinational repair 9.32E-04 * mismatch repair 1.45E-03

C PCNA D MCM4 30 Adult 40 Adult Aged * 30 Aged 20 MCM4 Adult ZT16 MCM4 Adult ZT20 20 10 **

epidermal cells 10 Dapi+ basal cells % of MCM4+ cells/ % of PCNA+ cells/ PCNA+ % of MCM4 Aged ZT16 MCM4 Aged ZT20 0 0

ZT0 ZT4 ZT8 ZT12 ZT16 ZT20 ZT24 ZT16 ZT20 E ZT4 ZT16 pRPA32/RPA2 3

2 * Adult Adult 1

0 Mean intensity per Dapi+ pixel p-RPA32/RPA2

Adult ZT4Aged ZT4 Aged Aged Adult ZT16Aged ZT16

F yH2Ax 3

2 Adult Adult *

1 yH2Ax

0 Mean intensity per Dapi+ pixel Dapi+ per intensity Mean Aged Aged ZT4 Adult Aged ZT4 G Adult ZT16Aged ZT16 8OHdG 10 **** * Adult Adult 8 6

8OHdG 4

Aged Aged Mean intensity per Dapi+ pixel 0

Length of genes down regulated in Aged H Genes Adult ZT4Aged ZT4 0.6 0h; 4h; 20 Adult ZT16Aged ZT16 8h; 12h; 1 ZT0; ZT4; ZT20 0.4 ALL ZT8; ZT12; ZT16 ALL 0.2 Gene density

0.0 23456 Length (log10 of bases)

(legend on next page)

Cell 170, 678–692, August 10, 2017 683 the bona fide quiescence state of muSCs throughout life (Bent- muscle, such as a decrease in the thickness of the cornified zinger et al., 2013; Lukjanenko et al., 2016; Rozo et al., 2016; envelope, thicker fur, and a higher number of muSCs (Figures Tierney et al., 2016). Loss of quiescence by anticipated prolifer- 5A and S6A–S6C). ation or senescence entry can affect stemness and regenerative Strikingly, CR led to the maintenance in the aged SCs of traits functions in muSCs (Chakkalakal et al., 2012; Sousa-Victor et al., associated with the young age and a strong reprogramming of 2014). However, we did not observe any signs of unsolicited acti- the rhythmic transcriptome of aged epSCs and aged muSCs vation in aged muSCs, which remained in a quiescent state (Figures 4A and 5B; Table S2). The average expression levels similar to adult muSCs, as demonstrated by the absence of of the genes that changed their oscillatory behavior in SCs expression of proliferative markers (MyoD or Ki67) (data from CR-fed mice was the same as in their control diet-fed coun- not shown). terparts; thus, the changes in rhythmicity were not due to loss or A progressive decline in basal autophagy in quiescent muSCs de novo gene expression after exposure to distinct nutritional underlies the loss of fitness and functionality in old mice (Garcıá- diets (Figures S6D and S6E). While the amplitude and period of Prat et al., 2016). Interestingly, we found that adult muSCs the core clock genes were unaffected by CR, we observed a expressed genes involved in autophagy in a rhythmic manner 4-hr advance in their zenith due to a food anticipation effect (Figure S5A). For instance, key autophagy genes, including that has been previously described for CR diets (Figures S6F Becn1, Flcn, Atg13, and Svip, peaked late at night or early in and S6G). the morning in adult muSCs. We confirmed by co-immunostain- GO analysis of the genes that were uniquely rhythmic in aged ing of LC3 and LAMP1, two autophagy and lysosomal markers, epSCs in control diet-fed mice revealed very similar categories that adult muSCs have higher levels of autophagy during the day as in the previous cohort of aged mice, including DNA damage, than at night (Figures 2F and S5D). In stark contrast, aged inflammation, and a predominant lack of epidermal homeostatic muSCs no longer showed autophagy as an enriched GO term categories (Figure 4B; Table S2). Strikingly, however, daily oscil- in the ‘‘day’’ or ‘‘night’’ gene set (Figures S5B). Importantly, auto- latory homeostatic functions were maintained in aged epSCs phagy was also significantly reduced throughout the 24-hr day in from CR-fed mice (keratinocyte development and epithelial cell aged muSCs (Figures 2F and S5D), reinforcing the notion that a proliferation) (Figure 4C; Table S2). These included genes decline in basal autophagy is a hallmark of the changes in timing involved in Bmp and TGFbeta signaling, as well those for of the function of quiescent muSCs with aging. Thus, our results EGFR, VDR, or Jarid2, all of which regulate epSC activation or indicate that aged muSCs retain their ability to remain quiescent differentiation (Solanas and Benitah, 2013)(Figure 4C). Aged and to rhythmically regulate their interaction with their niche. epSCs from mice fed either control diet or CR shared a common However, they significantly lose their capacity to rhythmically rhythmic transcriptome that was predominantly associated with recycle the damaged components that are likely generated on the core clock machinery and DNA replication (Figure 4D). How- a daily basis. ever, in aged epSCs from CR-fed mice (but not control diet-fed mice), replicative stress levels were low (Figure 4E), the correct Caloric Restriction Prevents the Age-Dependent (‘‘adult-like’’) expression pattern of all the DNA replication Reprogramming of Diurnal Oscillations of Aged SCs machinery was maintained (Figures 4F and S6H), the nocturnal Caloric restriction (CR) extends the lifespan of many organisms, S-phase peak was unaltered (Figure 4G), and the amount of including rodents (Froy and Miskin, 2010). CR also enhances the oxidized DNA was significantly lower (Figure 4H). Nonetheless, function of adult SCs, including the regenerative capacity of CR did not prevent the age-related rhythmic expression of aged muSCs (Cerletti et al., 2012; Mihaylova et al., 2014), certain genes involved in double-strand break repair or base- through yet-uncharacterized mechanisms. To address whether excision repair, suggesting that some age-related DNA damage CR regulates the rhythmic behavior of SCs during aging, we persists in aged epSCs even in CR-fed mice (Figure 4D). fed a large cohort of aged (60-week-old) mice (C57BL/6 pure Aged muSCs from CR-fed mice also showed a dramatic background) with either a 30% CR diet or a control diet ad libitum reprogramming of their rhythmic transcriptome as compared for 25 weeks. We then analyzed whole transcriptomes by gene to their control aged counterparts (Figure 5B). For instance, expression arrays using FACS-sorted epSCs and muSCs they displayed an oscillatory expression of genes involved in collected at six time points over a 24-hr period. Besides losing autophagy, including Tmem59, Pycard, Ulk3, Park2, and Dap1. weight, mice under CR conditions showed a significant amelio- Importantly, this maintains their ability to undergo rhythmic ration of aging-associated traits, both in the epidermis and the autophagy during the day, as seen in healthy adult mice

Figure 3. Aged epSCs Delay the Timing of DNA Replication and Show Signs of Replicative Stress (A) GO analysis performed separately on rhythmic genes with phase ZT12–ZT18 (‘‘night’’ rhythmic genes) in adult and aged epSCs. (B) Expression levels over time of the ZT16-subset genes Mcm2 and E2f1 in adult and aged epSCs. Error bars represent SD. (C) Quantification of PCNA-positive adult and aged epSCs every 4 hr from ZT0. D) Quantification of MCM4-positive adult and aged basal cells at the time points ZT16 and ZT20, and representative images. Scale bar, 10 mm. (E–G) Representative images and quantification of intensity levels of p-RPA, gH2ax, and oxidised DNA at ZT4 and ZT16 in adult and aged epSCs (normalized to adult epSCs at ZT4). Scale bar, 5 mm. (H) Length of early phase (ZT20, ZT0, ZT4) and late phase (ZT8, ZT12, ZT16) genes that were downregulated in aged epSCs, and of all genes (ALL) in the mouse genome. *p % 0.05, **p % 0.01, ***p % 0.001, ****p % 0.0001. Error bars represent ± SEM. ZT24 = ZT0. See also Figures S3 and S4 and Table S1.

684 Cell 170, 678–692, August 10, 2017 A Rhythmic oscillations Rhythmic genes in EpSC Normal Diet CR Diet

Apply JTK algorithm

toto identiidentifyfy rhythmic genesgenes 11657657 353 10101011 0 ZT0 ZT4 ZT8 ZT12 ZT16 ZT20 (p-val<0.005) Sorting of EpSCs at 6 time points of the day

E B Normal Diet CR Diet Adult ZT12

1 1 1 Normal Diet rhythmic genes P-value 0.5 0.5 0.5 0 0 0

−0.5 −0.5 −0.5

DNA repair 1.92E-10 −1 −1 −1 intra-S DNA damage checkpoint 9.21E-04 response to ER stress 1.24E-03 nucleotide-excision repair 1.08E-03 ND response to superoxide 3.33E-03 TNF superfamily cytokine production 5.47E-03 yH2Ax

C 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 CR

1 1 1 CR Diet rhythmic genes P-value 0.5 0.5 0.5 0 0 0

−0.5 −0.5 −0.5

translation 5.14E-08 −1 −1 −1 oxidation-reduction process 4.98E-05 epithelial cell proliferation 6.93E-05 keratinocyte development 6.00E-04 ND BMP signaling pathway 2.91E-04 regulation of Tgf beta1 production 1.16E-03 positive regulation of TOR signaling cascade 1.84E-03

wound healing 3.93E-03 p-RPA32/RPA2

0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 CR D

1 1 1 P-value 0.5 0.5 0.5 yH2Ax ZT12 pRPA32/RPA2 ZT12 Common rhythmic genes 0 0 0

−0.5 −0.5 −0.5 1.5 1.5

−1 cell cycle 1.57E-15 −1 −1

DNA replication 7.32E-14 sity en t l l e xe 1.0 cell cycle checkpoint 6.67E-06 i 1.0 ** *** circadian rhythm 1.84E-04 Dapi+ p

DSB repair 1.63E-04 Dapi+ pix

0.5 r 0.5 DNA integrity checkpoint 1.45E-03

Base-excision repair, gap-filling 3.00E-03 Normalised Meanin per Normalised Meanintensity pe DNA damage checkpoint 5.43E-03 0.0 0.0

0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 CR Diet CR Diet F Normal Diet Normal Diet

*** ***

G H ZT16 PCNA 8OHdG / 1.5 y 40 Normal Diet t ensi

CR Diet t l ** e 1.0 pix ean in

30 + * ND Dapi

r 0.5 20 8OHdG Normalised M pe 0.0 10 t number of Dapi+ basal cells

% of total number PCNA+ cells ZT0 ZT4 ZT8 ZT12 ZT16 ZT20 ZT24 CR CR Diet Normal Die

(legend on next page)

Cell 170, 678–692, August 10, 2017 685 (Boada-Romero et al., 2013; Durcan and Fon, 2015; Koren et al., phagy levels also remained identical irrespective of diet (Figures 2010; Suzuki et al., 2007; Young et al., 2009)(Figure 5C). These 6F and S7G; Table S3). Nonetheless, comparing the rhythmic results could provide a possible mechanistic explanation to the transcriptome in muSCs from HFD-fed mice with those from previously reported functional rejuvenating effect of CR on aged mice revealed a significant overlap of genes involved in aged quiescent muSCs (Cerletti et al., 2012). Of note, aged inflammation (regulation of inflammatory response) (Figure 6C). muSCs from CR-fed mice no longer expressed genes involved It is worth noting that even this relatively short 7-week interval in inflammation or mitochondrial DNA repair in a rhythmic of HFD led to a significant reduction in the number of muSCs, manner, but instead expressed an oscillating homeostatic tran- a hallmark of skeletal muscle aging (Figure 6D). Thus, HFD scriptome program (with GO terms of regulation of myotube strongly rewires the rhythmic output of adult SCs, most likely differentiation, myoblast proliferation, and skeletal muscle to allow them to cope with fatty acid metabolism, yet—unlike regeneration) (Figures 5D, 5E, and S5C; Table S2). Expression physiological aging—it does not prevent adult SCs from ex- of core timed functions related to circadian rhythms and cellular pressing their homeostatic transcriptome in an oscillatory daily adhesion that remained unaltered in aged muSCs was unaf- manner (Table S3). fected by CR (Figure 5F; Table S2). Overall, our data indicate that CR is significantly capable of preserving the correct timing Clock Disruption Does Not Recapitulate Traits of of adult SC functions that is lost during aging. Physiologically Aged SCs Our results show that physiological aging is associated with High-Fat-Diet-Induced Rewiring of the Rhythmic reprogramming rather than with arrhythmia. To substantiate Transcriptome Does Not Overlap with the this, we analyzed whether clock disruption recapitulates any Reprogramming of Aged SCs features of age-associated reprogramming of the rhythmic func- In contrast to CR, exposure to a prolonged high-fat diet (HFD) tion of SCs, using Bmal1 KO and Per1/Per2 double-KO (Per1/2 accelerates the development of many aging-related pathologies, dKO) mice. Strikingly, epSCs from these KO mice showed no including chronic inflammation, cardiovascular disease, dia- increases in oxidized DNA or replicative stress foci during the betes, and a higher predisposition to develop cancer (Lo´ pez- active phase of the day (Figures 7A–7C) and no changes in the Otıń et al., 2013). Moreover, HFD rewires the liver circadian expression of DNA replication genes (Figures 7D and 7E), as transcriptome and metabolome in an obesity-independent compared to WT mice. Likewise, muSCs from Bmal1 KO mice manner (Eckel-Mahan et al., 2013). We therefore investigated did not have altered levels of autophagy (Figure 7F). Thus, whether a HFD would reprogram the rhythmic transcriptome of although Bmal1 KO SCs display strong aging-like symptoms, adult SCs similar to physiological aging. We ad libitum fed adult these changes do not recapitulate many of the differences in (8-week-old) C57BL/6 mice a HFD or a control diet for 7 weeks, rhythmic processes observed during physiological aging. We prior to reaching obesity in HFD-fed mice (Figure S7A), and therefore conclude that the reprogramming of the rhythmic func- analyzed the oscillatory transcriptome of epSCs and muSCs tions observed during physiological aging cannot be attributed (as described above). As previously observed for the liver, HFD to an arrhythmic function of the circadian clock. induced a very potent reprogramming of the rhythmic daily transcriptome of adult epSCs and muSCs, which was not due to DISCUSSION major gains or losses in gene expression levels (Figures S7B– S7E; Table S3). In both SC types, many genes involved in this By analyzing in unprecedented detail the perturbations in the oscillatory transcriptome rewiring were related to fatty acid timing of epSCs and muSCs functions during aging, we have oxidation, response to oxidative stress, and mitochondrial orga- determined that aged SCs remain robustly rhythmic. Intriguingly, nization (Figures S7B and S7D; Table S3). However, HFD-fed however, the rhythmic functions of aged SCs are rewired such mice retained the non-age-associated rhythmic expression of that they adapt to the new conditions of stress associated genes involved in the homeostatic function of epSCs and muSCs with the aged environment. This reprogramming is likely gener- (Figures 6A and 6B). Unlike aging, HFD did not increase the ally associated with aging, as it also occurs in the liver (see levels of replicative stress or the amount of oxidized DNA in accompanying study by Sato et al., 2017) and intestine (data epSCs at ZT16 (Figures 6E and S7F). In muSCs, timed auto- not shown).

Figure 4. CR Prevents the Reprogramming of the Rhythmic Transcriptome of Aged epSCs (A) Experimental setup and number of rhythmic genes in epSCs from aged animals fed with normal diet (2,010 genes) or CR diet (1,363 genes). EpSCs were sorted every 4 hr from ZT0. Extracted RNA was processed for microarrays, and the JTK algorithm was used to determine daily rhythmic genes (p % 0.005). (B and C) GO analysis of genes exclusively rhythmic to normal diet (B) or CR diet (C). (D) Gene ontology analysis of common rhythmic genes; heatmaps show expression levels in normal-diet epSCs (left), CR-diet epSCs (center), and adult epSCs (right). (E) Representative images and quantification of intensity levels of p-RPA and gH2ax at ZT12 in normal diet and CR basal cells (normalized to normal-diet cells at ZT12). (F) Expression levels across time points of ZT16 subset genes Chek1 and E2f1 in normal-diet and CR epSCs. Error bars represent SD. (G) Quantification of PCNA-positive normal-diet and CR epSCs analyzed every 4 hr from ZT0. (H) Representative images and quantification of intensity levels of oxidised DNA at ZT16 are given for normal-diet and CR epSCs (normalized to normal-diet cells at ZT16). *p % 0.05, **p % 0.01, ***p % 0.001. Error bars represent ± SEM. Scale bar, 5 mm. ZT24 = ZT0. See also Figure S6 and Table S2.

686 Cell 170, 678–692, August 10, 2017 1 0.5 0 −0.5 −1 s ). 1 0.5 0 −0.5 −1 1 0.5 0 −0.5 −1 nd 687 iet’’).

mean_CT20 mean_CT20 mean_CT20 n

mean_CT16 mean_CT16 mean_CT16

mean_CT12 mean_CT12 mean_CT12

mean_CT8 mean_CT8 mean_CT8 Adult Adult Adult mean_CT4 mean_CT4 mean_CT4 ** mean_CT0 mean_CT0 mean_CT0 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 *

1 0.5 0 −0.5 −1 1 0.5 0 −0.5 −1

1 0.5 0 −0.5 −1 ZT16 CR Diet CR ZT16 ZT16 ND ZT16 N.S.

mean_CT20 mean_CT20 mean_CT20

mean_CT16 mean_CT16 *** mean_CT16

mean_CT12 mean_CT12 mean_CT12 mean_CT8 mean_CT8 Diet CR ZT4 mean_CT8

mean_CT4 mean_CT4 mean_CT4 ND ZT4

mean_CT0 mean_CT0 mean_CT0 0

60 40 20

(%) ) LC3 and (LAMP1

1 0.5 0 −0.5 −1 1 0.5 0 −0.5 −1 1 0.5 0 −0.5 −1

, 678–692, August 10, 2017 LAMP1 and LC3 Colocalizatio Colocalization 170

mean_CT20 mean_CT20 mean_CT20

mean_CT16 mean_CT16 mean_CT16

mean_CT12 mean_CT12 mean_CT12 Cell

mean_CT8 mean_CT8 mean_CT8

mean_CT4 mean_CT4 mean_CT4

mean_CT0 mean_CT0 mean_CT0 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 Normal Diet CR Diet Normal Diet CR Diet Normal Diet CR Diet m. m MERGE 0.005). % P-value P-value P-value 5.98E-03 6.52E-03 8.84E-03 3.99E-03 8.68E-03 1.87E-03 7.46E-03 8.15E-03 1.19E-06 7.00E-04 8.06E-03 1.61E-07 9.99E-04 2.48E-07 9.04E-03 2.20E-03 1.97E-03 2.98E-03 4.11E-03 5.78E-09 DAPI translation TOR signaling circadian rhythm rhythmic genes rhythmic genes rhythmic genes LAMP1 cytokine production response to hypoxia ribosome biogenesis myoblast proliferation regulation of cell cycle inflammatory response ROS metabolic process regulation of autophagy response to redox state mitochondrial DNA repair oxidative phosphorylation response to nutrient levels response to oxidative stress adherens junction assembly Common CR Diet only SEM. N.S., not signiﬁcant. Scale bar, 5 DAPI skeletal muscle tissue regeneration

regulation of myotube differentiation ± Normal Diet only Normal Diet LC3 skeletal muscle satellite cell differentiation E F D . t 1 2 Die

9 * 1 1921 Table S2 DAPI CR DietCR

0.001. Error bars represent and % N.S. t Pax7 S6

ie 586 586 D * BF and MuSC number al 0.01, ***p JTK p-val < 0.005 % Adult Diet Normal CR Diet 635 Normal Diet 1 1635 Figures S5 Rhythmic genes in MuSC 0

75 50 25 Cells/mg tissue Cells/mg 0.05, **p % ND CR ZT16 C ZT16 B processed for microarrays, and the JTK algorithm was used to determine oscillating genes (p (B) Number of rhythmic genes in aged muSCs from mice fed a normal diet (2,221 genes) or a CR diet (2,507 genes), sorted every 4 hr from ZT0. Extracted RNA wa Figure 5. CR Prevents(A) Reprogramming Number of of the muSCs per Daily mg Rhythmic of Transcriptome muscle tissue of in Aged normal-diet muSCs adult muSCs (‘‘adult’’), normal-diet aged muSCs (‘‘normal diet’’), and CR aged muSCs (‘‘CR d A (C) Autophagy levels. Representative images and percentageZT16. of colocalization Note of that LAMP1 LAMP1 and LC3 and puncta(D LC3 in and aged puncta muSCs E) are from GO normal-diet present or analysis in CR(F) of all mice GO genes at interstitial ZT4 exclusively analysis muscle a rhythmic of cells in and common*p aged in rhythmic normal-diet muscle genes; muSCs ﬁbers. heatmaps (D) muSCs show or boundaries CR-diet expression are aged levels indicated muSCs (dotted in (E). line). normal-diet muSCs (left), CR-diet muSCs (center) and adult muSCs (right See also A B Rhythmic genes in EpSC Rhythmic genes in MuSC JTK p-val < 0.05 JTK p-val < 0.005 JTK p-val < 0.05 Agedg Highg Fat Diet Agedg Highg Fat Diet

11745745 563 32423 11778778 445 32153213 5

Aged rhythmic genes P-value Aged rhythmic genes P-value DNA repair 4.77E-06 cytokine production 2.92E-03 DNA damage checkpoint 5.45E-05 mitochondrial DNA repair 1.17E-03 mitotic DNA damage checkpoint 2.64E-04 immune response 7.59E-03 double strand break repair 4.10E-04 cellular response to IL-6 6.72E-04 High Fat Diet rhythmic genes P-value ribosome biogenesis 1.04E-05 High Fat Diet rhythmic genes P-value response to oxidative stress 4.57E-06 oxidation-reduction process 9.31E-10 regulation of muscle cell differentiation 2.02E-05 fatty acid metabolic process 3.72E-07 autophagy 6.63E-05 response to oxidative stress 8.86E-06 fatty acid metabolic process 1.85E-04 skin development 5.45E-05 skeletal muscle tissue development 2.53E-04 wound healing 9.63E-04 DNA damage checkpoint 5.48E-04 mitochondrial transport 1.05E-03 regulation of mitochondrial organization 1.83E-03 epidermal cell differentiation 2.12E-03 response to ROS 2.55E-03 ribosome biogenesis 3.21E-03 regulation of fatty acid beta-oxidation 7.15E-03 mismatch repair 4.68E-03 respiratory electron transport chain 8.29E-03 response to hypoxia 5.50E-03 TOR signaling 9.46E-03 keratinocyte proliferation 6.32E-03 regulation of DNA repair 9.46E-03

Common rhythmic genes P-value C DNA repair 3.49E-16 DNA replication initiation 1.18E-15 Common rhythmic genes P-value rhythmic process 3.03E-06 circadian rhythm 1.94E-08 double strand break repair 3.85E-06 regulation of inflammatory response 1.91E-03 postreplication repair 2.48E-05 cytokine biosynthetic process 8.86E-03 recombinational repair 5.40E-05 DNA replication checkpoint 3.44E-04 hexose metabolic process 4.79E-04 D E F Number of MuSCs (a7itg+/CD34 bright) 8OHdG LAMP1 and LC3 Colocalization *** ty

80 i * 1.5 ) N.S. 50 (% ** ) 40 60 1.0 30

20 40 0.5 Colocalization 10 (LAMP1 and LC3 Cells/mg tissue 20 per Dapi+ pixel Normalised Mean intens Normalised Mean 0.0 0 D D D F N H HF 0 4 ZT4 NDT ZT4 ND Z ZT16 ZT16 ND ND HFD ZT16 ZT4 HFD ZT16 HFD

Figure 6. HFD Induces Rhythmic Transcriptome Reprogramming in Adult SCs that Is Distinct from Aging Reprogramming (A and B) Overlap between aged and HF-diet rhythmic genes in epSCs (A) and muSCs (B). GO analyses are shown for rhythmic genes exclusive to aged SCs (top) or SCs from HFD adult mice (middle) or common to both (bottom). Note that many transcripts involved in homeostatic functions remained rhythmic (included in GO categories such as keratinocyte proliferation, epidermal cell differentiation, and skin development, in HFD epSCs [A], and skeletal muscle tissue development, regulation of muscle cell differentiation, and autophagy in HFD muSCs [B]). (C) GO analysis showing oscillatory genes common between aged muSCs from normal diet mice and adult muSCs from HFD mice. (D) Number of muSCs per mg of muscle tissue in adult muSCs from normal-diet or HFD mice. (E) Quantiﬁcation of intensity levels of oxidised DNA at ZT4 and ZT16 (normalized to normal-diet epSCs at ZT4) in epSCs from adult mice fed with a normal diet or HFD. (F) Percentage of colocalization of LAMP1 and LC3 puncta in muSCs from normal diet or HFD mice at ZT4 and ZT16. *p % 0.05, **p % 0.01, ***p % 0.001. See also Figure S7 and Table S3.

688 Cell 170, 678–692, August 10, 2017 A B

(legend on next page)

Cell 170, 678–692, August 10, 2017 689 Healthy SCs segregate their predisposition to perform However, we show that this is unlikely to be related to direct different functions along the 24-hr day to maximize energetic disruption of the major clock components, as aged SCs still efficiency and performance efficacy while minimizing exposure have a clock machinery as robustly rhythmic as adult SCs, and to potentially harmful situations. Hence, the loss of the oscillatory neither Bmal1 KO nor Per1/2 dKO mice showed several of the transcriptome of SCs as they age can progressively carry dire traits associated with physiologically aged stem cells. Instead, consequences for their functions. Thus, aged human epSCs the aging-related rhythmic reprogramming in SCs seems to that have lost their ability to separate the timing of DNA replica- heavily rely on a metabolic entrainment. Indeed, we have found tion from that of maximal UV exposure, or that no longer regulate that CR has a profound protective effect on many of the rhythmic their daily NAD metabolism, have a significantly higher risk of changes that occur during physiological aging, which is consis- accumulating DNA damage than their younger homeostatic tent with the beneficial effects of CR on the function of aged stem counterparts (Janich et al., 2013; Stringari et al., 2015). Likewise, cells (Cerletti et al., 2012). Circadian rewiring has previously been liver hepatocytes that can no longer process nutrients at the time shown to occur in the liver of mice either fed a high-fat diet or with of food ingestion may have to shift to more oxidative forms of en- systemic signals secreted by lung-tumor cells, allowing the liver ergy production, which are potentially more harmful (see accom- to adapt to the metabolic changes required in both scenarios panying manuscript by Sato et al., 2017; Koike et al., 2012). (Eckel-Mahan et al., 2013; Masri et al., 2016). Thus, SCs might Nonetheless, some cellular functions remain strongly rhythmic be evolutionarily prepared to quickly reprogram their timed func- in aged SCs. For instance, aged epSCs still undergo daily rhyth- tions in response to different contexts of metabolic changes and mic cell divisions, despite having oxidized and damaged DNA. tissue-specific stresses. Importantly, this reprogramming ability We hypothesize that rhythmic functions that are essential for of SCs may underlie the previously observed beneficial effects of each subset of adult SCs are those that are more likely to be re- CR on SC function. Future studies will be required to identify tained during aging. Thus, in tissues with a high demand of which components are responsible for the aging-related rewiring cellular replenishment, like the epidermis, the clock machinery of the daily fluctuating functions of SCs and to determine appears to play a dominant role in ensuring a daily input of whether they could be therapeutically targeted to maintain the new cells at the right time of the day, irrespective of the presence proper timing of SC function during aging in humans. of damaged DNA. In contrast, in muSCs, the genes most resilient to loss of rhythmic control during aging were those involved in STAR+METHODS the physical interaction with the surrounding niche (myofiber, stroma, and niche cells)—which is essential for protection of Detailed methods are provided in the online version of this paper the key quiescent state in resting muscle, and hence for preser- and include the following: vation of stemness. At variance with this, and very importantly, our results identified a link between basal autophagy decline in d KEY RESOURCES TABLE quiescent muSCs and rhythmic rewiring during aging. Mainte- d CONTACT FOR REAGENT AND RESOURCE SHARING nance of a constitutive autophagy flux in resting muSCs has d EXPERIMENTAL MODELS AND SUBJECT DETAILS been shown to be fundamental for preservation of stemness B Animal models by preventing intracellular proteotoxicity, and age-associated d METHOD DETAILS autophagy decline leads to loss of quiescence and regenerative B FACS sorting and analysis of epidermal and muscle failure (Garcıá-Prat et al., 2016). In this context, the inability of stem cells aged quiescent muSCs to sustain a rhythmic autophagy may B Microarrays be at the basis of their failure to sustain proteostasis. Interest- B Immunofluorescence and immunohistochemistry ingly, loss of the daily rhythmic oscillation of autophagy genes B Microscopy and image analysis occurred as well in the aged liver, which is also predominantly B qRT-PCR a non-proliferative tissue (see accompanying study by Sato B Gene ontology et al., 2017). Hence, an additional important conclusion of these d QUANTIFICATION AND STATISTICAL ANALYSIS results is that although aging has a strong systemic component B Number of replicates used and occurs in all tissues, different types of SCs age differently B Quantification and according to the idiosyncrasies of their tissue of residence. B Statistical significance tests Mechanistically, we do not know the driving forces that B Identification of rhythmic genes change the rhythmic transcriptome of adult SCs as they age. d DATA AND SOFTWARE AVAILABILITY

Figure 7. Induced Circadian Arrhythmia Does Not Recapitulate Physiological Aging of Adult SCs (A) Representative images of intensity levels of gH2ax and p-RPA at ZT16 in 18-week-old wild-type and Bmal1 KO epSCs. (B) Representative images and quantification of intensity levels of oxidised DNA at ZT16 in wild-type and Bmal1-KO epSCs (normalized to wild-type ZT16). (C) Representative images of intensity levels of gH2ax and p-RPA at ZT16, in 27- to 39-week-old wild-type or Per1/2-dKO epSCs. (D and E) Fold-change of mRNA expression levels of the ZT16 subset genes E2f1, Mcm2, Mcm6, Chek1, and Pola2, normalized against B2m at ZT16, in wild-type or Bmal1 KO (D) and Per1/2-dKO (E) epSCs. (F) Autophagy levels. Representative images and percentage of colocalization of LAMP1 and LC3 puncta in wild-type and Bmal1 KO muSCs at ZT4 and ZT16. Note that LAMP1 and LC3 puncta are present in all interstitial muscle cells and in muscle fibers. MuSCs boundaries are indicated (dotted line). Scale bar, 5 mm. Error bars represent ± SEM; N.S., not significant.

690 Cell 170, 678–692, August 10, 2017 SUPPLEMENTAL INFORMATION Cerletti, M., Jang, Y.C., Finley, L.W.S., Haigis, M.C., and Wagers, A.J. (2012). Short-term calorie restriction enhances skeletal muscle stem cell function. Cell Supplemental Information includes seven ﬁgures and three tables and can be Stem Cell 10, 515–519. found with this article online at http://dx.doi.org/10.1016/j.cell.2017.07.035. Chakkalakal, J., and Brack, A. (2012). Extrinsic Regulation of Satellite Cell Function and Muscle Regeneration Capacity during Aging. J Stem Cell Res AUTHOR CONTRIBUTIONS Ther (Suppl 11 ), 001. Chakkalakal, J.V., Jones, K.M., Basson, M.A., and Brack, A.S. (2012). The Conceptualization, G.S., F.O.P., P.M.-C., P.S.-C., and S.A.B.; Investigation, aged niche disrupts muscle stem cell quiescence. Nature 490, 355–360. G.S., F.O.P., E.P., M.J., V.R.-B., and A.C.; Methodology, G.S., F.O.P., Chang, H.-C., and Guarente, L. (2013). SIRT1 mediates central circadian con- J.M.C., and P.-S.W.; Formal Analysis, D.D. and A.S.; Supervision, P.S.-C., trol in the SCN by a mechanism that decays with aging. Cell 153, 1448–1460. P.M.-C., and S.A.B.; Writing, P.M.-C. and S.A.B. Clayton, E., Doupe´ , D.P., Klein, A.M., Winton, D.J., Simons, B.D., and Jones, P.H. (2007). A single type of progenitor cell maintains normal epidermis. Nature ACKNOWLEDGMENTS 446, 185–189.

Research for this project in the lab of S.A.B. was supported by the European Gentleman, R., Carey, V.J., Huber, W., Irizarry, R.A., and Dudoit, S. (2005). Bio- Research Council (ERC), the Spanish Ministry of Economy and Development, informatics and Computational Biology Solutions Using R and Bioconductor and the Institute for Research in Biomedicine (IRB-Barcelona). F.O.P. is sup- (New York, NY: Springer New York). ported by a La Caixa International Ph.D. fellowship, and P-S.W. is supported Durcan, T.M., and Fon, E.A. (2015). The three ’P’s of mitophagy: PARKIN, by a Juan de la Cierva postdoctoral fellowship. Research in the lab of PINK1, and post-translational modifications. Genes Dev. 29, 989–999. P.M.C. is supported by the Spanish Ministry of Economy and Development Eckel-Mahan, K.L., Patel, V.R., de Mateo, S., Orozco-Solis, R., Ceglia, N.J., (SAF2015-67369-R), AFM, E-Rare/ERANET, Fundacio´ La Marato´ TV3, MDA, Sahar, S., Dilag-Penilla, S.A., Dyar, K.A., Baldi, P., and Sassone-Corsi, P. and Convenio CNIC-UPF. We are grateful for the excellent technical support (2013). Reprogramming of the circadian clock by nutritional challenge. Cell from the IRB Functional Genomics Core Facility, the PCB (Scientific Park of 155, 1464–1478. Barcelona) Animal Unit and FACS Unit, the IRB Advanced Digital Microscopy Eklund, A.C., and Szallasi, Z. (2008). Correction of technical bias in clinical mi- Unit, and IRB Biostatistics/Bioinformatics Core Facility, the IRB FACS unit, the croarray data improves concordance with known biological information. PCB (Scientific Park of Barcelona) Animal Unit, the IRB Imaging Unit, the IRB Genome Biol. 9, R26. Histology Unit for technical assistance and support, the UPF/CRG FACS Unit, and the UPF Advanced Light Microscopy Unit (especially to Jaume Boix for Farajnia, S., Michel, S., Deboer, T., vanderLeest, H.T., Houben, T., Rohling, microscopy advice and assistance). IRB Barcelona is the recipient of a Severo J.H., Ramkisoensing, A., Yasenkov, R., and Meijer, J.H. (2012). Evidence for Ochoa Award of Excellence from MINECO (Government of Spain). The neuronal desynchrony in the aged suprachiasmatic nucleus clock. DCESX/UPF is recipient of a ‘‘Marıá de Maeztu’’ Programme for Units of Excel- J. Neurosci. 32, 5891–5899. lence in R&D MDM-2014-0370 (Government of Spain). We thank Veronica Froy, O., and Miskin, R. (2010). Effect of feeding regimens on circadian Raker for manuscript editing. We apologize to the authors whose work was rhythms: implications for aging and longevity. Aging (Albany NY) 2, 7–27. not cited due to space limitations. Gaddameedhi, S., Selby, C.P., Kaufmann, W.K., Smart, R.C., and Sancar, A. (2011). Control of skin cancer by the circadian rhythm. Proc. Natl. Acad. Sci. Received: March 9, 2017 USA 108, 18790–18795. Revised: May 7, 2017 Garcıá-Prat, L., Martıńez-Vicente, M., Perdiguero, E., Ortet, L., Rodrı´guez- Accepted: July 21, 2017 Ubreva, J., Rebollo, E., Ruiz-Bonilla, V., Gutarra, S., Ballestar, E., Serrano, Published: August 10, 2017 A.L., et al. (2016). Autophagy maintains stemness by preventing senescence. Nature 529, 37–42. REFERENCES Gentleman, R.C., Carey, V.J., Bates, D.M., Bolstad, B., Dettling, M., Dudoit, S., Ellis, B., Gautier, L., Ge, Y., Gentry, J., et al. (2004). Bioconductor: open soft- Amaral, N., Vendrell, A., Funaya, C., Idrissi, F.-Z., Maier, M., Kumar, A., ware development for computational biology and bioinformatics. Genome Neurohr, G., Colomina, N., Torres-Rosell, J., Geli, M.-I., and Mendoza, Biol. 5, R80. M. (2016). The Aurora-B-dependent NoCut checkpoint prevents damage of anaphase bridges after DNA replication stress. Nat. Cell Biol. 18, Geyfman, M., Kumar, V., Liu, Q., Ruiz, R., Gordon, W., Espitia, F., Cam, E., 516–526. Millar, S.E., Smyth, P., Ihler, A., et al. (2012). Brain and muscle Arnt-like protein-1 (BMAL1) controls circadian cell proliferation and susceptibility to UVB- Bentzinger, C.F., Wang, Y.X., von Maltzahn, J., Soleimani, V.D., Yin, H., and induced DNA damage in the epidermis. Proc. Natl. Acad. Sci. USA 109, Rudnicki, M.A. (2013). Fibronectin regulates Wnt7a signaling and satellite 11758–11763. cell expansion. Cell Stem Cell 12, 75–87. Goetsch, K.P., Snyman, C., Myburgh, K.H., and Niesler, C.U. (2014). ROCK-2 Bernet, J.D., Doles, J.D., Hall, J.K., Kelly Tanaka, K., Carter, T.A., and Ol- is associated with focal adhesion maturation during myoblast migration. win, B.B. (2014). p38 MAPK signaling underlies a cell-autonomous loss of J. Cell. Biochem. 115, 1299–1307. stem cell self-renewal in skeletal muscle of aged mice. Nat. Med. 20, 265–271. Goodell, M.A., and Rando, T.A. (2015). Stem cells and healthy aging. Science 350, 1199–1204. Boada-Romero, E., Letek, M., Fleischer, A., Pallauf, K., Ramo´ n-Barros, C., and Pimentel-Muin˜ os, F.X. (2013). TMEM59 defines a novel ATG16L1-binding Hughes, M.E., Hogenesch, J.B., and Kornacker, K. (2010). JTK_CYCLE: an motif that promotes local activation of LC3. EMBO J. 32, 566–582. efficient nonparametric algorithm for detecting rhythmic components in genome-scale data sets. J. Biol. Rhythms 25, 372–380. Bonaconsa, M., Malpeli, G., Montaruli, A., Carandente, F., Grassi-Zucconi, G., and Bentivoglio, M. (2014). Differential modulation of clock gene expression in Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D., Antonellis, K.J., the suprachiasmatic nucleus, liver and heart of aged mice. Exp. Gerontol. Scherf, U., and Speed, T.P. (2003). Exploration, normalization, and summaries 55, 70–79. of high density oligonucleotide array probe level data. Biostatistics 4, 249–264. Bu´ a, S., Sotiropoulou, P., Sgarlata, C., Borlado, L.R., Eguren, M., Domıńguez, Ishii, A., and Lo, S.H. (2001). A role of tensin in skeletal-muscle regeneration. O., Ortega, S., Malumbres, M., Blanpain, C., and Me´ ndez, J. (2015). Deregu- Biochem. J. 356, 737–745. lated expression of Cdc6 in the skin facilitates papilloma formation and affects Janich, P., Pascual, G., Merlos-Sua´ rez, A., Batlle, E., Ripperger, J., Albrecht, the hair growth cycle. Cell Cycle 14, 3897–3907. U., Cheng, H.-Y.M., Obrietan, K., Di Croce, L., and Benitah, S.A. (2011). The

Cell 170, 678–692, August 10, 2017 691 circadian molecular clock creates epidermal stem cell heterogeneity. Nature Sato, S., Solanas, G., Peixoto, F., Bee, L., Symeonidi, A., Schmidt, M.S., 480, 209–214. Brenner, C., Masri, S., Benitah, S.A., and Sassone-Corsi, P. (2017). Circadian Janich, P., Toufighi, K., Solanas, G., Luis, N.M., Minkwitz, S., Serrano, L., Leh- reprogramming identifies metabolic pathways of aging. Cell 170, this issue, ner, B., and Benitah, S.A. (2013). Human epidermal stem cell function is regu- 664–677. lated by circadian oscillations. Cell Stem Cell 13, 745–753. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Janich, P., Meng, Q.-J., and Benitah, S.A. (2014). Circadian control of tissue Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. homeostasis and adult stem cells. Curr. Opin. Cell Biol. 31, 8–15. (2012). Fiji: an open-source platform for biological-image analysis. Nat. Koike, N., Yoo, S.-H., Huang, H.-C., Kumar, V., Lee, C., Kim, T.-K., and Taka- Methods 9, 676–682. hashi, J.S. (2012). Transcriptional architecture and chromatin landscape of the Segawa, M., Fukada, S., Yamamoto, Y., Yahagi, H., Kanematsu, M., Sato, M., core circadian clock in mammals. Science 338, 349–354. Ito, T., Uezumi, A., Hayashi, S., Miyagoe-Suzuki, Y., et al. (2008). Suppression Koren, I., Reem, E., and Kimchi, A. (2010). DAP1, a novel substrate of mTOR, of macrophage functions impairs skeletal muscle regeneration with severe negatively regulates autophagy. Curr. Biol. 20, 1093–1098. fibrosis. Exp. Cell Res. 314, 3232–3244. Lipp, H.-P., Litvin, O., Galsworthy, M., Vyssotski, A.L., Zinn, P., Rau, A.E., Neu- Shibukawa, Y., Yamazaki, N., Daimon, E., and Wada, Y. (2013). Rock-depen- hausser-Wespy, F., Wurbel, H., Nitsch, R., and Wolfer, D.P. (2005). Automated dent calponin 3 phosphorylation regulates myoblast fusion. Exp. Cell Res. 319, behavioral analysis of mice using INTELLICAGE: inter-laboratory comparisons 633–648. and validation with exploratory behavior and spatial learning. In Proceedings of Solanas, G., and Benitah, S.A. (2013). Regenerating the skin: a task for the het- Measuring Behavior 2005, L.P.J.J. Noldus, F. Grieco, L.W.S. Loijens, and P.H. erogeneous stem cell pool and surrounding niche. Nat. Rev. Mol. Cell Biol. 14, Zimmermann, eds., pp. 66–69. 737–748. Lo´ pez-Otıń, C., Blasco, M.A., Partridge, L., Serrano, M., and Kroemer, G. (2013). The hallmarks of aging. Cell 153, 1194–1217. Sousa-Victor, P., Gutarra, S., Garcıá-Prat, L., Rodriguez-Ubreva, J., Ortet, L., Ruiz-Bonilla, V., Jardı´, M., Ballestar, E., Gonza´ lez, S., Serrano, A.L., et al. Lukjanenko, L., Jung, M.J., Hegde, N., Perruisseau-Carrier, C., Migliavacca, (2014). Geriatric muscle stem cells switch reversible quiescence into senes- E., Rozo, M., Karaz, S., Jacot, G., Schmidt, M., Li, L., et al. (2016). Loss of fibro- cence. Nature 506, 316–321. nectin from the aged stem cell niche affects the regenerative capacity of skeletal muscle in mice. Nat. Med. 22, 897–905. Sousa-Victor, P., Garcıá-Prat, L., Serrano, A.L., Perdiguero, E., and Mun˜ oz- Mascre´ , G., Dekoninck, S., Drogat, B., Youssef, K.K., Brohee´ , S., Sotiropou- Ca´ noves, P. (2015). Muscle stem cell aging: regulation and rejuvenation. lou, P.A., Simons, B.D., and Blanpain, C. (2012). Distinct contribution of Trends Endocrinol. Metab. 26, 287–296. stem and progenitor cells to epidermal maintenance. Nature 489, 257–262. Stringari, C., Wang, H., Geyfman, M., Crosignani, V., Kumar, V., Takahashi, Masri, S., Papagiannakopoulos, T., Kinouchi, K., Liu, Y., Cervantes, M., Baldi, J.S., Andersen, B., and Gratton, E. (2015). In vivo single-cell detection of meta- P., Jacks, T., and Sassone-Corsi, P. (2016). Lung Adenocarcinoma Distally Re- bolic oscillations in stem cells. Cell Rep. 10, 1–7. wires Hepatic Circadian Homeostasis. Cell 165, 896–909. Suzuki, T., Franchi, L., Toma, C., Ashida, H., Ogawa, M., Yoshikawa, Y., Mim- McCroskery, S., Thomas, M., Maxwell, L., Sharma, M., and Kambadur, R. uro, H., Inohara, N., Sasakawa, C., and Nun˜ ez, G. (2007). Differential regulation (2003). Myostatin negatively regulates satellite cell activation and self-renewal. of caspase-1 activation, pyroptosis, and autophagy via Ipaf and ASC in J. Cell Biol. 162, 1135–1147. Shigella-infected macrophages. PLoS Pathog. 3, e111. Mihaylova, M.M., Sabatini, D.M., and Yilmaz, O¨ .H. (2014). Dietary and meta- Tetteh, P.W., Farin, H.F., and Clevers, H. (2015). Plasticity within stem cell bolic control of stem cell function in physiology and cancer. Cell Stem Cell hierarchies in mammalian epithelia. Trends Cell Biol. 25, 100–108. 14, 292–305. Tierney, M.T., Gromova, A., Sesillo, F.B., Sala, D., Spenle´ , C., Orend, G., and Montarras, D., L’honore´ , A., and Buckingham, M. (2013). Lying low but ready Sacco, A. (2016). Autonomous Extracellular Matrix Remodeling Controls a for action: the quiescent muscle satellite cell. FEBS J. 280, 4036–4050. Progressive Adaptation in Muscle Stem Cell Regenerative Capacity during Nakamura, T.J., Nakamura, W., Yamazaki, S., Kudo, T., Cutler, T., Colwell, Development. Cell Rep. 14, 1940–1952. C.S., and Block, G.D. (2011). Age-related decline in circadian output. J. Neurosci. 31, 10201–10205. Vahidi Ferdousi, L., Rocheteau, P., Chayot, R., Montagne, B., Chaker, Z., Flamant, P., Tajbakhsh, S., and Ricchetti, M. (2014). More efficient repair of Plikus, M.V., Vollmers, C., de la Cruz, D., Chaix, A., Ramos, R., Panda, S., and DNA double-strand breaks in skeletal muscle stem cells compared to their Chuong, C.M. (2013). Local circadian clock gates cell cycle progression of committed progeny. Stem Cell Res. (Amst.) 13(3 Pt A), 492–507. transient amplifying cells during regenerative hair cycling. Proc. Natl. Acad. Sci. USA 110, E2106–E2115. Vermeij, W.P., Dolle´ , M.E.T., Reiling, E., Jaarsma, D., Payan-Gomez, C., Bom- R Development Core Team (2008). R: A language and environment for statis- bardieri, C.R., Wu, H., Roks, A.J.M., Botter, S.M., van der Eerden, B.C., et al. tical computing (R Foundation for Statistical Computing). (2016). Restricted diet delays accelerated ageing and genomic stress in DNA- repair-deficient mice. Nature 537, 427–431. Roenneberg, T., Kuehnle, T., Juda, M., Kantermann, T., Allebrandt, K., Gordijn, M., and Merrow, M. (2007). Epidemiology of the human circadian clock. Sleep Young, A.R.J., Narita, M., Ferreira, M., Kirschner, K., Sadaie, M., Darot, Med. Rev. 11, 429–438. J.F.J., Tavare´ , S., Arakawa, S., Shimizu, S., Watt, F.M., and Narita, M. Rozo, M., Li, L., and Fan, C.-M. (2016). Targeting b1-integrin signaling en- (2009). Autophagy mediates the mitotic senescence transition. Genes Dev. hances regeneration in aged and dystrophic muscle in mice. Nat. Med. 22, 23, 798–803. 889–896. Zheng, B., Albrecht, U., Kaasik, K., Sage, M., Lu, W., Vaishnav, S., Li, Q., Sada, A., Jacob, F., Leung, E., Wang, S., White, B.S., Shalloway, D., and Tum- Sun, Z.S., Eichele, G., Bradley, A., et al. (2001). Nonredundant roles of the bar, T. (2016). Defining the cellular lineage hierarchy in the interfollicular mPer1 and mPer2 genes in the mammalian circadian clock. Cell 105, epidermis of adult skin. Nat. Cell Biol. 18, 619–631. 683–694.

692 Cell 170, 678–692, August 10, 2017 STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Anti-mouse CD34-biotin eBioscience 13-0341-85 APC-conjugated streptavidin BD PharMingen 554067 anti-CD49f-RPE AbD Serotec MCA699PE FITC-conjugated anti-CD45 Biolegend 103113/14 Alexa 647-conjugated anti-F4/80 AbD Serotec MCA497A647 APC-conjugated anti-CD31 Biolegend 102418 PE-conjugated anti-a7-integrin Ablab AB10STMW215 Pe-Cy7-conjugated anti-Sca-1 Biolegend 108113/14 anti-RPA32/RPA2 (phospho Ser4, Ser8) Abcam ab87277 mouse monoclonal anti-phospho-histone Millipore 05-636 H2A.X (Ser139) mouse monoclonal anti-8-hydroxyguanosine Abcam ab62623 mouse monoclonal anti-PCNA (PC10) Santa Cruz Biotechnology sc-56 rabbit polyclonal anti-MCM4 Dr. Juan Mendez (CNIO, Madrid, Spain) Bu´ a et al., 2015 rabbit polyclonal anti-caspase 3 Cell signaling Technology 9661 rabbit polyclonal anti-phosphohistone-3 (Ser10) Millipore 06-570 mouse monoclonal anti-Pax7 DSHB http://dshb.biology.uiowa.edu/PAX7 rat polyclonal anti-LAMP-1 Santa Cruz Biotechnology sc-19992 rabbit polyclonal anti-LC3 Novus Biologicals NB100-2331 rabbit polyclonal anti-53BP1 Abcam ab21083 anti-mouse Alexa Fluor 488 Molecular Probes A21202 anti-mouse Alexa Fluor 568 IgG1 Molecular Probes A21124 anti-rabbit Alexa Fluor 488 Molecular Probes A11008 anti-rat Alexa 568 Molecular Probes A11077 anti-rabbit Alexa Fluor 647 Molecular Probes A31573 BrightVision poly-HRP anti-rabbit IgG Immunologic DPVR-110HRP biotin-free Chemicals, Peptides, and Recombinant Proteins Trypsin 2.5% ThermoFisher Scientific 15090-046 Fetal bovine serum ThermoFisher Scientific 10270-106 Chelex Bio-rad 142-2842 EMEM Calcium free culture medium Lonza BE06-174G Ham’s F10 ThermoFisher Scientific 31550023 Collagenase II Sigma C6885 Collagenase D Roche 11088866001 Dispase II Roche 04942078001 BD Pharm Lyse BD 555899 TRIzol ThermoFisher Scientific A33250 SDS Sigma L3771 Proteinase K Sigma P2308 Dl-Dithiotreitol DTT Sigma D0632 Tris Base Roche 11814273001 Agencourt RNAclean XP Beckman Coulter A63987 DNase I Sigma AMPD1 (Continued on next page)

Cell 170, 678–692.e1–e6, August 10, 2017 e1 Continued REAGENT or RESOURCE SOURCE IDENTIFIER 10% Neutral buffered formalin (NBF) Sigma HT501128-4L Tri-sodium Citrate 2-hydrate Panreac 1316551210 Triton X-100 Sigma T8787 Goat serum ThermoFisher Scientific 16210064 DAPI Sigma D9542-10MG BSA Sigma A7906 Peroxidase-blocking solution Dako S2023 3-30-diaminobenzidine Dako K3468 Hematoxilin Dako S202084 Paraformaldehyde Electron Microscopy Sciences 15710 Mouse on Mouse (M.O.M) Fluorescein Kit Vector laboratories FMK-2201 Fluoromount aqueous mounting medium Sigma F4680-25ML Critical Commercial Assays WTA2 Sigma WTA2-50RXN Purelink Genomic DNA kit ThermoFisher Scientific K182002 GeneChip Mapping 10K 2.0 array kit Affymetrix 511060 Fluorescein M.O.M. immunodetection kit Vector FKM-2201 High Capacity cDNA Reverse Transcription Kit ThermoFisher Scientific 4374966 TaqMan Universal PCR Master Mix. ThermoFisher Scientific 4324018 Deposited Data Raw and analyzed microarray data This paper GEO: GSE84580 Experimental Models: Organisms/Strains Mouse: Bmal1 KO (B6) Dr. Aznar-Benitah’s laboratory N/A Mouse: Per1/2 double KO (B6:129S) Zheng et al., 2001 N/A Mouse: C57BL/6J Wild-type strain The Jackson Laboratory JAX:000664 Software and Algorithms Jonckheere-Terpstra-Kendall (JTK) algorithm Hughes et al., 2010 http://www.openwetware.org/ wiki/HughesLab:JTK_Cycle R v3.2.4 R Development Core Team (2008) http://www.R-project.org Bioconductor Gentleman et al., 2004 www.bioconductor.org Genomatix Genomatix GmbH www.genomatix.de Fiji v2.0.0-rc-14/1.49 g Schindelin et al., 2012 https://imagej.net/Fiji Photoshop CS5 Adobe N/A TMARKER v2.142 NEXUS Personalized Health Technologies N/A Prism 6 software Pad Software, Inc. N/A Python 2.7.11 Python Software Foundation https://www.python.org/ FACSDIVA BD Biosciences N/A Other Rodent Diet Caloric-Restricted Harlan TD.120686 Rodent Control diet for Caloric-Restricted diet Harlan TD.120685 Rodent High-fat diet Harlan TD.06414 Rodent Control diet for High-fat diet Harlan TD.120455 Taqman probe for E2f1 ThermoFisher Scientific Mm00432939_m1 Taqman probe for Check1 ThermoFisher Scientific Mm01176757_m1 Taqman probe for Mcm2 ThermoFisher Scientific Mm00484815_m1 Taqman probe for Mcm6 ThermoFisher Scientific Mm00484848_m1 Taqman probe for Pola2 ThermoFisher Scientific Mm00447142_m1 Taqman probe for B2m ThermoFisher Scientific Mm00437762_m1

e2 Cell 170, 678–692.e1–e6, August 10, 2017 CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulﬁlled by the Lead Contact, Salvador Aznar-Benitah ([email protected]).

EXPERIMENTAL MODELS AND SUBJECT DETAILS

Animal models Bmal1 KO, Per1/Per2 double KO (Zheng et al., 2001) and wild-type mice were bred and aged at the animal facilities of the Barcelona Science Park. All mice were of C57BL/6 background, except for Per1/Per2 double KO that is a mixed B6:129S background. For the experiments specified below, C57BL/6J mice were purchased from Charles River, in strict accordance with the Spanish and Euro- pean Union regulations. The Catalan Government approved the work protocols, in accordance with applicable legislation. Both male and female mice were used in each experiment unless stated otherwise. All mice were fed with the standard chow used at the animal facilities of the Barcelona Science Park, composed of 12.2%/w protein, 2.27%/w fatty acids and 82.08%/w carbohydrates, with 12.22 Kcal/g, unless stated otherwise. For the transcriptomic study of rhythmic oscillations in adult and aged mice, C57BL/6J female mice purchased from Charles River. We used adult (8-week-old) and aged adult (between 102–116-week-old) mice. Four mice were used per each of the six time points (ZT0, ZT4, ZT8, ZT12, ZT16 and ZT20) and per age group. For the transcriptomic study of daily rhythmic oscillations in calorie-restricted aged mice, C57BL/6 aged and non-aged adult mice were fed either ad libitum a control diet (Harlan TD.120685) or with a 30% food reduction as compared to the ad libitum group intake of calorie-restricted diet (Harlan TD.120686) (Harlan Industries, Indianapolis, IN, USA). The control diet was composed of 18.8%/w protein, 7.3% %/w fatty acids and 55.1%/w carbohydrates, with 3.6 Kcal/g. The restricted diet was composed of 32.9%/w protein, 12.7%/w fatty acids and 31.9%/w carbohydrates, with 3.7Kcal/g. Mice were weighed every two weeks. Calorie-restricted mice were housed individually to prevent food competition with cage mates. The experiment lasted for 25 weeks and started after an adaption period of 3 weeks, during which calorie-restricted animals were subjected to a 10% food reduction per week. The age groups were composed of 19–29-week-old and 55–69-week-old mice at the start of the adaption period. Four mice were used per each of the 6 time points (ZT0, ZT4, ZT8, ZT12, ZT16 and ZT20) and per diet group. For sample collection, one mouse per diet group was sacrificed per day, all within the same time point. Both muSCs and epSCs were collected for transcriptome analyses. For the transcriptomic study in adult mice fed a high-fat diet, eight-week-old female C57BL/6J mice were purchased from Charles River and fed ad libitum either a high-fat diet (Harlan TD. 06414) or a control diet (Harlan TD.120455). The control diet was composed of 18.8%/w protein, 7.3%/w fatty acids and 55.1%/w carbohydrates, with 3.6Kcal/g. The high-fat diet was composed of 23.5%/w protein, 34.3%/w fatty acids and 27.3%/w carbohydrates, with 5.1 Kcal/g. Mice were weighed every two weeks. The experiment lasted for 7 weeks, and four mice were used for each of the six time points (ZT0, ZT4, ZT8, ZT12, ZT16 and ZT20) and for each diet group. For sample collection, four mice per diet group were sacrificed each day (all of which belonged to the same time point). For physiological activity monitoring, mouse activity was measured in IntelliCage cages (Lipp et al., 2005). This method allows measurement of basal physiological activity in a social environment. Microchips were inserted in the back of 10 adult (13-week- old) and 10 old (83-week-old) C57BL/6J to monitor individual visits to the drinking corners. Only females were used in order to avoid fighting due to hierarchies established by male mice when pooled during adulthood. Adult and aged mice were caged in separate IntelliCage cages to avoid interference of activity between groups. Mice had continuous access to food and water and were adapted for a week to the cages before the experiment started. Mice were subjected to a 12hr light:12hr dark photoperiod for the first week to use as reference, then they were switched to constant darkness for two weeks, followed by two weeks of 12hr light:12hr dark photoperiod. The results for each mouse and each week were analised separately. In order to avoid interference of the difference in total activity between adult and aged mice (Figure S3A), the assessment of the rhythmicity of the activity was performed using the proportion of visits during each 4hr-bin divided by the total number of visits. To identify rhythmic behavior, the values of activity during each interval for each week were given to the Jonckheere-Terpstra-Kendall (JTK) algorithm (Hughes et al., 2010). Two of the aged mice had to be removed of the study due to health issues. To study the aged-related features in Bmal1 KO and Per1/Per2 double KO, wild-type, Bmal1 KO and Per1/Per2 double KO mice were fed control diet (Harlan TD.120685) composed of 18.8%/w protein, 7.3%/w fatty acids and 55.1%/w carbohydrates, with 3.6 Kcal/g. For Bmal1 KO, four 18-week-old mice were used per genotype; for Per1/Per2 double KO, four 27-39-week-old mice were used per genotype. All mice were sacrificed at ZT16.

METHOD DETAILS

FACS sorting and analysis of epidermal and muscle stem cells After sacriﬁce, 1cm2 skin samples were collected for histology. Skin samples were scraped with a scalpel to remove the hypodermis and then ﬂoated on trypsin 0,25% (GIBCO) dissolved in PBS for approximately 50 min at 37C, allowing the epidermis to be

Cell 170, 678–692.e1–e6, August 10, 2017 e3 separated from the dermis. Trypsin was inactivated with 15% chelated fetal bovine serum (GIBCO) diluted in EMEM medium without calcium (Lonza). Epidermis was mechanically disaggregated and dissociated and sequentially ﬁltered through a 100-mm and a 40-mm ﬁlter. Anti-mouse CD34-biotin (1:50, eBioscience 13-0341-85), streptavidin-APC (1:400, BD PharMingen 554067) and anti-human CD49f-RPE (1:200, AbD Serotec MCA699PE) were used for sorting and FACS analysis of CD34 negative/CD49f bright interfollicular stem cells.

Muscles were mechanically disaggregated and dissociated in Ham’s F10 media containing collagenase II 0.2% (Sigma) with CaCl2 (2.5 mM) at 37cC for 30cmin. Tissue solution was then centrifuged and further dissociated in a collagenase D/Dispase solution (Roche, 1.5 U/ml and 2.4 U/ml respectively) at 37cC for 60cmin and then filtered through a 40-mm filter. Cells were incubated in lysing buffer (BD Pharm Lyse) for 10cmin on ice, re-suspended in PBS with 2.5% goat serum and counted with a haemocytometer. FITC- conjugated anti-CD45 (1:300, Biolegend 103113/14), Alexa 647 anti-F4/80 (1:200, AbD Serotec MCA497A647), APC anti-CD31 (1:300, Biolegend 102418), Pe-Cy7 anti-Sca-1 (1:300, Biolegend 108113/14) and PE-conjugated anti-a7-integrin (1:200, Ablab AB10STMW215) were used to separate macrophages, endothelial cells, mesenchymal cells and muSCs populations, respectively. Sorting gates were strictly defined based on isotype control (fluorescence minus one) stains. To assess viability, cells were stained with DAPI (1 mg ml–1) immediately before sorting. Sorted muSCs were used for RNA extraction, microarray analysis, or cytospin for immunostaining. MuSCs were confirmed as a homogeneous population of Pax7+ cells (data not shown).

Microarrays EpSCs RNA was extracted using TRIzol (Sigma-Aldrich). cDNA library preparation and amplification were performed from 25 ng total RNA using WTA2 (Sigma-Aldrich), with 17 cycles of amplification. MuSCs were lysed for 15 min at 65C in a TrisHCl, SDS, proteinase K and DTT solution (final concentrations were 10mM, 0.5%, 1 mg/ml and 20 mM, respectively). RNA was purified using magnetic beads (RNAClean XP beads, Agencourt). All purified RNA was used for library preparation and amplification. Amplified cDNA libraries were prepared according to manufacturer’s (Sigma-Aldrich) recommendations for WTA2. In order to monitor the amplification, SYBR Green was added to the reaction, which was was stopped at 24 cycles when SYBR Green signal had reached a plateau. cDNA was purified using Purelink (Invitrogen). cDNA (8 mg) from both epSCs and muSCs was subsequently fragmented by DNaseI and biotinylated by terminal transferase obtained from GeneChip Mapping 10Kv2 Assay Kit (Affymetrix). Hybridization mixtures were prepared according to the Gene Atlas protocol (Affymetrix). Each sample target was hybridized to a Mouse Genome 430 PM array. After hybridization for 16 h at 45C, samples were washed and stained in the GeneAtlas Fluidics Station (Affymetrix). Arrays were scanned in a GeneAtlas Imaging Station (Affymetrix). All processing was performed according to manufacturer’s recommendations. CEL files were generated from DAT files using Affymetrix Command Console software. Microarrays processing was performed at IRB Barcelona Functional Genomics Core Facility. Processing of microarray samples was carried out separately for each dataset using R (http://www.R-project.org/)(R Development Core Team, 2008) and Bioconductor (Gentleman et al., 2004). Raw CEL files were normalized using RMA background correction and summarization (Irizarry et al., 2003). Standard quality controls were performed in order to identify abnormal samples (Gentleman et al., 2005) regarding: a) spatial artifacts in the hybridization process (scan images and pseudo-images from probe level models); b) intensity dependences of differences between chips (MvA plots); c) RNA quality (RNA digest plot); d) global intensity levels (boxplot of perfect match log-intensity distributions before and after normalization and RLE plots); and e) anomalous intensity profile compared to the rest of samples (NUSE plots, Principal Component Analyses). A total of 15 samples were excluded across all datasets due to quality issues (from the skin HFD array samples Skin_Control.CT4.4, Skin_Control.CT8.3, Skin_Control.CT8.4, Skin_HFD.CT8.3, Skin_HFD.CT8.4, Skin_Control.CT8.3 Skin_HFD.CT0.4 and Skin_Control.CT12.4; from the skin CR array samples Skin_ND_CT8.4, Skin_CR_CT12.5, Skin_CR_CT4.4 and Skin_Adult_ND_CT8.4; from the muscle CR array samples SC_Adult_ND_CT20.2, SC_Adult_ND_CT20.3, SC_Adult_ND_CT20.4 and SC_Aged_CR_CT20.5). Probesets were annotated using information available on the Affymetrix webpage (https://www.affymetrix.com/analysis/index.affx). Prior to downstream analyses, expression values were corrected for amplification and scanning batches using a linear model in which gender, time point, diet, age, and the pairwise interaction terms between the three last were included as covariates. Correction by metrics (Eklund and Szal- lasi, 2008) was then carried out for the muSCs calorie restriction dataset, and for both muSCs and epSCs high fat diet datasets. Addi- tionally, correction for extraction date was performed for the epSCs high fat diet dataset. Microarrays were analyzed at IRB Barcelona Biostatistics/Bioinformatics Core Facility.

Immunoﬂuorescence and immunohistochemistry Primary antibodies used are rabbit polyclonal anti-RPA32/RPA2 (phospho Ser4, Ser8) (1:100, Abcam ab87277), mouse monoclonal anti-phospho-histone H2A.X (Ser139) (1:200, Millipore 05-636), mouse monoclonal anti-8-hydroxyguanosine (1:2000, Abcam ab62623), mouse monoclonal anti-PCNA (PC10) (1:200, Santa Cruz sc-56), rabbit polyclonal anti-MCM4 (1:500, kindly provided by Dr. Juan Mendez, CNIO, Madrid, Spain (Bu´ a et al., 2015)), rabbit polyclonal anti-caspase 3 (1:500, Cell Signaling Technology, 9661S), rabbit polyclonal anti-phosphohistone-3 (Ser10) (1:500, 06-570 EMD, Millipore), mouse monoclonal anti-Pax7 (1:20, DSHB), rat polyclonal anti-LAMP-1 (1:200, Santa Cruz Biotechnology sc-19992), rabbit polyclonal anti-LC3 (1:100, Novus Biologicals NB100-2331) and rabbit polyclonal anti-53BP1 (1:200, Abcam ab21083).

e4 Cell 170, 678–692.e1–e6, August 10, 2017 Mouse back skin was fixed in 10% NBF for 3 h at room temperature and then processed for embedding in paraffin blocks. Antigen retrieval was performed for 20 min at 97C with citrate (pH 6) on 3-micron tissue sections. Sections were permeabilised for 20 min with a 0.05% Triton X-100 in PBS solution, and blocked with a 10% goat serum in PBS solution for 1 h at room temperature. Primary antibody incubation was done overnight at 4C. Secondary antibody incubation was done at room temperature for 2 h. Secondary antibodies used are anti-mouse Alexa Fluor 488, anti-rabbit Alexa Fluor 647 (1:500, Molecular Probes). Nuclei were stained with DAPI (Invitrogen). All washes were with PBS except when staining for phospho-histone H2A.X and RPA32/RPA2, in which case washes were done with PBS-T 0.4% BSA. To stain with MCM4, phosphohistone-3 or caspase3, endogenous peroxidase was quenched after antigen retrieval with a 10 min incubation with peroxidase-blocking solution (Dako REAL S2023). After primary antibody incubation (for 60, 90 and 120 min, respectively) at room temperature, a BrightVision poly-HRP anti-rabbit IgG biotin-free (Immunologic, DPVR- 110HRP) was used. Antigen–antibody complexes were revealed with 3-30-diaminobenzidine (K3468, Dako). Sections were counter- stained with hematoxylin (Dako, S202084). Tibialis anterior muscles were frozen in isopentane cooled with liquid nitrogen and stored at 80C until analysis, at which point 10-mm sections were cut and immunostained. Cryosections were rinsed once with PBS and fixed in 2%–4% paraformaldehyde for 10 min at room temperature. Sections were rinsed three times for 5 min with PBS, permeabilized with 0.5% Triton X-100 in PBS for 10 min at room temperature followed again by rinsing them three times with PBS. Sections were blocked in PBS supplemented with 5% goat serum, 5% BSA and 1:40 mouse on mouse blocking reagent (Vector labs) for 1 h at room temperature. Incubation with primary antibodies was carried out overnight at 4C. The next day, sections were rinsed three times with PBS followed by incubation with secondary antibodies coupled to Alexa 488, Alexa-568 or Alexa-647 fluo- rochromes (1:500, Molecular Probes) for 1 h at room temperature. Sections were rinsed again with PBS and nuclei were counter- stained with DAPI (1 mg ml–1) in PBS before mounting with Fluoromount (Sigma). Slides were stored at 4C until analysis. For all the stainings, at least four biological replicates were used.

Microscopy and image analysis Fluorescence pictures from four biological replicates were acquired using either a Leica TCS SP5 or a Zeiss LSM780 confocal microscope (63 3 /1.40 oil objective at 512 3 512 or 1024 3 1024 pixel resolution) and processed using the Fiji v2.0.0-rc-14/ 1.49 g software (ImageJ) (Schindelin et al., 2012). PCNA-stained sections (from the calorie restriction study) and MCM4-stained sections were imaged using a NanoZoomer 2.0HT (Hamamatsu, Japan) (20 3 objective at 0.46 mm/pixel). Caspase 3–stained sections were imaged with a upright Nikon E800 microscope (60 3 /1.4 oil objective at 2070 3 1548 pixel resolution). qRT-PCR RNA was extracted using TRIzol and converted into cDNA by reverse transcriptase using the high-capacity cDNA reverse transcription kit (AppliedBiosystems). Gene expression was quantiﬁed by quantitative real-time PCR using TaqMan Master Mix and the following TaqMan probes (ThermoFisher Scientiﬁc): E2f1 (Mm00432939_m1), Chek1 mouse (Mm01176757_m1), Mcm2 mouse (Mm00484815_m1), Mcm6 mouse (Mm00484848_m1) and Pola2 mouse (Mm00447142_m1), and (for normalization) B2m mouse (Mm00437762_m1). A LightCycler 480 instrument (Roche) was used. Four biological replicates were used in each assay.

Gene ontology GO analyses was done using Genomatix (https://www.genomatix.de/), using the Gene Ranker package, with a p value of 0.01. The categories belonging to ‘‘biological processes’’ (BP) were used to perform GO comparisons. Heatmaps of the GO genes were performed by extracting the genes from the corresponding GO terms, along with the corresponding expression values from the arrays. For each gene, the mean expression value of all replicates per time point was calculated and normalized on a [–1,1] scale. Then, genes were ordered according to their lag obtained from the JTK output ﬁle. Heatmaps were plotted using R version 3.2.4.

QUANTIFICATION AND STATISTICAL ANALYSIS

Number of replicates used The number of biological and/or technical replicates for each experiment is stated in the ‘‘Methods details’’ section and the ﬁgure legends.

Quantiﬁcation PCNA and phosphohistone 3–positive nuclei from epidermis of adult and aged mice were manually counted. MuSC LC3–positive autophagosomes, and LAMP1-positive lysosomes were manually counted, and the co-localization of both vesicles was determined on digital images with respect to the total cellular area (background was reduced with Photoshop CS5 [Adobe], with brightness and contrast adjustments applied to the entire image). Phospho-histone H2A.X, RPA32/RPA2 and 8-hydroxyguanosine signal intensity levels were determined using Fiji v2.0.0-rc-14/1.49 g software (ImageJ). PCNA-stained sections (from the calorie restriction study) and MCM4-stained sections were quantiﬁed using TMARKER v2.142 software (NEXUS Personalized Health Technologies).

Cell 170, 678–692.e1–e6, August 10, 2017 e5 Distance between hair follicles and size of the cornified layer were quantified by measuring these parameters on calibrated images on Fiji v2.0.0-rc-14/1.49 g software (ImageJ). To compare the muscle SC abundance between mice, we calculated the number of SCs per mg of tissue by multiplying the SC proportion generated by FACS to the total number of cells counted following digestion over the initial weight of isolated skeletal muscle (to normalize the efficiency of the enzymatic digestion). Muscle SC proportion was calculated over the living cells (gated with DAPI) as shown in gating scheme. Total number of mononuclear cells after tissue enzymatic digestion was counted using a haemocytometer.

Statistical signiﬁcance tests Statistical analysis was performed using Prism 6 software (Pad Software). Experimental groups were compared by unpaired t test with Welch’s correction. The ZT16 time point of the rhythmic genes in the different conditions was compared by two-sided t test (scipy.stats package). *, p % 0.05, **, p % 0.01, ***, p % 0.001, ****, p % 0.0001. Error bars represent ± standard error of the mean (SEM), unless otherwise stated in the ﬁgure legends.

Identification of rhythmic genes To identify genes with a daily rhythmic expression from the different sets of gene expression data, the Jonckheere-Terpstra-Kendall (JTK) algorithm was used (Hughes et al., 2010). For all array sets except for the calorie restriction dataset, a permutation-based p value of less than 0.05 was considered significant. Applying a 0.05 threshold to the calorie restriction array identified non-rhythmic genes as rhythmic. Therefore, the threshold was adjusted to the average p value of the known rhythmic genes—in this case, 0.005. The expression levels for all rhythmic genes at each time point were calculated as the average expression of the biological replicates. The error bars represent standard deviation. Amplitude and phase estimations of oscillating genes were extracted from the JTK algorithm. Kernel-density estimations (KDE) were computed and plotted for the oscillation amplitudes from each array using Python Seaborn library. Phase estimations were performed by computing the corresponding histogram from each array, and by fitting the parametric distribution to the data, using the Python Seaborn library. For the gene length graph, early and late phase genes were identified according to their phase estimated by JTK. Early phase genes were defined as those phasing between ZT20 and ZT4, and late phase genes, those between ZT8 and ZT16. The gene lengths were extracted from the UCSC annotation, and a KDE for each category was computed, along with the estimation for the whole mouse gene set. Plotting was done with the Seaborn library (version 0.7.1) in Python (2.7.11 within Anaconda 4.0.0; 64-bit).

DATA AND SOFTWARE AVAILABILITY

All array expression data ﬁles have been uploaded to the NCBI GEO database at GEO: GSE84580.

e6 Cell 170, 678–692.e1–e6, August 10, 2017 Supplemental Figures

A All events Keratinocytes SSC-A FSC-W Singlets

Keratinocytes

FSC-A FSC-A Singlets Alive cells

HF SCs CD45 CD34 Dapi epSCs Alive cells

FSC-A CD49f (a6Integrin) CD49f (a6Integrin) B All events 250 5 5 5 10 10 (x 1.000) (x 1.000) Macrophages 10 Muscle stem cells 4 4 4 10 10 10 3 3 3 10 10 10 CD45 DAPI SSC-A FSC-H Endothelial cells 2 2 2 a7-integrin 50 100 150 200 50 100 150 200 250 010 010 2 2

-10 Mesenchymal cells -10 -100 0 10 4 4 5 2 2 3 4 5 50 100 150 200 250 50 100 150 200 250 50 100 150 200 250 -10 010 10 -10 010 10 10 10 (x 1.000) (x 1.000) (x 1.000) FSC-A FSC-A FSC-A F4/80 CD31 C

rhythmic Adult Adult rhythmic

Figure S1. FACS Isolation Strategy and Gene Expression Controls, Related to Figures 1 and 2 (A) Representative examples are shown of the FACS strategy and gating scheme for epSC and hair follicle SC isolation from mice; and FACS analysis of epSCs verifying that no CD49f-positive cells were also CD45-positive in epidermal preparations.

(legend continued on next page) (B) Representative example of the FACS strategy and gating scheme for muSC isolation from mice. All cellular populations present in skeletal muscle are indicated. (C) Expression levels in adult epSCs of genes exclusively rhythmic in aged epSCs (left) and expression levels in aged epSCs of genes exclusively rhythmic in adult epSCs (right). (D) Expression levels over time of genes controlled by the secondary circadian clock loop in adult and aged adult epSCs. Error bars represent SD. ZT24 = ZT0. A 6000 **** s sit

Vi 4000 of

2000 Number

0 Adult Aged B Adult Aged 0.4 0.3

s L:D t L:D 0 isi

0.3 V D:D 1 0.2 D:D 1 D:D 2 D:D 2 0.2 L:D 1 L:D 1 0.1L:D 2 0.1 Proportion of L:D 2 Proportion of Visits

0.0 0.0

ZT0-4 ZT4-8 ZT0-4 ZT4-8 ZT8-12 ZT8-12 ZT12-16 ZT16-20 ZT20-24 ZT12-16 ZT16-20 ZT20-24

C Adult Aged

L:D 0 D:D 1 0.4 0.4 isits

0.3 V 0.3

0.2 0.2

0.1 0.1 Proportion of Proportion of Visits Proportion of 0.0 0.0

ZT0-4 ZT4-8 ZT0-4 ZT4-8 ZT8-12 ZT8-12 ZT12-16ZT16-20ZT20-24 ZT12-16ZT16-20ZT20-24

D:D 2 L:D 1 0.4 0.4

Adult s t 0.3 Aged Visi 0.3 0.2 0.2

0.1 0.1 Proportion of Visits Proportion of 0.0 0.0

ZT0-4 ZT4-8 ZT0-4 ZT4-8 ZT8-12 ZT8-12 ZT12-16ZT16-20ZT20-24 ZT12-16ZT16-20ZT20-24 D L:D 2 0.4 JTK adj. p-val Adult Aged 0.3 L:D 0 3.82E-11 9.94E-12 0.2 D:D 1 9.25E-19 2.47E-10 5.28E-15 8.40E-09 0.1 D:D 2

Proportion of Visits L:D 1 1.97E-12 5.14E-10 0.0 L:D 2 4.20E-14 5.14E-10

ZT0-4 ZT4-8 ZT8-12 ZT12-16ZT16-20ZT20-24

(legend on next page) Figure S2. Aged Mice Retain Circadian Physiological Activity, Related to Figure 1 (A) Total number of visits to the drinking corners of adult and aged mice during experiment (5 weeks). (B) Proportion of visits for each week of experiment comparing adult and aged mice. (C) Proportion of visits for adult and aged mice during each week of the experiment. Bins of 4 hr were accumulated and plotted together. Error bars represent ± SEM. (D) Adjusted p values obtained when applying the JTK algorithm on the values of number of visits for every week and age group. L:D, 12hr light:12hr dark photoperiod. D:D, constant darkness. ****p % 0.0001. AB Adult rhythmic genes with phase 0-6 P-value carbohydrate metabolic process 1.79E-05 Distribution of rhythmic genes Distribution of rhythmic genes cytokinesis 2.41E-05 Adult Aged oxidation-reduction process 4.17E-04 regulation of BMP signaling pathway 1.76E-03 response to oxidative stress 5.49E-03 wound healing 6.68E-03 mitochondrion organization 7.73E-03

Gene density Gene density C

Aged rhythmic genes with phase 0-6 P-value cytokinesis 1.73E-07 Day Night Day Night ZT ZT mitochondrial genome maintenance 1.11E-05 mitochondrion organization 2.19E-03 negative regulation of G0 to G1 transition 6.69E-03 D

** * ** * ***

* *** *** * **

** * ** ** *

**** * *

* * * * **

E * * ZT16 subset rhythmic genes P-value * * DNA replication 8.89E-28 cell cycle 2.57E-19 cell division 8.31E-10 DNA repair 1.06E-09 double strand break repair 1.72E-08 DNA conformation change 1.07E-07 DNA damage checkpoint 1.43E-07 replication fork processing 9.99E-05 **

(legend on next page) Figure S3. Aged epSCs Delay the Timing of DNA Replication and Show Signs of Replicative Stress, Related to Figure 3 (A) Phase distribution of adult and aged rhythmic genes in epSCs, and selection of rhythmic genes with phase ZT0–ZT6 (‘‘day’’ genes) and ZT12–ZT18 (‘‘night’’ genes). (B and C) GO analysis performed separately on rhythmic genes with phase ZT0–ZT6 (‘‘day’’ set) in adult (B) or aged (C) epSCs. (D) Expression levels across time points of ZT16-subset genes in adult and aged epSCs. (E) GO analysis performed separately on ZT16-subset genes. *p % 0.05, **p % 0.01, ***p % 0.001. Error bars represent SD. ZT24 = ZT0. A

B p-H3 20 Adult Aged 15 **

% of pos cells 5

0 4 8 6 0 T0 T T Z Z Z ZT12 ZT1 ZT2 ZT24

C Caspase 3 staining

ZT4 ZT8

Adult

Positive control

Aged

BF Pax7 yH2Ax DAPI D yH2Ax BF Pax7 53BP1 DAPI 53BP1

Adult Adult

Myonucleus

Aged muSC Aged

Aged Myonucleus

JTK=n.s.

(legend on next page) Figure S4. Controls Regarding Oxidative Metabolism Timing, Mitosis, and Apoptosis in epSCs and Replicative Stress in muSCs, Related to Figures 2 and 3 (A) Expression levels over time of oxidative metabolism representative genes in adult and aged epSCs. Error bars represent SD. (B) Quantiﬁcation of phosphohistone H3 (p-H3)-positive adult and aged basal cells across time points ZT0, ZT4, ZT8, ZT12, ZT16 and ZT20. **p % 0.01. Error bars represent ± SEM. ZT24 = ZT0. (C) Representative images of Caspase 3 staining in adult and aged epSCs, at ZT4 and ZT8, and C57BL/6J mouse ovary tissue as a positive control. Scale bar, 15 mm. (D) Representative images of yH2Ax and 53BP1 levels in muSCs (Pax7-positive), with myonucleus as a positive control. MuSC boundaries are indicated (dotted lines). Bright ﬁeld (BF). Scale bar, 5 mm. (E) Expression levels of Xpa and Ercc4 genes over time in adult and aged muSCs. Xpa and Ercc4 were no longer detected as rhythmic in aged muSCs. Error bars represent SD. A F EpSC 140 Normal Diet CR Diet 120

100

Weight (%) Weight 80

60 1 3 5 7 9 11 13 15 17 19 21 23 Time (weeks)

B Cornified layer thickness 15

10 **** **** 5 G MuSC micrometer 0

N Diet Young CR diet C Distance between HF 120 100 *** **** 80

micrometer 60

YoungN Diet CR Diet D Expression level of exclusively CR Diet rhythmic genes in Normal Diet 14 H *** *** 12 ***

10 8 6

Log2(expression) 4 *** 2 *** ** *** *** ZT0 ZT4 ZT8 ZT12 ZT16 ZT20 Expression level of exclusively Normal Diet rhythmic genes in CR Diet 14 12 * *** *** 10 * *** 8 6

Log2(expression) 4 2 ZT0 ZT4 ZT8 ZT12 ZT16 ZT20 ** ** *** *** E ** Expression level of exclusively CR Diet rhythmic genes in Normal Diet 14 12

10 * ** * 8 * 6

Log2(expression) 4 2 ZT0 ZT4 ZT8 ZT12 ZT16 ZT20 *** Expression level of exclusively Normal Diet *** ** rhythmic genes in CR Diet * 14 12

10 8 * *** *** 6

Log2(expression) 4 * 2 ZT0 ZT4 ZT8 ZT12 ZT16 ZT20

(legend on next page) Figure S5. CR Partially Rescues the Differences in Function Segregation along the Day Observed between Adult and Aged muSCs; Representative Images of Quantification of Autophagy Levels in Figure 2F, Related to Figures 2 and 5 (A–C) Phase distribution of rhythmic genes in muSCs from adult (A) and aged (B) normal diet-fed mice, and aged CR-fed mice (C). GO analysis performed separately on rhythmic genes with phase ZT0-ZT6 (‘‘day’’ genes) or those with phase ZT12-ZT18 (‘‘night’’ genes) in adult (A), aged (B), and calorie-restricted aged (C) muSCs. (D) Representative images of LC3 and LAMP1 levels and their colocalization in adult and aged muSCs (Pax7-positive), at ZT4 and ZT16. MuSCs boundaries are indicated (dotted lines). Bright field (BF). Scale bar, 5 mm. See quantification in Figure 2F. ABAdult Aged

Gene density

Day ZT Night Day ZT Night Adult rhythmic genes with phase 0-6 P-value Aged rhythmic genes with phase 0-6 P-value cell migration 9.88E-06 locomotion 1.04E-08 regulation of transcription, DNA templated 2.09E-05 small GTPase mediated signal transduction 4.66E-05 cellular lipid metabolic process 2.21E-05 regulation of immune effector process 4.90E-05 skeletal muscle tissue regeneration 9.37E-05 wound healing 7.42E-05 activation of NF-kappaB-inducing kinase activity 9.10E-04 circadian rhythm 4.70E-04 Wnt signaling pathway 5.39E-03 response to transforming growth factor beta 5.04E-04 circadian regulation of gene expression 5.58E-03 mitochondrion organization 8.36E-04 response to nutrient levels 5.76E-03 mitochondrial DNA repair 1.31E-03 response to nutrient levels 1.35E-03 Adult rhythmic genes with phase 12-18 P-value negative regulation of cellular response to hypoxia 3.44E-03 small GTPase mediated signal transduction 1.97E-08 transcription, DNA-templated 5.67E-08 Aged rhythmic genes with phase 12-18 P-value negative regulation of Wnt signaling pathway 3.44E-05 transcription, DNA-templated 3.01E-07 locomotion 3.58E-04 Ras protein signal transduction 2.87E-05 lipid metabolic process 3.44E-03 locomotion 4.17E-05 regulation of mitochondrion organization 4.03E-03 small GTPase mediated signal transduction 2.34E-04 skeletal muscle organ development 4.48E-03 cytokine production 3.66E-04 negative regulation of TOR signaling 6.03E-03 oxidation-reduction process 8.63E-04 immune system process 6.86E-03 lipid metabolic process 4.14E-03 cellular response to insulin stimulus 7.26E-03 positive regulation of autophagy 7.58E-03 C CR Diet

D BF Pax7 DAPI LC3 DAPI LAMP1 DAPI MERGE Day ZT Night

CR Diet rhythmic genes with phase 0-6 P-value ZT4 transcription, DNA-templated 2.44E-05 Adult circadian rhythm 1.75E-04 immune system development 6.37E-04 locomotion 6.72E-04 lipid biosynthetic process 7.68E-04 negative regulation of interleukin-10 production 1.85E-03 ZT4 cellular response to amino acid starvation 2.35E-03 Aged negative regulation of translation 2.42E-03 TOR signaling 3.97E-03

CR Diet rhythmic genes with phase 12-18 P-value locomotion 1.73E-06 regulation of transcription, DNA-templated 1.30E-05 ZT16 NADP metabolic process 5.97E-04 Adult insulin receptor signaling pathway 7.96E-04 immune system process 1.97E-03 response to hypoxia 4.11E-03 small GTPase mediated signal transduction 4.37E-03 response to decreased oxygen levels 4.99E-03 ZT16 Aged intracellular lipid transport 6.00E-03 negative regulation of mitochondrion organization 8.18E-03 response to DNA integrity checkpoint signaling 9.84E-03

(legend on next page) Figure S6. Controls for the CR Protocol, Related to Figures 4 and 5 (A) Percentage of weight relative to week 1 of mice fed either a normal diet or a CR diet. (B and C) Quantiﬁcation of the corniﬁed layer thickness and distance between hair follicles from adult mice (fed a normal diet) or aged mice fed a normal (N) or CR diet. Error bars represent ± SEM. (D and E) Gene expression levels in normal-diet, aged epSCs (D) and muSCs (E) showing genes exclusively rhythmic in CR-diet, aged epSCs (D) and muSCs (E); and in CR-diet, aged epSCs (D) and muSCs (E) showing genes exclusively rhythmic in normal-diet, aged epSCs (D) and muSCs (E). (F and G) Expression levels across time points of core clock genes (Arntl, Npas2, Per2, Per3, Nr1d1, Cry1 and Cry2) in aged epSCs (F) and in aged muSCs (G) from mice fed a normal diet or a CR diet. (H) Expression levels across time points of ZT16-subset genes in aged epSCs from mice fed either a normal or a CR diet. Error bars represent SD. *p % 0.05; **p % 0.01; ***p % 0.001; ****p % 0.0001. ZT24 = ZT0. A D Rhythmic genes in MuSC F ZT16 JTK p-val < 0.05 yH2ax

130 Normal Diet Normal Diet High Fat Diet High Fat Diet 120 ) ND

110 22832283 927927 27342734 ght (%

Wei 100

90 High Fat Diet rhythmic genes P-value HFD 1 3 5 mitochondrion organization Time (weeks) 1.17E-09 response to toxic stress 2.93E-05 p-RPA32/RPA2 cellular response to oxidative stress 6.29E-05 oxidative phosphorylation 7.58E-04 ribosome biogenesis 1.33E-03 response to ROS 2.57E-03 ND B autophagy 3.39E-03 Rhythmic genes in EpSC interferon-gamma production 4.24E-03 TOR signaling 6.83E-03 JTK p-val < 0.05 Common rhythmic genes P-value HFD Normal Diet High Fat Diet circadian rhythm 1.57E-06 skeletal muscle tissue development 2.32E-06 8OHdG fatty acid metabolic process 6.88E-05 15811581 795 3009 regulation of inflammatory response 1.17E-04 triglyceride biosynthetic process 2.18E-03 wound healing 3.94E-03 ND regulation of DNA damage checkpoint 5.67E-03 High Fat Diet rhythmic genes P-value E Expression level of exclusively HF Diet rhythmic genes in Normal Diet fatty acid oxidation 1.37E-05 14 response to oxidative stress 1.23E-05 12 mitochondrion organization 9.79E-05 10 HFD regulation of translation 2.43E-04 8 6 ZT4 Common rhythmic genes P-value Log2(expression) 4 8OHdG DNA replication 6.87E-26 2 chromosome segregation 3.50E-17 ZT0 ZT4 ZT8 ZT12 ZT16 ZT20 DNA repair 8.30E-15 Expression level of exclusively Normal Diet rhythmic genes in HF Diet 14 circadian rhythm 8.68E-15 12 ND DNA integrity checkpoint 1.25E-05 epithelium development 4.73E-03 10 8 6

Log2(expression) 4 2 Expression level of exclusively HF Diet HFD C ZT0 ZT4 ZT8 ZT12 ZT16 ZT20 rhythmic genes in Normal Diet 14 G BF Pax7 DAPI LC3 DAPI LAMP1 DAPI MERGE 12

10 8 ZT4 6 ND

Log2(expression) 4 2 ZT0 ZT4 ZT8 ZT12 ZT16 ZT20

Expression level of exclusively Normal Diet ZT4 rhythmic genes in HF Diet HFD 14 12

10 8 ZT16 6 ND

Log2(expression) 4 2 ZT0 ZT4 ZT8 ZT12 ZT16 ZT20

ZT16 HFD

(legend on next page) Figure S7. HFD Induces a Reprogramming of the Daily Rhythmic Expression of Genes in epSCs and muSCs Independently of Obesity, Related to Figure 6 (A) Percentage of weight relative to week 1 of adult mice fed either a normal diet or a high-fat (HF) diet. (B) Number of rhythmic genes in adult epSCs from mice fed a normal diet (2,376 genes) or HF diet (3,804 genes), sorted at time points ZT0, ZT4, ZT8, ZT12, ZT16 and ZT20. RNA was extracted and processed for microarrays. Rhythmic genes were determined based on the JTK algorithm (p % 0.05). GO analysis is shown for genes exclusively rhythmic in HF-diet, adult epSCs (top), or for rhythmic genes common to both normal-diet and HF-diet adult epSCs. (C) Gene expression levels from normal-diet, adult epSCs showing rhythmic genes exclusively in HF-diet, adult epSCs (top); and from HF-diet, adult epSCs showing genes exclusively rhythmic in normal-diet, adult epSCs (below). (D) Number of rhythmic genes in muSCs from adult mice fed a normal diet (3,210 genes) or a high-fat (HF) diet (3,661 genes), sorted at time points ZT0, ZT4, ZT8, ZT12, ZT16 and ZT20. RNA was extracted and processed for microarrays. Rhythmic genes were determined based on the JTK algorithm (p % 0.05). GO analyses are shown for genes exclusively oscillating in HF-diet, adult muSCs (top), and rhythmic genes common between normal-diet and HF-diet, adult muSCs. (E) Gene expression levels in normal-diet, adult muSCs showing genes exclusively rhythmic in HF-diet, adult muSCs (top); and in HFD, adult muSCs showing genes exclusively rhythmic in normal-diet, adult muSCs (bottom). (F) Representative images of intensity levels of yH2ax and p-RPA, at ZT16, and oxidised DNA at ZT4 and ZT16 from adult mice fed with a normal diet or HF diet. Scale bar, 5 mm. (G) Representative images of LAMP1 and LC3 puncta in normal diet and High Fat (HF) diet muSCs at ZT4 and ZT16. See quantiﬁcation in Figure 6F. Scale bar, 5 mm.