Boosting ATM activity alleviates aging and extends lifespan in a mouse model of progeria

DNA damage accumulates with age (Lombard et al., 2005). However, whether and how robust DNA repair machinery promotes longevity is elusive. Here, we demonstrate that ATM-centered DNA damage response (DDR) progressively declines with senescence and age, while low dose of chloroquine (CQ) activates ATM, promotes DNA damage clearance, rescues age-related metabolic shift, and prolongs replicative lifespan. Molecularly, ATM phosphorylates SIRT6 deacetylase and thus prevents MDM2-mediated ubiquitination and proteasomal degradation. Extra copies of Sirt6 extend lifespan in Atm-/- mice, with restored metabolic homeostasis. Moreover, the treatment with CQ remarkably extends lifespan of Caenorhabditis elegans, but not the ATM-1 mutants. In a progeria mouse model with low DNA repair capacity, long-term administration of CQ ameliorates premature aging features and extends lifespan. Thus, our data highlights a pro-longevity role of ATM, for the first time establishing direct causal links between robust DNA repair machinery and longevity, and providing therapeutic strategy for progeria and age-related metabolic diseases.


Introduction
A variety of metabolic insults can generate DNA lesions in mammalian cells, which, if incorrectly repaired, can lead to somatic mutations and cell transformation (Vijg, 2014). If unrepaired, such lesions can accumulate and constantly activate the DNA damage response (DDR), a unique feature and mechanism of senescence (Halliwell and Whiteman, 2004;Tanaka et al., 2006). Ataxia telangiectasia mutated (ATM), a serine/threonine protein kinase, is one of the key regulators of DDR (Guleria and Chandna, 2016). Upon DNA damage, self-activated ATM phosphorylates downstream transducers and effectors, promoting DNA repair (Bakkenist and Kastan, 2003;Paull, 2015). H2AX is a well-documented phosphorylation target of ATM; phosphorylated H2AX at S139 (gH2AX) is widely recognized as a hallmark of DNA damage (Burma et al., 2001). Accompanied by decline in DNA repair function, gH2AX-enriched DNA damage foci accumulate in senescent cells and in tissues from aged animals (Maslov and Vijg, 2009), supporting causal links between defective DDR and aging. In human fibroblasts, a dramatic decline of homologous recombination (HR) efficiency, attributable to defective recruitment of Rad51, has been observed . Similar defects in HR also have been observed in Hutchinson-Gilford progeria syndrome (HGPS), which is predominantly caused by a LMNA C1024T mutation .
In addition to DNA damage accumulation, inherited loss-of-function mutations in essential components of DNA repair machinery accelerate aging in humans and mice (Hoeijmakers, 2009b). Patients suffering from ataxia telangiectasia (A-T) develop prominent aging features in their second decades (Boder and Sedgwick, 1958;Shiloh and Lederman, 2017). Werner syndrome, Bloom's syndrome and Rothmund-Thomson syndrome are all progeria syndromes caused by mutations of genes that directly regulate DNA repair (Balajee et al., 1999;Cooper et al., 2000;Lebel et al., 1999;Li and Comai, 2000). Homozygous disruption of Atm in mice causes many premature aging features of A-T, such as growth retardation, infertility, neurodegeneration, immunodeficiency and cancer predisposition (Barlow et al., 1996). Mouse models deficient in DNA repair factors, including DNA-PKcs, Ku70, Ku80, DNA ligase IV, Artemis or Ercc1 etc., phenocopy premature aging features (Hasty, 2005;Hoeijmakers, 2009a), supporting the suggestion that defects in DNA repair accelerate aging. However, whether and how robust DNA repair machinery promotes longevity is poorly understood.
Metabolic disturbance is another antagonistic hallmark of aging (Ló pez- Otín et al., 2013). Although DNA repair deficiency is implicated in aging and age-related diseases including metabolic disorders (Ló pez-Otín et al., 2016;Shimizu et al., 2014), the mechanistic link between decreased DNA repair machinery and metabolic reprogramming during aging is poorly understood. Notably, in response to oxidative stress, ATM phosphorylates Hsp27, shifting glucose metabolism from glycolysis to the pentose phosphate pathway (PPP) (Cosentino et al., 2011;Krüger and Ralser, 2011). eLife digest As cells live and divide, their genetic material gets damaged. The DNA damage response is a network of proteins that monitor, recognize and fix the damage, which is also called DNA lesions. For example, an enzyme called ATM senses when DNA is broken and then begins a process that will get it repaired, while another enzyme known as SIRT6 participates in the actual mending process.
When organisms get older, the repair machinery becomes less efficient, and the number of DNA lesions and errors increases. The accumulation of DNA damage may cause the 'symptoms' of old age -from cancer, to wrinkles and the slowing down of the body's chemical processes. In fact, individuals with defective ATMs (who thus struggle to repair their DNA) age abnormally fast; conversely, SIRT6 promotes longevity. If declining repair mechanisms cause aging, would boosting the DNA damage response slow down this process?
Chloroquine is a drug used to combat malaria, but it can also enhance the activity of ATM without damaging DNA. Qian, Liu et al. show that chloroquine helps cells remove broken DNA and keep dividing for longer. In animals, the drug increases the lifespan of worms and prolongs the lives of mice who have mutations that make them age quicker.
Qian, Liu et al. also demonstrate that ATM works by chemically altering the pro-longevity enzyme SIRT6. These changes make SIRT6 more stable, and keep it safe from cellular processes that destroy it. In addition, mice that are genetically engineered to lack ATM can survive for longer if they also produce extra SIRT6. These experiments show that enhancing the DNA damage response can slow down aging, thus linking the DNA repair machinery to longevity.
Progeria is a group of rare genetic conditions with inefficient DNA repair; people with progeria age fast and die young. The results by Qian, Liu et al., if confirmed in humans, could provide a new way of treating these diseases.
Inactivating ATM enhances glucose and glutamine consumption by inhibiting P53 and upregulating c-MYC (Aird et al., 2015). However, the role of ATM in age-onset metabolic disturbances is as yet unclear.
Here, we identified a progressive decline in ATM-centered DNA repair machinery during aging, along with shunted glucose metabolism to glycolysis. DNA damage-free activation of ATM by chloroquine (CQ) promotes DNA damage clearance, rescues age-related metabolic shift, and alleviates cellular senescence. Mechanistically, ATM phosphorylates and stabilizes pro-longevity protein SIRT6. Extra copies of Sirt6 attenuate metabolic abnormality and extend lifespan in Atm-/-mice. Importantly, long-term treatment of CQ restores metabolic reprogramming and extends the lifespan of nematodes and a progeria mouse model.

ATM activation alleviates replicative senescence
In searching for genes/pathways that drive senescence, we employed human primary endothelial cells, which underwent replicative senescence at passage 21, with increased p21 expression and bgalactosidase activity (Figure 1-figure supplement 1a-b). By RNAseq analysis, a gradual decline of ATM-centered DNA repair machinery was identified (Figure 1-figure supplement 1c-e). Western blotting analysis confirmed progressively downregulated protein levels of ATM and its downstream target NBS1 and RAP80 in senescent human skin fibroblasts (HSFs) (Figure 1a). Mouse embryonic fibroblasts (MEFs) with limited growth capacity and senescent phenotypes when cultured in vitro (Parrinello et al., 2003;Samper et al., 2003;Sherr and DePinho, 2000), and brain tissues from aged mice also showed progressive decline of ATM, NBS1, and RAP80 (Figure 1b-c). Concomitantly, upregulation of gH2AX, indicating accumulated DNA damage, and an increase in p16 Ink4a were observed in senescent HSFs, MEFs, and aged brain tissues (Figure 1a (Figure 1-figure supplement 2b). These data indicate that ATM decline retards DDR and drives senescence.
Other than DNA damage, ATM is activated by chloroquine (CQ), an antimalarial drug that modulates chromatin confirmation (Bakkenist and Kastan, 2003). We confirmed that a low dose of CQ increased the level of pS1981 auto-phosphorylation of ATM but not gH2AX (Figure 1-figure supplement 2c). We then investigated whether activating ATM by CQ can ameliorate senescence. As shown, the CQ treatment activated ATM (pS1981), promoted clearance of DNA damage (gH2AX), and inhibited apoptosis (cleaved Casp3) in HSFs ( Figure 1g). Also, the CQ treatment suppressed bgalactosidase activity, which was abrogated if ATM was knocked down (Figure 1h-i). Importantly, CQ treatment extended the replicative lifespan of HSFs ( Figure 1j). Likewise, CQ treatment activated Atm, cleared up accumulated DNA damage, suppressed b-galactosidase activity ( Figure 1k and An ATM-SIRT6 axis underlies age-associated metabolic reprogramming A-T patients lacking functional ATM display features of premature aging, accompanied by insulin resistance and glucose intolerance (Bar et al., 1978;Espach et al., 2015). Senescent cells exhibit impaired mitochondrial respiration, but enhanced glycolysis producing more lactate (Hagen et al., 1997;Lenaz et al., 2000). As such, we wondered whether ATM decline triggers an age-associated metabolic shift. Levels of glycolytic genes LDHB and PDK1 were dramatically increased in senescent MEFs and HSFs (Figure 2a and  Immunoblots showing protein levels of ATM, NBS1, and RAP80 in mouse embryonic fibroblasts (MEFs). (c) Immunoblots showing protein levels of ATM, NBS1, and RAP80 in brain tissues isolated from 3-, 10-, and 18-month-old male mice. (d) SA-b-Gal staining in HSFs treated with sh-ATM or scramble Figure 1 continued on next page inhibitory effect on glycolysis was diminished when ATM was depleted in HepG2 cells ( Figure 2b). These data suggest a role for ATM in inhibiting glycolysis.
To examine how ATM regulates glycolysis, we performed RNA-Seq in Atm-/-MEF cells, and revealed a significant upregulation of glycolytic pathways (Figure 2c, and Figure 2-source data 1). Specific genes were validated by q-PCR ( Figure 2-figure supplement 1c). As p53 is critical in glycolysis (Kruiswijk et al., 2015;Schwartzenberg-Bar-Yoseph et al., 2004), we further analyzed metabolomics of Atm-/-and control MEFs in p53 null background. As shown, the metabolic profile exhibited a clear shift, i.e. mitochondrial electron transport chain and intermediates of TCA cycle were reduced, while intermediates of glycolysis were elevated ( Figure 2b). Additionally, ATM depletion in HEK293 cells, HSFs, and MEFs, significantly downregulated SIRT6 protein level, with little effect on SIRT1 or SIRT7 (Figure 2i and Figure 2-figure supplement 2f-g). Thus, these data suggest that ATM decline triggers an age-associated metabolic shift via SIRT6-mediated chromatin remodeling.
Other than metabolic abnormality, depleting Sirt6 leads to premature aging features and shortened lifespan (Mostoslavsky et al., 2006a), whereas extra copies of Sirt6 promote longevity in male mice (Kanfi et al., 2012). Given that Sirt6 was destabilized in Atm null mice, we wondered shRNA. Scale bar, 100 mm. (e) Quantification of SA-b-Gal-positive staining of (d) from five views randomly captured for each group. Data represent means ± SEM. ***p<0.001. (f) Immunoblots showing increased gH2AX and unaffected LC3I/II in HSFs treated with sh-ATM or scramble shRNA. (g) Immunoblots showing protein levels of pS1981 ATM, gH2AX, and cleaved caspase-3 in HSFs treated with 10 mM of CQ for indicated time. (h) SA-b-Gal staining in HSFs expressing either scramble or ATM shRNA treated with 1 mM CQ or DMSO (12 hr). Scale bar, 100 mm. (i) Quantification of SA-b-Galpositive staining of (h) from five views randomly captured for each group. Data represent means ± SEM. ***p<0.001; 'N.S.' indicates no significant difference. (j) HSFs at passage 20 were continuously cultured with 1 mM CQ or DMSO, and cell number was calculated at each passage. Data represent means ± SEM. ***p<0.01. (k) Immunoblots showing protein levels of gH2AX, p62, and LC3 in MEFs treated with 1 mM CQ or DMSO. Note that CQ had little effect on the expression levels of p62 and LC3. (l) MEFs at passage one were continuously cultured in 20% O 2 with 1 mM CQ or DMSO, and cell number was determined at each passage. Data represent means ± SEM. ***p<0.01. DOI: https://doi.org/10.7554/eLife.34836.003 The following source data and figure supplements are available for figure 1:  . Given that little difference was observed in glucose metabolism between young wildtype (WT) and Sirt6-transgenic mice (Kanfi et al., 2012), these data suggest a contributory role of the Atm-Sirt6 axis in the age-associated metabolic reprogramming.

ATM phosphorylates and stabilizes SIRT6
Next, we examined how ATM regulates SIRT6. Significantly, overexpression of ATM increased SIRT6 level, but this was abolished when ATM was S1981A-mutated to block dimeric ATM dissociation (Bakkenist and Kastan, 2003;Berkovich et al., 2007) (Figure 3a). Moreover, in addition to CQ, hypotonic buffer (20 mM NaCl), low glucose (LG), DNA-damaging agent camptothecin (CPT), and doxorubicin (Dox) all activated ATM and concomitantly increased SIRT6 protein level ( . These data implicate a direct regulation of SIRT6 stability by ATM kinase activity. To confirm this, we first performed co-immunoprecipitation (Co-IP) in cells transfected with various FLAG-sirtuins. Interestingly, ATM was predominantly associated with SIRT6 among seven sirtuins ( Figure 3b). The interaction was further confirmed at both ectopic and endogenous levels ( Figure 3c  To determine whether ATM physically binds to SIRT6, 10 consecutive recombinant GST-ATM proteins were obtained and the binding to purified His-SIRT6 was analyzed. As shown (Figure 3e), His-SIRT6 bound predominantly to GST-ATM-4 (residues 770-1102) and relatively weakly to GST-ATM-1 (residues 1-250); both belong to the N-terminal HEAT repeat domain of ATM.
We next examined whether ATM phosphorylates SIRT6. Firstly, we found that CQ or CPT treatment significantly enhanced the binding of SIRT6 to ATM (Figure 3f and  . Moreover, ectopic ATM significantly increased the p-S/TQ level of SIRT6, but this was abolished in the case of S1981A mutant (Figure 3h). Consistently, a pronounced reduction of p-S/TQ level of SIRT6 was observed in cells lacking ATM or treated with KU55933, a selective and specific ATM kinase inhibitor (Berkovich et al., 2007;Hickson et al., 2004) (Figure 3i-j). The decrease in p-S/TQ level was primarily attributable to loss of ATM, as it was restored by ectopic FLAG-ATM in a dose-dependent manner ( Figure 3-figure supplement 1j). Indeed, SIRT6 has one evolutionarily conserved S 112 Q 113 motif ( Figure 3k). We therefore constructed S112A and S112D mutants, which resemble hypo-and hyper-phosphorylated SIRT6 respectively. As shown, these mutations almost abolished the pS/T-Q level of FLAG-SIRT6 (Figure 3l). The in vitro kinase assay showed that ATM could phosphorylate GST-SIRT6, but not S112A (Figure 3-figure supplement 1k). Furthermore, compared with SIRT6 S112A, ectopic S112D exhibited a much higher inhibitory effect on glycolytic gene expression in sh-SIRT6 HepG2 cells (Figure 3m and Figure 3-figure supplement 1l), and enhanced chromatin association of SIRT6 ( Figure 3n). Collectively, the data suggest that ATM directly phosphorylates SIRT6 at Serine 112.
We next examined whether ATM is involved in regulating SIRT6 protein stability. Notably, compared with WT or vehicle control, the degradation rate of ectopic and endogenous SIRT6 was largely increased in ATM KO HEK293 cells, Atm-/-MEFs, and cells incubated with KU55933 in the presence of cycloheximide (CHX) (Figure 4a-b and Figure 4-figure supplement 1a-c). Recently, MDM2 was demonstrated to ubiquitinate SIRT6 and promote its proteasomal degradation (Thirumurthi et al., 2014). We therefore examined the polyubiquitination level of SIRT6. As shown, the ubiquitination level of FLAG-SIRT6 in ATM KO cells was significantly elevated compared with WT ( Figure 4-figure supplement 1d). While S112A mutant markedly enhanced the polyubiquitination level of SIRT6, S112D had little effect (Figure 4-figure supplement 1e). Moreover, S112A accelerated SIRT6 degradation, whereas S112D retarded it (Figure 4c-d), indicating that the Ser112 phosphorylation by ATM regulates SIRT6 ubiquitination and thus protein stability. Indeed, ectopic MDM2 enhanced the polyubiquitination level of FLAG-SIRT6 (Figure 4-figure supplement 1f). In the case of ATM depleted or SIRT6 S112A mutant, the binding capacity of SIRT6 to MDM2 was enhanced ( Figure 4e and Figure 4-figure supplement 1g). In searching for key residues that are polyubiquitinated by MDM2, we identified two clusters of lysine residues, i.e. K143/145 and K346/349, which are conserved across species. We then generated KR mutations of these residues, and found that K346/ 349R remarkably reduced the polyubiquitination level of SIRT6 (Figure 4-figure supplement 1h). Individual KR mutation showed that K346R significantly blocked MDM2-mediated ubiquitination and (k) Alignment of protein sequence of human SIRT6 and orthologues in mouse, rat, fruit fly, Xenopus, and C. elegans. A conserved S 112 Q 113 motif was highlighted. (l) Immunoblots showing p-S/T Q level of Flag-SIRT6, Flag-SIRT6 S112A, or Flag-SIRT6 S112D in HEK293 cells. (m) Quantitative RT-PCR analysis of mRNA levels of indicated glycolytic genes in sh-SIRT6 HepG2 cells re-expressing SIRT6, SIRT6 S112A, or 112D mutation. Data represent means ± SEM. *p<0.05, **p<0.01, ***p<0.001. 'ns' indicates no significant difference. (n) Immunoblots showing SIRT6 protein level in total cell extract (TCE) and chromatin-enriched fractions (P2). Densitometry analysis was performed to determine the relative ratio of SIRT6/H2B within chromatin fractions. DOI: https://doi.org/10.7554/eLife.34836.014 The following figure supplement is available for figure 3:   Figure 4 continued on next page degradation of SIRT6, whereas K349R hardly affected it (Figure 4f-g). More importantly, K346R restored the increased ubiquitination and accelerated protein degradation of SIRT6 S112A (Figure 4-figure supplement 1i-j). Collectively, these data indicate that K346 is subject to MDM2mediated ubiquitination, which is inhibited by ATM-mediated S112 phosphorylation.

Activating ATM via CQ promotes longevity
The cellular data suggest a pro-longevity function of ATM. We then tested it at organismal level. We employed Caenorhabditis elegans, which have a short lifespan of approximate 30 days. Nematodes deficient for atm-1, an orthologue of mammalian ATM, and WTs were exposed to various doses of CQ (see Materials and methods). Significantly, the period treatment with CQ (1.0 mM) extended the median lifespan (~14%) of C. elegans (Figure 5a). The lifespan-extending effect was abolished in atm-1 KO (Figure 5b) or in SIRT6 homolog sir-2.4 KD nematodes ( Figure 5-figure supplement 1ab). The data suggest that CQ promotes longevity in an ATM-and SIRT6-dependent manner. We further examined the beneficial effect of CQ in a HGPS model, i.e. Zmpste24-/-mice, which has a shortened lifespan of 4-6 months (Pendás et al., 2002) and impaired ATM-mediated DNA repair signaling (Liu et al., 2013a). We found that the level of Atm was dramatically reduced in Zmpste24-/-MEFs and tissues (Figure 5c and Physiologically aged mice frequently develop aging-associated metabolic disorders, with high glucose and lactate (Houtkooper et al., 2011). Given that ATM declines with age, and activation of ATM by CQ inhibits glycolysis in senescent cells and Zmpste24-/-mice, we intraperitoneally administrated 12-month-old 'old' male mice with low-dose CQ (3.5 mg/kg) twice a week. Remarkably, compared with the saline-treated group, CQ treatment inhibited glycolysis, lowered serum lactate level, and attenuated body weight decline (Figure 5i and Figure 5-figure supplement 1g-h), implicating potential benefits of CQ in physiologically aged mice. Generally, these data demonstrate a lifespanextending benefit of ATM activation by CQ.

Discussion
DNA damage accumulates with age, and defective DDR and DNA repair accelerates aging. However, whether boosting DNA repair machinery promotes healthiness and longevity is still unknown. DNA damage stimulates DDR, but if persistent, it instead leads to senescence. Therefore, if enhancing DDR efficacy possibly promotes longevity, it must be DNA damage free. The antimalarial drug CQ can intercalate into the internucleosomal regions of chromatin, unwind DNA helical twist, and thus activate ATM without causing any DNA damage (Bakkenist and Kastan, 2003;Krajewski, 1995). We demonstrate that long-term treatment with CQ activates ATM, improves DNA repair, restores age-related metabolic shift, alleviates cellular senescence, and extends lifespan of nematodes and Zmpste24 null mice. Mechanistically, ATM phosphorylates the longevity gene SIRT6 (Tasselli et al., 2017), and prevents MDM2-mediated ubiquitination and proteasomal degradation of SIRT6. To our knowledge, this is the first study to establish direct causal links between robust DNA repair machinery and longevity. In support of this notion, DNA repair efficacy has been shown to be enhanced in long-lived naked mole rat (MacRae et al., 2015), and human longevity has been shown to be associated with single nucleotide polymorphisms (SNPs) in DNA repair genes/pathways (Debrabant et al., 2014;Soerensen et al., 2012). Interestingly, the heterozygous rather than homozygous status of a SNP, albeit both enhance the transcription of ATM, is associated with longevity in Chinese and Italian populations (Chen et al., 2010;Piaceri et al., 2013). Therefore, in future study, it would be worthwhile evaluating whether Atm can promote longevity in model organisms, and, if so, how many extra copies are required.  Moskalev et al., 2013). Metabolic reprogramming from the TCA cycle to glycolysis is prominent in both physiological and pathological aging (Feng et al., 2016;Shimizu et al., 2014). Why senescent cells become glycolytic is poorly understood. The crosstalk between cellular metabolism and DDR is not well elucidated. Upon genotoxic stress, ATM represses the rapamycin-sensitive mammalian target of the rapamycin (mTORC1) pathway (Alexander et al., 2010), but activates the pentose phosphate pathway (PPP) (Cosentino et al., 2011), suggesting that cell metabolism may be key downstream of DDR signaling. Moreover, it is recognized that deficiency in DNA repair machinery such as ATM, WRN, and Ercc1, accelerates aging and causes severe metabolic disorders (Garinis et al., 2008;White and Vijg, 2016). In this study, we showed that boosting ATM activity using a low dose of CQ enhances genomic stability, attenuates age-onset metabolic reprogramming, alleviates senescence, and extends lifespan in mice. The data demonstrate for the first time that enhanced DNA repair machinery (ATM-SIRT6 axis) promotes longevity. Considering that ATM and SIRT6 function not only in maintenance of genome integrity but also as homeostatic protein modifiers, the pro-longevity role of ATM is most likely benefited from enhanced DNA repair and metabolic homeostasis, but it is difficult to determine which is more important.
Recently, Bohr's group identified that increased consumption of NAD + by an early DDR factor poly (ADP-ribose) polymerase (PARP1), owing to accumulated DNA damage, accelerated aging in Atm mutant mice (Fang et al., 2016). NAD + serves as a cofactor of sirtuins, including SIRT1 and SIRT6. Therefore, this work establishes a linear causal link between deficient DDR, DNA damage accumulation, consumption of NAD + , decline in sirtuin activity, and aging. Moreover, administration of nicotinamide mononucleotide or nicotinamide riboside ameliorates age-related function decline and extends lifespan in mice (Mills et al., 2016;Zhang et al., 2016). Here, we found that ATM decline during aging causes DNA damage accumulation and enhances glycolysis, both of which consume most of the NAD + , providing an explanation for the low NAD + level in Atm-/-mice and physiologically aged mice.
Closely resembling normal aging, HGPS has attracted numerous efforts in understanding of molecular mechanisms and developing therapeutic strategies (De Sandre-Giovannoli et al., 2003;Eriksson et al., 2003). We and others have found that HGPS and Zmpste24 null cells undergo premature senescence because of defective chromatin remodeling (Ghosh et al., 2015;Krishnan et al., 2011;Liu et al., 2013a;Liu et al., 2013b), delayed DDR, and impaired DNA repair Liu et al., 2013b;Varela et al., 2005). Specifically, Atm-Kap-1 signaling is compromised (Liu et al., 2013a), and SIRT6 protein level and deacetylase activity are reduced in progeria cells (Endisha et al., 2015;Ghosh et al., 2015). Here we showed that Atm is significantly downregulated, which explains the reduced SIRT6, delayed DDR, and metabolic shift in progeria cells and mice. It would be interesting to investigate whether ectopic Atm or Sirt6 could rescue progeroid features in these mice. Nevertheless, the activation of ATM via CQ remarkably improves glucose homeostasis, DNA damage clearance, and running endurance, and extends lifespan in progeria mice. It would be worthwhile evaluating the pro-longevity benefits of CQ in physiological aging.
CQ is an FDA-approved and clinically used medicine for treatment of malaria (2015). Via activation of ATM, long-term treatment of CQ protects against atherosclerosis, improves insulin sensitivity, and rescues glucose tolerance in type 2 diabetes (T2D) (Emami et al., 1999;Razani et al., 2010;Schneider et al., 2006). The lysosomotropic property of CQ also makes it a potent inhibitor of autophagy (Yang et al., 2013). The application of CQ for antimalarial treatment (500 mg/week, maximum 0.8 mM in plasma) and for cancer therapy (100-500 mg/day) (Kimura et al., 2013) is attributed to its inhibitory action on autophagy. Of note, CQ also the attenuates inflammatory response by inhibiting autophagy (Szatmári-Tó th et al., 2016;Whelan et al., 2017;Wu et al., 2018), requiring a high dosage of 50 mg/kg for mice. In the current study, we used a low dose of CQ to activate ATM, i.e. 1-10 mM for cell line and 3.5 mg/kg twice a week for mice (Schneider et al., 2006). The results showed that at such low doses, CQ has no toxicity and little effect on basal autophagic activity. Moreover, a low dose of CQ prolongs lifespan in progeroid mice, but exhibits little effect on Atm KO background, supporting an ATM-dependent pro-longevity function of CQ. Unfortunately, we could not test the CQ effect in Sirt6-/-mice with only 1-month lifespan (Mostoslavsky et al., 2006b). Here, we addressed the pro-longevity benefits of CQ-activated ATM, most likely attributable to improved DNA repair and glucose metabolism. Given that ATM also displays anti-inflammatory function (Erttmann et al., 2017;Shoelson, 2006), we could not rule out an anti-inflammatory effect in lifespan extension observed in CQ-treated mice.
In conclusion, our data establish direct causal links between robust DNA repair machinery and longevity. In line with DNA damage theory of aging, we propose that DNA damage activates DDR; however, its constant activation causes senescence; defective ATM-SIRT6 axis underlies premature aging, exemplified by HGPS and A-T mouse models, which are rescued by treatment of CQ and Sirt6 transgene, respectively; in physiological aging, DNA damage-free activation of ATM by CQ stabilizes SIRT6, thus promoting longevity in nematodes and most likely also in mice ( Figure 5-figure  supplement 2). Our findings provide a novel therapeutic strategy for HGPS, and could facilitate clinical trials of CQ as an effective treatment for age-related diseases.

Mice
Zmpste24-/-mice and Atm-/-mice have been described previously (Barlow et al., 1996;Pendás et al., 2002). Sirt6-transgenic mice (Sirt6-tg) of C57BL/6J background were constructed by injecting cloned mSirt6 cDNA with CAG promoter into fertilized eggs. Primers for genotyping of Sirt6 transgenic allele were as follows: forward: 5'-CTGGTTATTGTGCTGTCTCATCAT-3'; reverse: 5'-CCGTCTACGTTCTGGCTGAC-3'. Atm-/-mice were crossed to Sirt6-tg mice to get Atm-/-;Sirt6-tg mice. Chloroquine (CQ) experiments were conducted as described (Schneider et al., 2006). Briefly, 12-month-old wild-type C57BL/6J male mice, 2-month-old Zmpste24-/-, and Atm-/-male mice were administered with CQ (Sigma, St. Louis, MO) in 0.9% saline twice per week at 7 mg/kg body weight, and the control group was treated with saline alone. At least 8 weeks after treatment of CQ, mice were subjected to functional tests. Body weight and lifespan was recorded. The survival rate was analyzed using the Kaplan-Meier method and statistical comparison was performed using the Logrank Test. Mice were housed and handled in the laboratory animal research center of Shenzhen University. All experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC). The protocols were approved by the Animal Welfare and Research Ethics Committee of Shenzhen University (Approval ID: 201412023).
C. elegans survival assay C. elegans nematode survival assay was performed according to standard protocols (Kenyon et al., 1993). Briefly, wild-type and atm-1 null nematodes (100 to 150 per group) synchronized to prefertile young adult stage were exposed to NGM plates containing the indicated dosage of CQ. After 1-day incubation, animals were transferred to fresh incubation plates without CQ for another 2 days. This procedure was repeated every 3 days. Nematodes that showed no response to gentle stimulation were recorded as dead. The survival data were analyzed using the Kaplan-Meier method and statistical comparison was performed using the Log-rank Test.

Cell lines
HEK293 (CRL-1573), HepG2 (HB-8065), and U2OS (HTB-96) cells were purchased from ATCC. Human skin fibroblasts HSFs (F2-S) and primary MEFs were prepared as described previously . Immortalized Atm-/-; p53-/-and Sirt6-/-MEFs were provided as a kind gift from Dr. Yosef Shiloh (Tel Aviv University, Israel) and Dr. Raul Mostoslavsky (Massachusetts General Hospital Cancer center, USA), respectively. These cell lines were authenticated by short tandem repeat (STR) profile analysis and genotyping, and were mycoplasma free. Cells were cultured in Gibco DMEM (Life Technologies, USA) with 10% fetal bovine serum (FBS), 100 U/ml penicillin and streptomycin (P/S) at 37˚C in 5% CO 2 and atmospheric oxygen conditions. For CQ experiments, cells were maintained in the medium containing 1 mM chloroquine for 12 hr, and then grown in new fresh medium for 48 hr.

Protein extraction and western blotting
For whole cell protein extraction, cells were suspended in five volumes of suspension buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, protease inhibitor cocktail), and then five volumes of 2X SDS loading buffer were added and incubated at 98˚C for 6 min. Mice tissues were homogenized with 1 ml of ice-cold tissue lysis buffer (25 mM TrisHCl, pH 7.5, 10 mM Na 3 VO 4 , 100 mM NaF, 50 mM Na 4 P 2 O 7 , 5 mM EGTA, 5 mM EDTA, 0.5% SDS, 1% NP-40, protease inhibitor cocktail). After homogenization and sonication, lysates were centrifuged at 16,000 g for 15 min. The clean supernatant was carefully transferred to new tubes. Protein concentrations were determined using a bicinchoninic acid (BCA) assay method (Pierce, Rockford, IL) and were normalized with lysis buffer for each sample. Samples were denatured in 1X SDS loading buffer by boiling at 98˚C for 6 min. Proteins were separated by loading to SDS-polyacrylamide gels, and then were transferred to PVDF membrane (Millipore). The protein levels were determined by immunoblotting using respective antibodies. The ImageJ program was used for densitometric analysis of immunoblotting, and the quantification results were normalized to the loading control.

Protein degradation assay
HEK293 cells (WT and ATM-deficient cells) were transfected with Flag-SIRT6 alone or together with Mdm2. 48 hr later, the cells were treated with 50 mg/ml of cycloheximide (CHX, Sigma-Aldrich), a translation inhibitor. For endogenous SIRT6 protein degradation assay, ATM wild-type and null MEFs were grown in 6 cm plates, and were treated with 50 mg/ml CHX for indicated time points. Cells were collected and the protein levels were determined by western blotting, the subsequent quantification was performed with ImageJ software.

In vivo ubiquitination assay
In vivo ubiquitination assay was performed by transfecting HEK293 cells in 6 cm dishes with 1 mg Myc-ubiquitin, 2 mg Flag-SIRT6 or its mutations, and/or 1 mg MDM2 vector. 48 hr after transfection, cells were lysed in the buffer (25 mM Tris-HCl pH 8.0, 250 mM NaCl, 10 mM Na 3 VO 4 , 1 mM EDTA, 10% glycerol, protease inhibitor cocktail, and 0.1 mM phenylmethylsulphonyl fluoride), and then incubated with Flag-M2 beads (Sigma-Aldrich) overnight at 4˚C. Beads were washed with lysis buffer three times, bound proteins were eluted by adding 1.5 Â SDS loading buffer. The ubiquitin levels were analyzed by immunoblotting.
In vitro kinase assay HEK293T cells were transfected with 10 mg of FLAG-ATM and then treated with CPT. Activated ATM was immune-purified from the cell extracts with FLAG beads (Sigma, M8823). GST-SIRT6 or the S112A mutant was purified from bacteria. Kinase reactions were initiated by incubating purified ATM with GST-SIRT6 in the kinase buffer with or without 1 mM ATP for 120 min at 30˚C. After reaction, proteins were blocked by SDS loading buffer. The membrane was then subjected to western blotting with antibodies against p-S/TQ.

Immunoprecipitation
Cells under indicated treatments were totally lysed in lysis buffer containing 20 mM HEPES, pH 7.5, 150 mM NaCl, 10 mM Na 3 VO 4 , 10% glycerol, 2 mM EDTA, protease inhibitor cocktail, and 0.1 mM phenylmethylsulphonyl fluoride. After sonication and centrifugation, the supernatant was collected and incubated with H3K9ac (Millipore, 2 mg/sample) overnight at 4˚C with a gentle rotation. Protein A/G agarose (Pierce, 10 ml/sample) were added to the tubes and rotated at 4˚C for 2 hr. Beads were precipitated by centrifugation at 1000 g for 15 s and washed three times with cold lysis buffer. The pellet was resuspended in 1.5 Â SDS loading buffer and incubated at 98˚C for 6 min. The supernatants were collected and used for western blotting.

GST pull-down assay
A series of GST fusion proteins of truncated ATM, which together spanned the full length of ATM, were constructed into pGEX4T-3 vector. For GST pull-down, bacterially expressed 6 Â His tagged SIRT6 was separately incubated with various GST-ATM fragments in a buffer of 150 mM NaCl, 20 mM Tris-HCl [pH 7.5], 5 mM MgCl 2 , 0.2 mM EDTA, 10% glycerol, 0.2% NP-40, and protease inhibitors (Roche Complete). GST-fusion proteins were then precipitated by adding Glutathione Sepharose fast flow (GE Healthcare). After washing twice with TEN buffer (0.5% Nonidet P-40, 20 mM Tris-HCl [pH 7.4], 0.1 mM EDTA, and 300 mM NaCl), glutathione agarose beads were analyzed by western blotting and coomassie staining.

EdU (5-ethynyl-2'-deoxyuridine) incorporation assay
EdU incorporation assays were conducted in HSF cells to estimate cell proliferation using the Click-iT EdU Alexa Fluor 488 Kit (Invitrogen, USA). HSF cells, infected by the respective lentiviruses containing shNC and shATM, were cultured in a six-well plate containing the coverslips in the presence of 10 mM EdU for 12 hr. Cells were fixed in 3.7% formaldehyde followed by 0.5% Triton X-100 permeabilization, and then stained with Alexa Fluor picolyl azide. Five random views were captured to calculate the positive staining rate for each group.

Growth curves and SA-b-gal assays
Cell population doublings were monitored using a Coulter Counter. SA-b-galactosidase assay in primary cells was performed using Senescence b-galactosidase staining Kit (#9860, CST) according to the manufacturer's instructions. Five views were captured randomly to calculate the positive staining rate for each group.

RNA preparation and Real-Time qPCR
Total RNA was extracted from cells or mouse tissues using Trizol reagent RNAiso Plus (TaKaRa, Japan) following the phenol-chloroform extraction method. Purified total RNA was used to obtain cDNA using PrimeScript RT Master Mix (Takara, Japan) following this method: 37˚C for 30 min, and 85˚C for 5 s. The gene expression was analyzed with the CFX Connected Real-Time PCR Detection System (BioRad) with SYBR Ex Taq Premixes (Takara, Japan). Gene expression levels were normalized to actin.

Lactate assay
Mouse serum was five-fold diluted, and lactate concentration was determined with the Lactate Colorimetric Assay Kit (BioVision).

Endurance running test
Zmpste24-/-mice were treated for 8 weeks with chloroquine or saline before running on a Rota-Rod Treadmill (YLS-4C, Jinan Yiyan Scientific Research Company, Shandong, China) to test the effect of chloroquine on fatigue resistance. Mice were placed on the rotating lane, and the speed was gradually increased to 10 r/min. When mice were exhausted and safely dropped from the rotating lane, the time latency to fall was automatically recorded.

Metabolite analysis
Wild-type and ATM KO cells were grown in normal medium for 24 hr, and methanol-fixed cell pellets were analyzed by a two liquid chromatography-tandem mass spectrometry (LC-MS) method as described (Luo et al., 2007).

Immunofluorescence microscopy
The cells were fixed using 4% paraformaldehyde at room temperature for 15 min, permeabilized by 0.5% Triton X-100 at room temperature for 10 min, blocked using 10% FBS/PBS, and then incubated with primary antibodies diluted in PBS containing 2% BSA overnight at 4˚C. The primary antibodies were detected using an Alexa-488-conjugated anti-mouse secondary antibody (Invitrogen). The nuclei were stained using DAPI in anti-fade mounting medium. Images were captured using a Zeiss LSM880 confocal/multiphoton microscope.

ChIP assay
Cells were fixed in 1% formaldehyde for 10 min at room temperature. The crosslinking reaction was quenched with 0.125 M glycine. After washing with PBS, cells were lysed with lysis buffer (50 mM TrisÁHCl pH 8.0, 2 mM EDTA, 15 mM NaCl, 1% SDS, 0.5% deoxycholate, protease inhibitor cocktail, 1 mM PMSF), followed by sonication and centrifugation. The supernatant was collected and precleared in dilution buffer (50 mM Tris-HCl pH 8.0, 2 mM EDTA, 150 mM NaCl, 1% Triton X-100) with protein A/G Sepharose and pre-treated salmon DNA. The precleared samples were incubated overnight with H3K9ac antibody (2 mg/sample, Millipore) or appropriate control IgGs (Santa Cruz), and protein A/G Sepharose (Invitrogen). After washing sequentially with a series of buffers, the beads were heated at 65˚C to reverse the crosslink. DNA fragments were purified and analyzed. Real-time PCR was performed with primers as described (Zhong et al., 2010):

Statistical analysis
Statistical analyses were conducted using two-tailed Student's t-test between two groups. All data are presented as mean ± S.D. or mean ± S.E.M. as indicated, and a p value < 0.05 was considered statistically significant. ( The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.