Ablation of XP-V gene causes adipose tissuesenescence and metabolic abnormalitiesYih-Wen Chena, Robert A. Harrisb, Zafer Hatahetc, and Kai-ming Choua,1

aDepartment of Pharmacology and Toxicology, Indiana University School of Medicine, Indianapolis, IN 46202; bRichard Roudebush Veterans Affairs MedicalCenter and the Department of Biochemistry and Molecular Biology, Indiana University School of Medicine, Indianapolis, IN 46202; and cDepartment ofBiological and Physical Sciences, Northwestern State University of Louisiana, Natchitoches, LA 71497

Edited by James E. Cleaver, University of California, San Francisco, CA, and approved June 26, 2015 (received for review April 12, 2015)

Obesity and the metabolic syndrome have evolved to be majorhealth issues throughout the world. Whether loss of genomeintegrity contributes to this epidemic is an open question. DNApolymerase η (pol η), encoded by the xeroderma pigmentosum(XP-V) gene, plays an essential role in preventing cutaneous cancercaused by UV radiation-induced DNA damage. Herein, we demon-strate that pol η deficiency in mice (pol η−/−) causes obesity withvisceral fat accumulation, hepatic steatosis, hyperleptinemia,hyperinsulinemia, and glucose intolerance. In comparison to WTmice, adipose tissue from pol η−/− mice exhibits increased DNAdamage and a greater DNA damage response, indicated by up-regulation and/or phosphorylation of ataxia telangiectasia mu-tated (ATM), phosphorylated H2AX (γH2AX), and poly[ADP-ribose]polymerase 1 (PARP-1). Concomitantly, increased cellular senes-cence in the adipose tissue from pol η−/− mice was observed andmeasured by up-regulation of senescence markers, including p53,p16Ink4a, p21, senescence-associated (SA) β-gal activity, and SA se-cretion of proinflammatory cytokines interleukin 6 (IL-6) and tu-mor necrosis factor α (TNF-α) as early as 4 wk of age. Treatment ofpol η−/− mice with a p53 inhibitor, pifithrin-α, reduced adipocytesenescence and attenuated the metabolic abnormalities. Further-more, elevation of adipocyte DNA damage with a high-fat diet orsodium arsenite exacerbated adipocyte senescence and metabolicabnormalities in pol η−/− mice. In contrast, reduction of adiposeDNA damage with N-acetylcysteine or metformin ameliorated cel-lular senescence and metabolic abnormalities. These studies indi-cate that elevated DNA damage is a root cause of adipocytesenescence, which plays a determining role in the developmentof obesity and insulin resistance.

DNA polymerase η | obesity | senescence | DNA damage | adipose tissue

The human genome is constantly challenged by exogenous andendogenous DNA damaging agents. To ensure genome in-

tegrity, human cells have faithful DNA replication machineryand DNA repair systems that are coordinated by DNA damageresponse networks. In response to different extents or types ofDNA lesions, the DNA damage response activates appropriatecellular responses, including transient or permanent (senes-cence) cell cycle arrest, or apoptosis, to minimize the detrimentaleffects of DNA lesions (1). Reduction or deficiency in DNArepair/replication enzyme activity is well documented to increasevulnerability for the development of cancer, neurodegenerativediseases, and aging (2). In addition, defective DNA repair en-zymes are associated with the metabolic symptom; for example,DNA glycosylase (Neil1)- and OGG1-deficient mice are obese(3–5), and nucleotide excision repair protein ERCC1-XPF de-ficiency causes lipodystrophy (6). Furthermore, DNA damageresponse protein ataxia telangiectasia mutated (ATM) sup-presses JNK activity through p53 signaling and mediates an an-tioxidant action that has been suggested to be relevant to themetabolic syndrome (7). Nucleotide excision repair XP-A pro-tein may affect metabolism by altering mitochondrial function(8). To date, the mechanisms that link genome instability andmetabolic dysregulation have not been elucidated.

DNA polymerase η (pol η) is a specialized lesion bypass poly-merase that faithfully replicates across UV-induced cyclobutanepyrimidine dimers (9) to rescue stalled DNA replication forksfrom potential breakages and mutations. Defects in the geneencoding pol η produce a variant form of the autosomal recessivedisease xeroderma pigmentosum (XP-V) (9). Patients with XP-Vare highly sensitive to sunlight and prone to cutaneous cancer (9).In addition to skin, pol η is expressed in most tissues (10). Theexpression of pol η correlates with the effectiveness of anticancertherapeutic agents that exert their activity by introducing DNAlesions that block the progression of DNA replication (11). Inaddition to exogenous introduced DNA lesions, pol η contributesto genomic stability during unperturbed DNA replication (12) andreplicates across reactive oxygen species (ROS)-induced oxidativeDNA lesions 8-oxoG (7,8-dihydro-8-oxoguanine), thymine glycol,and lipid peroxidation DNA adducts generated during endoge-nous metabolic processes (13, 14). ROS generated from endoge-nous metabolism or exogenous sources has been associated withcancer, aging, and the metabolic syndrome.Therefore, the initial hypothesis for our studies was that pol η

contributes to reduce tumorigenesis by managing oxidative DNAlesions in other organs. However, in a 2-y study, no increase in theincidence of tumors of any type was found with pol η KO (pol η−/−)mice (15). Instead, we found pol η−/− mice develop metabolicabnormalities that induce obesity.

Results and DiscussionMetabolic Abnormalities in Pol ηMice.Compared with littermateWTmice, pol η−/− mice gained more weight with age, particularly after48 wk, with no genotype difference in food or water consumption

Significance

The metabolic syndrome has evolved to be a major health issueglobally. The association between genome integrity and meta-bolic abnormalities is not well understood. Our results indicatethat increased DNA damage and persistent activation of theDNA damage response induce adipocyte senescence in DNApolymerase η knockout (pol η−/−) mice. Suppression of adipocytesenescence with a p53 inhibitor, pifithrin-α, alleviated metabolicabnormalities in pol η−/− mice. An increase or decrease in DNAdamage affected the senescence status of adipocytes accord-ingly, which was also in concordance with the severity of met-abolic abnormalities in the pol η−/− mice. Our current resultsindicate that reduced genome integrity plays a causative role inprovoking adipocyte senescence that leads to development ofobesity and insulin resistance.

Author contributions: Y.-W.C., R.A.H., Z.H., and K.-m.C. designed research; Y.-W.C., Z.H.,and K.-m.C. performed research; Y.-W.C., R.A.H., and K.-m.C. analyzed data; and Y.-W.C.,R.A.H., and K.-m.C. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: kmchou3@gmail.com.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1506954112/-/DCSupplemental.

E4556–E4564 | PNAS | Published online August 3, 2015 www.pnas.org/cgi/doi/10.1073/pnas.1506954112

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

Fig. 1. Pol η−/− mice develop metabolic abnormalities. (A) Autopsy of WT and pol η−/− mice at 4 (4W) and 72 (72W) wk of age; (B) body weight gain (n = 12);(C) body fat percentages (n = 12); (D) circulating leptin and (E) insulin levels; (F) HOMA-IR (n = 8); (G) glucose tolerance test (GTT) with area under curve (AUC)(n = 8); (H) liver stained with Oil Red O (Scale bars: 50 μM); (I) H&E staining of adipose macrophage infiltration; (J) F4/80+/CD11b+-positive cells (n = 6) inadipocytes from mice at 72 wk (w) of age; (K) IL-6; (L) TNF-α; (M) adiponectin in plasma (n = 8); (N) H&E staining of visceral fat histological sections; (O)quantitation of adipocyte size (n = 400); and (P) expression of PPAR-γ, SREBP1-p125, and SREBP1-p68 subunits. Histological pictures and blots are repre-sentative of six independent experiments. The graphical data, which are shown as the mean ± SEM, were analyzed by the Student t test (*P < 0.05).

Chen et al. PNAS | Published online August 3, 2015 | E4557

GEN

ETICS

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

(Fig. 1 A–C and Fig. S1A). Autopsies revealed increased visceral fatencasing the abdominal organs of pol η−/−mice (Fig. 1A). By 72 wk,pol η−/− mice had gained an average of 55% more weight than WTlittermates (14.0 g vs. 9.0 g, respectively; Fig. 1B), with 37.5% morebody fat (38.5% vs. 28.0%, respectively; Fig. 1C). Pol η−/− mice alsodeveloped hyperleptinemia (Fig. 1D), hyperinsulinemia (Fig. 1E),higher blood glucose (Fig. S1B), higher homeostasis model assess-ment of insulin resistance (HOMA-IR) (Fig. 1F), reduced energyexpenditure (Fig. S1C), and reduced glucose tolerance (Fig. 1G)compared with WT littermates. Hepatic steatosis was also appar-ent in the pol η−/− mice, as evidenced by enlarged and light-colored livers (Fig. 1A), increased micro- and macrovesicles (Fig.S1D), and increased Oil Red O staining (Fig. 1H).Infiltration of macrophages into adipose tissue is an established

characteristic of obesity (16), which was apparent in the adiposetissue of pol η−/− mice, as evidenced by crown-like structures inhistological sections (Fig. 1I) and an increased number of F4/80+/CD11b+-positive cells compared with WT mice (17) (Fig. 1J).Consistent with these findings, the inflammatory cytokines IL-6,TNF-α, and monocyte chemoattractant protein-1 (MCP-1) wereincreased and the antiinflammatory cytokine adiponectin (16) wasdecreased in the plasma and adipocytes (Fig. 1 K–M and Fig. S2A)of pol η−/− mice.The adipocytes from pol η−/− mice developed hypertrophy with

age (4.6-fold larger than WT at 72 wk old; Fig. 1 N and O) andexhibited elevated expression of the master adipogenic regulator

genes SREBP1 and PPAR-γ (18) (Fig. 1P and Fig. S2B). Theadministration of diets high in fat or fructose promoted greaterobesity and exacerbated the metabolic abnormalities in the pol η−/−

mice relative to their WT littermates (Fig. S3), which further in-dicated that pol η−/− mice are prone to the development of meta-bolic abnormalities. Together, these characteristics of obesity withconcomitant development of glucose intolerance and reducedinsulin sensitivity in the pol η−/− mice suggest a preventive rolefor pol η in the onset of metabolic abnormalities.

Increased DNA Breaks and DNA Damage Response in Pol η−/− Mice.To our knowledge, there is no information in the literature as tohow a deficiency of a DNA lesion bypass polymerase causesmetabolic abnormalities. Autopsy and histopathological analysisindicated that visceral adipose tissue accumulation was the mostprominent change in major organs between pol η−/− and WTmice (Fig. 1 A–C). In addition to its vital role in energy storage,adipose tissue plays essential roles in the immune response andthe secretion of hormones and cytokines. Adipose tissue isclosely associated with metabolic dysfunction, inflammation, andaging-related health issues.Previous studies demonstrated that ROS are selectively in-

creased in the adipose tissue in obese mice (19), which implies ahigher amount of DNA damage in adipose tissue and a potentiallinkage between oxidative DNA damage, fat accumulation, andobesity. However, it is not clear whether increased DNA damage

Fig. 2. Elevated DNA damage and DNA damage responses in the adipocytes from pol η−/− mice. (A) Alkaline comet assay. (Scale bars: 100 μM.) (B) Quan-titation of comet assay (n = 200). (C) Immunofluorescence staining of phosphorylated γH2AX. (Scale bars: 25 μM.) DNA damage protein expression (D) andquantitation (E) in WT, pol η−/− adipocyte control (Ctl), and pol η−/− adipocytes treated with NAC or Met (n = 6). The data, which are shown as the mean ±SEM, were analyzed by one-way ANOVA (*P < 0.05).

E4558 | www.pnas.org/cgi/doi/10.1073/pnas.1506954112 Chen et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

can be a causal factor for obesity development. We thereforepostulated that increased genome instability in adipose tissues,due to pol η ablation, leads to the development of the observedmetabolic disorders in pol η−/− mice. To test this hypothesis, analkaline comet assay was performed, and, indeed, the resultsindicated greater DNA damage in pol η−/− adipocytes as early as4 wk of age compared with pol η−/− adipocytes from WT mice ofthe same age (Fig. 2 A and B). These results were further sup-ported by increased staining of 8-oxoG (13) and a higher amountof DNA double-strand breaks, as indicated by phosphorylatedγH2AX (20) in the adipose tissue from pol η−/− mice (Fig. 2 C–Eand Fig. S4). Concurrently, up-regulation of DNA damage re-sponse proteins, including ATM, poly[ADP-ribose] polymerase 1(PARP-1), p53, and p21, at the protein, mRNA, and phos-phorylation levels, was observed in adipocytes from pol η−/− mice(Fig. 2 D and E and Fig. S5A).Activation of the DNA damage response has been demon-

strated to up-regulate the inflammatory response through themaster regulator of inflammation gene NF-κB (21). The ob-served up-regulation of ATM, p53, and PARP-1 in the pol η−/−

mice may induce up-regulation of NF-κB via IL-6 and TNF-α(21), which would provide an explanation for the elevated in-flammatory responses observed in pol η−/− mice (Fig. 1 J and K).Indeed, expression of NF-κB (p65) at both the protein andmRNA levels was higher in adipocytes from pol η−/− mice (Fig. 2D and E and Fig. S5A). Concomitantly, the expression of IκBα,an inhibitor of NF-κB (22), was reduced (Fig. 2 D and E). Thesefindings indicate that ablation of pol η increases the amount ofDNA damage and DNA damage response, which contributes tothe elevated inflammatory responses of pol η−/− mice. In addi-tion, IL-6 and TNF-α increase oxidative stress, which is likely toinduce additional DNA damage (16), and subsequently propa-gate a vicious cycle between DNA damage and inflammation.

DNA Damage-Mediated Adipocyte Senescence in Pol η−/− Mice. Wenext addressed how elevated DNA damage and inflammatoryresponses lead to the observed metabolic abnormalities in polη−/− mice. As shown in Fig. 2, adipocytes from pol η−/− miceexhibited a persistently higher level of DNA damage and DNAdamage response in contrast to WT mice. Persistent DNAdamage is causally involved in cellular senescence (23) and in thesenescence-associated secretory phenotype (SASP) that resultsin the secretion of inflammatory cytokines IL-6, TNF-α, andMCP-1 (24). The observed adipocyte hypertrophy; up-regulationof p53 and p21; and higher circulating levels of IL-6, TNF-α, andMCP-1 observed with pol η−/− mice are established characteris-tics of senescence (Figs. 1 J, K, M, and N and 2 D and E and Fig.S5A). A previous report indicated that obese human subjectshave more senescent cells in their s.c. adipose tissue (25), im-plying a correlation between adipocyte senescence and obesity.To determine the senescent status in the adipose tissues, twocellular senescence markers, senescence-associated (SA) β-galactivity and p16Ink4a expression, were examined (25). Notably, asearly as 4 wk of age, which is before the onset of obesity, and upto 72 wk of age, the adipose tissue of pol η−/− mice displayedsignificantly greater amounts of SA–β-gal activity and p16Ink4a

expression compared with WT mice (Fig. 3 A–D). These dataindicate that senescence in adipose tissue precedes the devel-opment of metabolic abnormalities in pol η−/− mice. Concomi-tantly, reduced expression of the proliferation marker Ki-67 andapoptosis markers Bax, Noxa, and Puma (26) was observed (Fig.3 C and D), suggesting decreased proliferation and decreasedapoptosis in the adipose tissue from pol η−/− mice compared withWT. The role of DNA damage and adipose tissue cell senes-cence on metabolic dysfunction was further supported by theexacerbation of the metabolic abnormalities with a high-fat dietstudy. The high-calorie diet produced a similar amount of DNAdamage and prominent elevation of SA–β-gal activity within

20 wk compared with the regular chow diet after 72 wk in mice ofboth genotypes (Fig. S6).Adipose tissue consists of a complex mixture of adipocytes,

preadipocytes, macrophages, neutrophils, lymphocytes, endothe-lial cells, and stem cells (25). The differentiation of preadipocytesinto mature adipocytes plays a key role in the development ofobesity (25). Because the mature adipocytes are responsible forlipid storage, we hypothesized that pol η deficiency inducessenescence mainly in the mature adipocytes of adipose tissue.To test this hypothesis, we monitored the SA–β-gal activitythrough the differentiation of preadipocytes to mature adipocytes.Whereas no apparent SA–β-gal activity was detected in pre-adipocytes from mice of both genotypes, differentiation of thesecells into mature adipocytes resulted in increases in SA–β-gal ac-tivity and lipid accumulation with preadipocytes from pol η−/−

mice (Fig. 3 E and F). These findings suggest that mature adi-pocytes are the cells that become senescent and accumulate lipidin the adipose tissue of pol η−/− mice, suggesting senescent adi-pocytes play a causative role in the development of obesity inpol η−/− mice.

Suppression of Adipocyte Senescence Alleviates Metabolic Abnormalities.Pol η has the ability to handle various oxidative DNA lesions, in-cluding lipid peroxidation products that impede the progression ofDNA replication (13, 14). Both pol η and p53 are required for re-covery from DNA damage-stalled DNA replication forks (27). Inthe pol η-deficient XP-V cells, p53 plays a major role in mediatingthe downstream cellular events in response to DNA damage (27).The observed up-regulation and activation of p53 (Fig. 2D and E) inthe adipose tissue of pol η−/− mice also support adipose senescenceresults (Fig. 3 A and B). More importantly, p53 activation is essentialfor the establishment of cell senescence in adipose tissue and re-quired for the development of insulin resistance (28). We thereforehypothesize that suppression of p53 activity will be a reasonableapproach to examine the importance of adipocyte senescence in thedevelopment of metabolic abnormalities. This hypothesis was testedwith the p53 inhibitor pifithrin-α (PFT-α) (29). PFT-α treatmentexhibited a more significant inhibitory effect on senescence in theadipose tissue of pol η−/−mice (Fig. 3G–I) compared with WTmice.Similarly, PFT-α also has a more profound impact in reducing thebody weight, body fat accumulation, and circulating levels of insulin,and it caused a marked improvement in glucose tolerance in pol η−/−

mice (Fig. 3 J–M). These results demonstrate the importance of p53activation in mediating adipocyte senescence and suggest a criticalrole of adipocyte senescence in the observed metabolic abnormali-ties in pol η−/− mice.

Magnitude of the DNA Damage Affects Adipocyte Senescenceand Severity of Metabolic Abnormalities. To test further whetherDNA damage is causative for the development of metabolicabnormalities in pol η−/− mice, the antioxidant N-acetylcysteine(NAC) was administered to mice as a way to reduce oxidativestress and oxidative DNA lesions (30). NAC significantly re-duced the amount of DNA damage, as measured by the cometassay and the activation of ATM, γ-H2AX, p53, p21, and NF-κB.NAC also attenuated adipocyte senescence and secretion ofSASP IL-6, TNF-α, and MCP-1 (Fig. 2 B, D, and E and Fig.S5A), and reduced adipocyte hypertrophy. Moreover, with sim-ilar uptake (Fig. S5B), NAC attenuated and suppressed themetabolic abnormalities in pol η−/− mice with no apparent im-pact on the WT mice (Fig. S5 C–H). The latter is likely due to alow amount of endogenous DNA damage in the WT mice fed aregular chow diet. Similarly, administration of metformin (Met),the drug used to treat diabetes (31), also suppressed the activationof the DNA damage response, the increase in inflammatory fac-tors, and the metabolic abnormalities in the pol η−/− mice, perhapsbecause it enhances antioxidant defenses and reduces inflamma-tion (32) (Fig. 2 D and E and Fig. S5). Furthermore, both NAC

Chen et al. PNAS | Published online August 3, 2015 | E4559

GEN

ETICS

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

Fig. 3. Cellular senescence in adipose tissues and metabolic abnormalities. (A and B) SA–β-gal activity and quantitation from WT or pol η−/− mice at differentages (n = 8). (Scale bar: 1 cm.) (C and D) Expression of senescent and apoptotic proteins and quantitation (n = 6). PFT-α suppressed adipose senescence andmetabolic abnormalities. (E and F) SA–β-gal activity (Top) and Oil Red O staining (Bottom) of SVF (mainly preadipocytes) and mature adipocytes through thedifferentiation process. (Scale bars: 200 μM.) SA–β-gal activity and quantitation (n = 6) (G and H), IL-6 (I), body weight gain (J), body fat (K), plasma insulin (L),and GTT analysis (n = 6) (M) are shown. C, control; HF, high fat; P, PFT-α.The graphical data were shown as the mean ± SEM. The data were analyzed by one-way ANOVA (*P < 0.05).

E4560 | www.pnas.org/cgi/doi/10.1073/pnas.1506954112 Chen et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

and Met reduced the DNA lesions and metabolic abnormalitiesthat were induced by a high-fat diet in both WT and pol η−/− mice(Fig. S6).In contrast to the positive effects of antioxidants, sodium ar-

senite, a well-established oxidative stress inducer (33), increasedDNA lesions, elevated adipocyte senescence, and promotedmetabolic dysfunction in pol η−/− mice (Figs. S7 and S8). Thesefindings support the notion that endogenous DNA damage iscausative of the observed obesity and metabolic abnormalities ofpol η−/− mice. The opposite effects of antioxidants (NAC or Met)(Figs. S5 and S8) and oxidative stress inducers (a high-fat diet orsodium arsenite) (Figs. S6–S8) reinforce the importance of adi-pocyte senescence in responding to DNA damage and the onsetof metabolic abnormalities in pol η−/− mice. Notably, NAC andMet also suppressed DNA damage and the metabolic changesinduced by a high-fat diet in WT mice, suggesting that metabolicabnormalities may be a general response to elevated DNAdamage (Fig. S6B). In addition, because cell senescence pro-motes aging (25) and Met improves life span (32), our datasupport the evidence that Met defers the aging process (32).

Impact of Pol η on Cellular Senescence and Adipogenesis UsingKnockdown and Salvage Studies with Cells in Culture. To confirmfurther that lack of pol η activity is responsible for the cell se-nescence and metabolic changes observed in pol η−/− mice, twoadditional systems were used. First, pol η expression was down-regulated by the transfection of pol η-specific shRNA into NIH3T3L1 cells (3T3L1-shPol η) (Fig. S9A). The 3T3L1-shPol η cellsexhibited elevated amounts of DNA damage (Fig. S9B) and in-creased activation of ATM, p53, p21, γ-H2AX, and p16Ink4a ex-

pression, but they exhibited reduced expression of Ki-67 (Fig. 4Aand Fig. S9C), increased secretion of IL-6 (Fig. 4B), and greateractivity of SA–β-gal than 3T3L1-NTC-shRNA–transfected cells(Fig. 4C). Furthermore, the elevated SA–β-gal activity in the3T3L1-shPol η cells was suppressed by PFT-α (Fig. S9 D and E).The 3T3L1-pol η-shRNA–transfected cells also displayed up-regulated levels of SREBP1 and PPAR-γ and accumulated morelipid droplets (Fig. 4 A and D). Second, the effect of pol η wasexamined with a salvage study by independently transfectingpol η−/− mouse embryo fibroblast (MEF) cells with vectors thatexpress human WT pol η or an R61F mutant form of humanpol η with reduced polymerase activity (34). The expression ofWT-pol η reduced the DNA damage response, secretion of IL-6,SA–β-gal activity, and accumulation of lipid droplets (Fig. 4 E–Hand Fig. S9E). By contrast, the expression of the less active pol ηmutant R61F (34) only partially reduced these parameters (Fig.4 E–H and Fig. S9F). These findings suggest that the DNA lesionbypass activity of pol η protects adipocytes from early adipocytesenescence and lipid accumulation. In addition, because MEFcells were derived from embryos, the elevated amount of DNAdamage and SA–β-gal activity in the primary pol η−/− MEF cellscompared with WT cells further supports the notion that chronicDNA damage-mediated senescence, which occurs before theaccumulation of body fat, creates conditions that lead to obesityand metabolic abnormalities.The association between tumorigenesis and reduced genome

integrity, due to exposure to exogenous or endogenous DNAdamaging agents or genetic defects in DNA repair/replicationenzymes, is well documented for many tissues (35). However,the potential impact of genomic integrity on adipose tissue has

Fig. 4. Correlation between pol η expression, DNA damage, cell senescence, and lipid accumulation in 3T3L1 and MEF cell systems. (A–D) 3T3L1cells. (A)Quantitation of expression and activation of DNA damage, senescence, and adipogenic protein markers. (B) IL-6 secretion. (C) SA–β-gal activity. (D) Oil Red Ostaining (n = 6). (E–H) MEF cells. (E) Quantitation of expression and activation of DNA damage, senescence, and adipogenic protein markers in WT, pol η−/−,pol η−/−-hPol η WT, and pol η−/−-hPol η R61F MEF cells. (F) IL-6 secretion (n = 6). (G) SA–β-gal activity. (H) Oil Red O staining. The images presented were in thesame scale. The graphical data are shown as the mean ± SEM. The data were analyzed by the Student t test (*P < 0.05).

Chen et al. PNAS | Published online August 3, 2015 | E4561

GEN

ETICS

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

received little attention, at least in part because adipose tissuetumors are usually benign. Adipose tissue is vital for energyhomeostasis and serves as an essential endocrine and immuneorgan (25). Adipose tissue is the main site for lipid peroxidationand acts as a major “sink” to store and handle free radical-induced peroxidation products (36). Oxidative DNA lesions pro-duced by lipid peroxidation products block DNA replicativepolymerases and stall the progression of DNA replication (13,14), which triggers cell cycle arrest and leads to different cellularresponses, depending on the magnitude of DNA damage. MildDNA damage causes temporary pausing of the cell cycle to allowthe DNA repair system to repair the damage. Medium or severeDNA damage induces cellular senescence or cell death (apo-ptosis), which also depends on the tissue type. Our results sup-port previous studies that have shown pol η bypasses oxidativeDNA lesions (13, 14) because pol η−/− mice exhibit an increasedamount of DNA lesions and elevated DNA damage responses.The increase in SA–β-gal activity; up-regulation of p53, p16, andp21; and increased SASP circulating and cellular levels of IL-6and TNF-α that occur at a very early age in pol η−/− mice provideevidence for a causative role of adipocyte senescence in obesityand the development of insulin resistance. These results alsoimply that senescence is the primary response to DNA damage inadipose tissue, which may explain the low incidence of malignantcancer in adipose tissue, because cellular senescence counteractstumorigenesis. Although increased inflammatory responses, in-cluding infiltration of macrophages into the adipose tissue andincreased levels of IL-6 or TNF-α, are commonly observed inadipose tissue from obese subjects or mouse models (37), theseobservations may be a consequential phenomenon, because tis-sue samples were harvested from subjects or mice with anestablished obesity phenotype. Deficiency of the DNA repairenzyme Neil1 or OGG1 also induces the metabolic syndromeand obesity in mice (3, 4), at least in part because of impairedmitochondrial function due to DNA damage (3, 4). Given theessential role of mitochondria in energy and ROS production, itwill be worthwhile to investigate the mitochondrial function ofthese DNA repair/replication mice at different ages to understandits correlation with the development of the metabolic syndrome.In summary, our findings suggest that pol η plays an important

role in maintaining genome stability and preventing cellular se-nescence in adipose tissue. This study extends the function of polη from its established role in suppressing tumorigenesis in sun-exposed skin to reducing the negative effects of endogenousDNA damage on adipose tissue metabolism. It would be im-portant to investigate the detailed mechanisms of how the de-ficiency of pol η increased DNA damage in adipose tissue andthe types of lesions that contribute to the observed metabolicabnormalities. It is not clear from the available literature whetherpatients with XP-V are prone to the development of obesity andthe metabolic syndrome. Given that variation in the expressionlevel of the XP-V gene affects the sensitivity to DNA-damagingagents in various tissues (10, 38–40), an epidemiological study willbe necessary to examine the correlation between the pol η ex-pression level and the metabolic syndrome. In addition to pol η,other translesion synthesis polymerases have been shown to han-dle various types of DNA lesions (41), and it will be interesting todetermine whether KO mice for other lesion bypass polymerasesare prone to obesity and metabolic abnormalities. The currentresults also suggest that senescence in adipose tissue caused bychronic DNA damage plays a causative role in the development ofmetabolic abnormalities. Our data with mice suggest that adipo-cyte senescence plays an important role in the development ofobesity. Because senescent preadipocytes are present in severelyobese subjects (25), the molecular mechanism responsible for se-nescence cells in adipose tissue requires additional study. Ourmodel of adipose tissue senescence and obesity due to DNAdamage provides an additional testable paradigm for other DNA

repair enzyme-deficient mice that develop obesity, such as Neil1and OGG1 (3, 4).Studies have been conducted with pol η−/− mice to determine

the importance of pol η on mutation frequency, carcinogenesis,and somatic hypermutation of Ig genes (42–44). Most of thestudies were performed with mice at an age (2–6 mo) at whichthe metabolic abnormalities observed in the present study couldhave been easily missed (42–44). Mice were followed for 24 moin one study, but the focus was on mutagenesis, and neither bodyweights nor metabolic abnormalities were reported (45). Themetabolic abnormalities that occur in pol η−/− mice develop withage, and the phenotype is less severely affected than in theleptin-deficient (ob/ob) or leptin-receptor–mutated (db/db)mouse model (46, 47). The findings with pol η−/− mice suggestthat the vulnerability for metabolic abnormalities is increased bya loss of genome integrity with age. Our findings also suggest thatDNA damage-induced adipocyte senescence and metabolic ab-normalities are not limited to genetic defects due to DNAdamage repair or response enzymes. A high-fat or high-sugardiet and environmental toxicants that have a direct or indirectimpact on genome integrity may also cause obesity and themetabolic syndrome by inducing DNA damage and adipocytesenescence. On the other hand, diets that are rich in antioxidantcomponents may prevent oxidative DNA lesions and the devel-opment of senescent adipocytes, which may prevent the devel-opment of obesity and the metabolic syndrome in response toproinflammatory diets and environmental toxins. These findings,together with other DNA repair enzyme-defective mouse models(3, 4, 6–8), suggest that reduction of genome integrity, regardlessof whether it is induced by defective DNA repair or exogenousagents, contributes causatively to the development of the meta-bolic syndrome. Given that most people become overweight ormoderately obese with age rather than severely obese (48), themechanistic relationship between loss of genome integrity anddevelopment of the metabolic syndrome with age needs to befurther studied. New insight may also be gained into the ap-parent link between obesity and cancer.

Materials and MethodsAnimals. All animal experimental protocols were approved by the InstitutionalAnimalCareandUseCommittee, IndianaUniversity SchoolofMedicine.WTC57BL/6mice were purchased from The Jackson Laboratory. Pol η−/− mice were obtainedfrom the National Cancer Institute (B6;129-Polhtm1Rak/NCI) and backcrossed for 12generations with WT C57BL/6 mice. All experiments were conducted with malelittermates maintained in the animal facility at 25 °C with a 12-h light/dark cycle.For bodyweight gain and body fat percentagemeasurements, mice wereweighedweekly, and weight gain was calculated by subtracting the original weight at 8 wkof age from the weight of the individual mice. Body fat content was analyzedusing dual-energy X-ray absorptiometry scanning (PIXImus β mouse densitometer;Lunar Corp.). Treatments were startedwhenmice were 8wk of age, with a normalchow diet, high-fat diet (Harlan Laboratories, Inc.), or high-fructose diet (normalchow with 20% fructose in drinking water). Agents, including NAC (1 mg/mL;wt/vol), Met (1.5 mg/mL; wt/vol), or sodium arsenite (2.5 mg/L; wt/vol), were ad-ministered through the drinking water. PFT-α (2.2 mg/kg; wt/body wt; Sigma–Aldrich) was injected i.p. three times a week for 4 wk. For histological analysis, micewere euthanized with CO2, and tissues were harvested and fixed in 4% parafor-maldehyde (PFA). Samples were subjected to H&E staining for visualization. Thecross-sectional area of fat tissue was used to measure the size of adipocytes uti-lizing ImageJ software (NIH). For Oil Red O staining, fresh liver tissue was immersedand frozen in optimal cutting temperature compound, followed by sectioning andstaining with Oil Red O and Harris hematoxylin (49). For energy expendituremeasurement, mice were individually housed in metabolic chambers (Lab Master;TSE Systems). Oxygen consumption, respiratory exchange ratio, and food/drinkintake were recorded and calculated as previously described (50, 51).

Blood Glucose and Glucose Tolerance Test. Mice were fasted overnight beforea glucose tolerance test was performed. Blood glucose was measured using aglucometer (Contour; Bayer Healthcare) at 0, 30, 60, 90, and 120min after i.p.injection of glucose (2 mg/g). The data were analyzed by one-way ANOVAusing GraphPad Prism 5.0 software (GraphPad Software), and the area underthe curve was calculated by the trapezoid method.

E4562 | www.pnas.org/cgi/doi/10.1073/pnas.1506954112 Chen et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

ELISA.Mice were fasted overnight, and plasma samples were collected from afacial vein for ELISA analyses using ultrasensitive mouse leptin and insulinELISA kits (Crystal Chem, Inc.). The mouse IL-6 and TNF-α kits were purchasedfrom Biolegend, Inc., and the mouse adiponectin ELISA kit was obtainedfrom Life Technologies Invitrogen. The HOMA-IR was calculated from theproduct of fasting insulin concentration (micro-international units per mil-liliter) and plasma glucose (milligrams per deciliter) divided by 405 (52).

Isolation of Adipocytes and Macrophage Detection. Briefly, collected fat padswere minced and digested with type I collagenase before centrifugationusing similar procedures as described (53). The pellet is the stromal vascularfraction (SVF), and the floating cells were collected as the mature adipocytefraction. The mature adipocyte fraction was collected and incubated withAlexa Fluor 488-conjugated anti-mouse F4/80 and R-PE–conjugated anti-mouse CD11b (Molecular Probes, Life Technologies) before being analyzedby flow cytometry (FACSCalibur; BD Biosciences). Cells stained with bothF4/80+ and CD11b+ were defined as the macrophage fraction. The amountof macrophages was calculated as the percentage of F4/80+/CD11b+ amongall live cells.

Preparation of Protein Lysates and mRNA for Real-Time PCR. Adipose tissueswere homogenized with modified radioimmunoprecipitation assay buffer,followed by centrifugation at 4 °C. The fat layer was removed, and cell ly-sates were subjected to Western blot analysis. Total RNA was isolated frommature adipose tissue fractions using TRIzol (Molecular Research Center,Inc.), and cDNA was prepared using a QuantiTect Reverse Transcription kitand a QuantiFast SYBR Green PCR kit (both from Qiagen). Real-time PCR wasperformed using a StepOne Plus Real-Time PCR System (Applied Biosystems).The mouse primers were designed and validated using a QuantiTect PrimerAssay System (Qiagen). Relative mRNA expression represents the target genemRNA level compared with the corresponding control mRNA level.

Western Blot Analysis and Immunofluorescence Staining. Whole-cell lysates(50 μg) were resolved by SDS/PAGE and transferred onto nitrocellulosemembranes (Bio-Rad) before being incubated with primary antibodies,followed by addition of HRP-conjugated secondary antibody (JacksonImmuno Research Lab). Specific proteins were detected using SuperSignalWest Femto chemiluminescence substrates (Thermo Scientific) and docu-mented with a Fujifilm LAS-4000 Image Analyzer (GE Life Science). Theprimary antibodies used for Western blotting were the following anti-mouse antibodies: anti-ATM, anti–phospho-γ-H2AX, and anti–Parp-1 an-tibodies from Cell Signaling Technology; anti–phospho-ATM (Ser1981) andanti–Ki-67 antibodies from Millipore; anti-p53, anti–phospho-p53 (Ser15),anti-p21, anti–phospho-p21, anti–PPAR-γ, anti-Bax, anti-Noxa, anti-Puma,anti-Actinin, and anti-Actin antibodies from Sigma–Aldrich; and anti-p16Ink4a from Abcam; anti–SREBP1-p68, anti–SREBP1-p125, anti–NF-κB-p65, anti–phospho-NF-κB-p65, anti-IκBα, and anti–phospho-IκBα anti-bodies from Santa Cruz Biotechnology. The samples derived from the sameexperiment, and the blots were processed in parallel. Loading controls,positive and negative controls, and molecular markers were run on thesame blot, and the loading controls were used for normalization. Thequantitation of protein expression represents relative protein level com-

pared with the corresponding control after normalization with actin oractinin (ImageJ software). Blots are representative of six independentexperiments. For immunofluorescence staining, cells were fixed with 4%PFA, followed by incubation in permeabilization buffer (0.5% saponin,0.1% BSA, and 1% NaN3 in PBS). The cells were then stained with rabbitanti-mouse phospho-γ-H2AX and DyLight488–anti-rabbit IgG (JacksonImmune Research laboratory) before monitoring under a Leica florescencemicroscope (Leica DMI6000B) using Leica imaging LAS.AF software. Theimages represent results from four independent experiments.

Single-Cell Gel Electrophoresis Assay (Comet Assay) and Immunohistochemistry.Adipocytes isolated from the adipose tissue of WT and pol η−/− mice (53) weresubjected to a single-cell gel electrophoresis assay (Trevigen, Inc.). The data,which were analyzed by CometScore 1.5 software (Triek Corp.), are presentedas the percentage of DNA in the tail. For detection of 8-oxoG, the fixed adi-pose tissue sections were deparaffinized and rehydrated before being in-cubated with mouse anti–8-oxoG antibody (Trevigen, Inc.), followed byincubation with biotin-conjugated anti-mouse IgG. The 3.3′-diaminobenzidine(Thermo Scientific) reaction was used to visualize the bound secondary anti-body. The sections were counterstained with methyl green for the cell nucleus.Images were acquired using a Leica microscope (DM2500) and Leica imagingLAS software.

SA–β-Gal Staining and Apoptosis Assessment. SA–β-gal activity in visceraladipose tissues fromWT and pol η−/− mice was determined using an SA–β-galstaining kit from Cell Signaling Technology. The percentage of cells positivefor SA–β-gal was quantitated as described previously (28).

Cell Culture. Murine 3T3-L1 cells (American Type Culture Collection) weretransfected with nontarget shRNA control or shRNA (Sigma–Aldrich) togenerate 3T3L1-NTC and 3T3-sh-pol η cell lines. Embryos from WT and polη−/− mice were harvested on day 13.5 for the generation of MEF cells (54).The established pol η−/− MEF cells were transfected with either WT humanpol η (hPol η WT) or the R61F mutant form of human pol η (hPol η R61F) (55).The protein expression level of pol η was examined by Western blot analysiswith anti-human or anti-mouse pol η antibodies from Biorbyt. The harvestedSVF cells from mice adipose tissue, 3T3L1 transfectants, and MEF cells werecultured and differentiated into adipocytes (56), and then examined for SA–β-gal staining and Oil Red O staining.

Statistical Analyses. The staining images and blots represent results from fourto six independent experiments. The graphical data, which are shown as themean ± SEM, were analyzed by one-way ANOVA for multiple comparisontests or by the Student t test for individual pairwise comparisons. Thethreshold for statistical significance was P < 0.05 vs. the correspondingcontrol values. Analyses were performed with statistical software (GraphPadPrism 5.0).

ACKNOWLEDGMENTS. We thank Dr. Richard Honkanen (University of SouthAlabama) for suggestions and comments during the study. This study wassupported, in part, by NIH Grant R01 CA 112446 (to K.-m.C.) and a VeteransAffairs Merit Award (to R.A.H.).

1. Korwek Z, Alster O (2014) [The role of the DNA damage response in apoptosis and cell

senescence.] Postepy Biochem 60(2):248–262. Polish.2. Ishida T, Ishida M, Tashiro S, Yoshizumi M, Kihara Y (2014) Role of DNA damage in

cardiovascular disease. Circ J 78(1):42–50.3. Vartanian V, et al. (2006) The metabolic syndrome resulting from a knockout of the

NEIL1 DNA glycosylase. Proc Natl Acad Sci USA 103(6):1864–1869.4. Sampath H, et al. (2012) 8-Oxoguanine DNA glycosylase (OGG1) deficiency increases

susceptibility to obesity and metabolic dysfunction. PLoS One 7(12):e51697.5. Sampath H, et al. (2011) Variable penetrance of metabolic phenotypes and devel-

opment of high-fat diet-induced adiposity in NEIL1-deficient mice. Am J Physiol

Endocrinol Metab 300(4):E724–E734.6. Karakasilioti I, et al. (2013) DNA damage triggers a chronic autoinflammatory re-

sponse, leading to fat depletion in NER progeria. Cell Metab 18(3):403–415.7. Schneider JG, et al. (2006) ATM-dependent suppression of stress signaling reduces

vascular disease in metabolic syndrome. Cell Metab 4(5):377–389.8. Fang EF, et al. (2014) Defective mitophagy in XPA via PARP-1 hyperactivation and

NAD(+)/SIRT1 reduction. Cell 157(4):882–896.9. Masutani C, et al. (1999) Xeroderma pigmentosum variant (XP-V) correcting protein

from HeLa cells has a thymine dimer bypass DNA polymerase activity. EMBO J 18(12):

3491–3501.10. Thakur M, et al. (2001) DNA polymerase eta undergoes alternative splicing, protects

against UV sensitivity and apoptosis, and suppresses Mre11-dependent recombination.

Genes Chromosomes Cancer 32(3):222–235.

11. Chen YW, Cleaver JE, Hanaoka F, Chang CF, Chou KM (2006) A novel role of DNApolymerase eta in modulating cellular sensitivity to chemotherapeutic agents. MolCancer Res 4(4):257–265.

12. Rey L, et al. (2009) Human DNA polymerase eta is required for common fragile sitestability during unperturbed DNA replication. Mol Cell Biol 29(12):3344–3354.

13. Haracska L, Yu SL, Johnson RE, Prakash L, Prakash S (2000) Efficient and accuratereplication in the presence of 7,8-dihydro-8-oxoguanine by DNA polymerase eta. NatGenet 25(4):458–461.

14. Yang IY, Hashimoto K, de Wind N, Blair IA, Moriya M (2009) Two distinct translesionsynthesis pathways across a lipid peroxidation-derived DNA adduct in mammaliancells. J Biol Chem 284(1):191–198.

15. Lin Q, et al. (2006) Increased susceptibility to UV-induced skin carcinogenesis in po-lymerase eta-deficient mice. Cancer Res 66(1):87–94.

16. Sell H, Habich C, Eckel J (2012) Adaptive immunity in obesity and insulin resistance.Nat Rev Endocrinol 8(12):709–716.

17. Austyn JM, Gordon S (1981) F4/80, a monoclonal antibody directed specifically againstthe mouse macrophage. Eur J Immunol 11(10):805–815.

18. Miard S, Fajas L (2005) Atypical transcriptional regulators and cofactors of PPAR-gamma. Int J Obes 29(Suppl 1):S10–S12.

19. Furukawa S, et al. (2004) Increased oxidative stress in obesity and its impact onmetabolic syndrome. J Clin Invest 114(12):1752–1761.

20. Burma S, Chen BP, Murphy M, Kurimasa A, Chen DJ (2001) ATM phosphorylates his-tone H2AX in response to DNA double-strand breaks. J Biol Chem 276(45):42462–42467.

Chen et al. PNAS | Published online August 3, 2015 | E4563

GEN

ETICS

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

21. Piret B, Schoonbroodt S, Piette J (1999) The ATM protein is required for sustainedactivation of NF-kappaB following DNA damage. Oncogene 18(13):2261–2271.

22. Jacobs MD, Harrison SC (1998) Structure of an IkappaBalpha/NF-kappaB complex. Cell95(6):749–758.

23. Fumagalli M, Rossiello F, Mondello C, d’Adda di Fagagna F (2014) Stable cellular se-nescence is associated with persistent DDR activation. PLoS One 9(10):e110969.

24. Rodier F, et al. (2009) Persistent DNA damage signalling triggers senescence-associ-ated inflammatory cytokine secretion. Nat Cell Biol 11(8):973–979, and correction(2009) 10:1272.

25. Tchkonia T, et al. (2010) Fat tissue, aging, and cellular senescence. Aging Cell 9(5):667–684.

26. Athar M, Back JH, Kopelovich L, Bickers DR, Kim AL (2009) Multiple molecular targetsof resveratrol: Anti-carcinogenic mechanisms. Arch Biochem Biophys 486(2):95–102.

27. Cleaver JE, et al. (2002) Polymerase eta and p53 jointly regulate cell survival, apoptosisand Mre11 recombination during S phase checkpoint arrest after UV irradiation. DNARepair (Amst) 1(1):41–57.

28. Minamino T, et al. (2009) A crucial role for adipose tissue p53 in the regulation ofinsulin resistance. Nat Med 15(9):1082–1087.

29. Komarov PG, et al. (1999) A chemical inhibitor of p53 that protects mice from the sideeffects of cancer therapy. Science 285(5434):1733–1737.

30. De Flora S, Izzotti A, D’Agostini F, Balansky RM (2001) Mechanisms of N-acetylcysteinein the prevention of DNA damage and cancer, with special reference to smoking-related end-points. Carcinogenesis 22(7):999–1013.

31. Rojas LB, Gomes MB (2013) Metformin: An old but still the best treatment for type 2diabetes. Diabetol Metab Syndr 5(1):6.

32. Martin-Montalvo A, et al. (2013) Metformin improves healthspan and lifespan inmice. Nat Commun 4:2192.

33. Flora SJ (2011) Arsenic-induced oxidative stress and its reversibility. Free Radic BiolMed 51(2):257–281.

34. Katafuchi A, et al. (2010) Critical amino acids in human DNA polymerases eta andkappa involved in erroneous incorporation of oxidized nucleotides. Nucleic Acids Res38(3):859–867.

35. Bartsch H, Dally H, Popanda O, Risch A, Schmezer P (2007) Genetic risk profiles forcancer susceptibility and therapy response. Recent Results Cancer Res 174:19–36.

36. Murdolo G, et al. (2013) Oxidative stress and lipid peroxidation by-products at thecrossroad between adipose organ dysregulation and obesity-linked insulin resistance.Biochimie 95(3):585–594.

37. Bai Y, Sun Q (2015) Macrophage recruitment in obese adipose tissue. Obes Rev 16(2):127–136.

38. Zhou W, et al. (2013) Expression of DNA translesion synthesis polymerase η in headand neck squamous cell cancer predicts resistance to gemcitabine and cisplatin-basedchemotherapy. PLoS One 8(12):e83978.

39. Teng KY, et al. (2010) DNA polymerase η protein expression predicts treatment re-sponse and survival of metastatic gastric adenocarcinoma patients treated withoxaliplatin-based chemotherapy. J Transl Med 8:126.

40. Ceppi P, et al. (2009) Polymerase eta mRNA expression predicts survival of non-small

cell lung cancer patients treated with platinum-based chemotherapy. Clin Cancer Res

15(3):1039–1045.41. Broyde S, Wang L, Rechkoblit O, Geacintov NE, Patel DJ (2008) Lesion processing:

High-fidelity versus lesion-bypass DNA polymerases. Trends Biochem Sci 33(5):

209–219.42. Delbos F, Aoufouchi S, Faili A, Weill JC, Reynaud CA (2007) DNA polymerase eta is the

sole contributor of A/T modifications during immunoglobulin gene hypermutation in

the mouse. J Exp Med 204(1):17–23.43. Martomo SA, Saribasak H, Yokoi M, Hanaoka F, Gearhart PJ (2008) Reevaluation of

the role of DNA polymerase theta in somatic hypermutation of immunoglobulin

genes. DNA Repair (Amst) 7(9):1603–1608.44. Stallons LJ, McGregor WG (2010) Translesion synthesis polymerases in the prevention

and promotion of carcinogenesis. J Nucleic Acids, 10.4061/2010/643857.45. Busuttil RA, Lin Q, Stambrook PJ, Kucherlapati R, Vijg J (2008) Mutation frequencies

and spectra in DNA polymerase eta-deficient mice. Cancer Res 68(7):2081–2084.46. Chen H, et al. (1996) Evidence that the diabetes gene encodes the leptin receptor:

Identification of a mutation in the leptin receptor gene in db/db mice. Cell 84(3):

491–495.47. Pelleymounter MA, et al. (1995) Effects of the obese gene product on body weight

regulation in ob/ob mice. Science 269(5223):540–543.48. Finkelstein EA, et al. (2012) Obesity and severe obesity forecasts through 2030. Am J

Prev Med 42(6):563–570.49. Novikoff AB, Novikoff PM, Rosen OM, Rubin CS (1980) Organelle relationships in

cultured 3T3-L1 preadipocytes. J Cell Biol 87(1):180–196.50. Weir JB (1949) New methods for calculating metabolic rate with special reference to

protein metabolism. J Physiol 109(1-2):1–9.51. Bruss MD, Khambatta CF, Ruby MA, Aggarwal I, Hellerstein MK (2010) Calorie re-

striction increases fatty acid synthesis and whole body fat oxidation rates. Am J

Physiol Endocrinol Metab 298(1):E108–E116.52. Matthews DR, et al. (1985) Homeostasis model assessment: Insulin resistance and

beta-cell function from fasting plasma glucose and insulin concentrations in man.

Diabetologia 28(7):412–419.53. Fernyhough ME, et al. (2004) Primary adipocyte culture: Adipocyte purification

methods may lead to a new understanding of adipose tissue growth and develop-

ment. Cytotechnology 46(2-3):163–172.54. Lei Y (2013) Generation and culture of mouse embryonic fibroblasts. Methods Mol

Biol 1031:59–64.55. Chen YW, et al. (2008) Human DNA polymerase eta activity and translocation is

regulated by phosphorylation. Proc Natl Acad Sci USA 105(43):16578–16583.56. Elmendorf JS, Chen D, Pessin JE (1998) Guanosine 5′-O-(3-thiotriphosphate) (GTPgammaS)

stimulation of GLUT4 translocation is tyrosine kinase-dependent. J Biol Chem 273(21):

13289–13296.

E4564 | www.pnas.org/cgi/doi/10.1073/pnas.1506954112 Chen et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020