p53 is required for chloroquine-induced atheroprotection but not insulin sensitization.

An intact genotoxic stress response appears to be atheroprotective and insulin sensitizing. ATM, mutated in ataxia telangiectasia, is critical for the genotoxic stress response, and its deficiency is associated with accelerated atherosclerosis and insulin resistance in humans and mice. The antimalarial drug chloroquine activates ATM signaling and improves metabolic phenotypes in mice. p53 is a major effector of ATM signaling, but it is unknown if p53 is required for the beneficial effects of chloroquine. We tested the hypothesis that the cardiometabolic effects of chloroquine are p53-dependent. ApoE-null mice with or without p53 were treated with low-dose chloroquine or saline in the setting of a Western diet. After 8 weeks, there was no p53-dependent or chloroquine-specific effect on serum lipids or body weight. Chloroquine reduced plaque burden in mice wild-type for p53, but it did not decrease lesion extent in p53-null mice. However, chloroquine improved glucose tolerance, enhanced insulin sensitivity, and increased hepatic Akt signaling regardless of the p53 genotype. These results indicate that atheroprotection induced by chloroquine is p53-dependent but the insulin-sensitizing effects of this agent are not. Discrete components of the genotoxic stress response might be targeted to treat lipid-driven disorders, such as diabetes and atherosclerosis.

About 13% of American adults have diabetes ( ‫ف‬ 40% are undiagnosed) and an additional 30% have prediabetes (impaired fasting glucose or impaired glucose tolerance) ( 1 ). These prevalence numbers, which have increased sub-regular insulin (0.75 units/kg body weight) (Eli Lilly and Co., Indianapolis, IN). Tail vein blood for these tests was assayed at 0, 30, 60, and 120 min using a glucose meter (Contour, Bayer Healthcare, Mishawaka, IN). A published formula [0.25 (fasting value) + 0.5 (1/2-h value) + 0.75 (1-h value) + 0.5 (2-h value)] ( 15 ) was used to calculate area under the curve (AUC) for each GTT and ITT. The en face and aortic root cross-section techniques previously described ( 16 ) were used to quantify atherosclerosis throughout the aorta as well as at the vessel origin. En face regions were analyzed as the aortic arch (from the aortic valve to the left subclavian artery), thoracic aorta (from the left subclavian artery to the fi nal intercostal arteries), and abdominal aorta (from the intercostals arteries to the bifurcation of the iliac arteries).

Macrophage culture
Macrophages were isolated essentially as described ( 17 ) by injecting mice intraperitoneally with a 4% solution of thioglycollate media (Sigma, St. Louis, MO) on day 1. On day 5, peritoneal macrophages were isolated, washed, counted, and plated in RPMI with 10% FBS. Macrophages were subsequently harvested at various times to isolate protein using standard techniques.

Statistical analyses
Statistical signifi cance of differences was calculated using the Student t -test for parametric data (Western blot densitometric quantifi cation, serum metabolites, GTT, and ITT results) or the Mann-Whitney test for nonparametric data (atherosclerosis quantifi cation).

RESULTS
Because macrophages are involved in atherosclerosis as well as insulin resistance ( 6 ), the stress kinase JNK has been shown to promote atherosclerosis ( 18,19 ), and the increased JNK signaling in ATM-defi cient macrophages is suppressed by chloroquine in an ATM-dependent manner ( 9 ), the absence of p53 might be expected to be associated with hyperactive JNK signaling in macrophages like that seen with ATM defi ciency. We assessed this possibility by comparing Thr183/Tyr185 JNK phosphorylation, known JNK-activation sites ( 20 ), in thioglycollate-elicited peritoneal macrophages from our wild-type and p53-defi cient mice. p53 defi ciency was associated with increased JNK phosphorylation following treatment with either of two classic proinfl ammatory stimuli, LPS and TNF ␣ ( Fig. 1 ). A similar response was seen with ATM-null macrophages (see supplemental Fig. I ), suggesting that ATM defi ciency and p53 defi ciency share proinfl ammatory and proatherogenic antecedents and raising the possibility that the presence of p53 is critical for propagating a chloroquine-initiated ATM signal that is atheroprotective. of the metabolic syndrome. We recently demonstrated that the ATM activator chloroquine, safely used in humans for more than a century to treat malaria as well as rheumatoid arthritis and systemic lupus erythematosus, decreases atherosclerosis and improves insulin sensitivity in several diet-induced and genetic murine models ( 9 ). The chloroquine effect appears to involve Jun N-terminal kinase (JNK), a major mediator of infl ammation, insulin resistance, and vascular disease, because chloroquine suppresses macrophage JNK activation in an ATM-dependent manner ( 9 ). p53, dubbed "guardian of the genome," ( 12 ) is a major downstream effector of ATM signaling ( 3 ). Because much of the genotoxic stress response fl ows through the ATM/ p53 axis, p53 might be expected to play an important role in the cardiometabolic consequences of insulin resistance and diabetes. Aside from the observation that p53-null mice are susceptible to atherosclerosis ( 13 ), it is not known if p53 mediates the relationship between insulin resistance and vascular disease. To test the hypothesis that the cardiometabolic effects of the ATM activator chloroquine are p53-dependent, we compared the effects of chronic administration of chloroquine in apoE-null mice that were wild-type for p53 and apoE-null mice that were p53defi cient.

Animals
The Washington University Animal Studies Committee approved protocols, and this research was conducted in conformity with the PHS Policy on Humane Care and Use of Laboratory Animals. p53 +/ Ϫ mice (Taconic Laboratories) were backcrossed with apoE Ϫ / Ϫ mice (C57BL/6 background) to generate p53 Ϫ / Ϫ apoE null and p53 +/+ apoE null littermates used for experiments that were >N7 C57BL/6. p53 genotyping utilized forward and reverse wild-type primers (5 ′ -GTGTTTCATTAGTTCCCCACCTTGAC-3 ′ and 5 ′ -ATGGGAGGCTGC CAGTCCTAACCC-3 ′ , respectively) and forward and reverse knockout primers (5 ′ -GTGGGAGGGA-CAAAAGT TCGAGGCC-3 ′ and 5 ′ -TTTACGGAGCCCTGGC G-CTCGATGT-3 ′ , respectively). All experiments were performed with male mice, in part due to a low yield of p53 Ϫ / Ϫ apoE Ϫ / Ϫ female births. ATM Ϫ / Ϫ and ATM +/+ mice (in the apoE-null background) were genotyped as described ( 9 ). Mice housed in a specifi c pathogen-free barrier facility were weaned at 3 weeks of age to a standard mouse chow providing 6% calories as fat, then started on a Western-type diet containing 0.15% cholesterol providing 42% calories as fat (TD 88137, Harlan, Madison, WI) at ‫ف‬ 8 weeks of age. At the time of initiation of the experimental diet, mice were treated with either 0.9% saline (control) or chloroquine diphosphate (Sigma, St. Louis, MO) in saline at 3.5 mg/ kg body weight 2 times per week after a loading dose of 7 mg/kg. Chloroquine or saline was continued with the diet intervention for a total of 8 weeks.

Analytical procedures and lesion quantifi cation
Cholesterol, triglycerides, glucose, and free (nonesterifi ed) fatty acids (FFA) were assayed in serum following a 4-h fast as described ( 14 ). Glucose tolerance tests (GTT) and insulin tolerance tests (ITT) were separated by at least a week using fasted mice that were injected with 10% D-glucose (1 g/kg) or human

p53-defi ciency and chloroquine do not affect weight gain or serum metabolites
Over the 8 weeks of high-fat diet feeding, all of the mice were overtly normal in appearance, and each experimental group manifested similar weight gain (see supplemental Fig. III ). There were no signifi cant differences between experimental groups for fasting levels of common metab-To directly address the potential p53-dependence of chloroquine on atherosclerosis as well as insulin resistance, we used the 2 × 2 study design shown in Fig. 2 . ApoE-null mice, roughly half that were p53 +/+ ( Fig. 2, top left ) and half that were p53 Ϫ / Ϫ ( Fig. 2, top right ), were started on a Western-type diet and randomized to receive biweekly injections of placebo versus a low dose of chloroquine previously shown to decrease atherosclerosis and cause insulin sensitization in apoE-null mice ( 9 ). Serum, collected at baseline and after 1 and 2 months of the experimental diet, was analyzed for several metabolic parameters. The primary endpoints of this study were insulin sensitivity (determined by insulin tolerance/glucose tolerance testing and confi rmed by Akt signaling) and the extent of aortic atherosclerosis (determined by assessment of both en face and aortic root lesions).

Chloroquine activates the ATM/p53 axis in vivo
Chronic administration of low-dose chloroquine to mice activates the ATM/p53 axis in many tissues, including adipose, liver, and the vasculature ( 9 ). To confi rm that mice in these experiments were achieving the same effects, we immunoblotted tissues for phosphorylation of p53 on its Serine-15 residue (a specifi c phosphorylation site for ATM) ( 21 ). Supplemental Fig. II shows representative blots of increased p53-phosphorylation in the adipose tissue of chloroquine-treated wild-type mice compared with saline-treated mice, indicating that active treatment mice were indeed receiving the drug and that there was an impact on the ATM/p53 axis in these animals.  of high-fat feeding. Glucose excursions were similarly lower during glucose tolerance testing in both wild-type and p53null cohorts treated with chloroquine, and area under the curve values were signifi cantly lower in the chloroquine groups ( Fig. 4A , B ). Hypoglycemic responses were similarly lower during insulin tolerance testing in both wild-type and p53-null cohorts treated with chloroquine, and area under the curve values were signifi cantly lower in the chloroquine groups ( Fig. 4C, D ). Because the liver is the major insulin-sensitizing site with activation of ATM, we assayed hepatic Akt, a major mediator of insulin sensitivity in vivo, and one of its downstream targets, GSK3 ␤ . As determined by phosphorylation status, Akt was activated (phosphorylated Akt is active), and GSK3 ␤ was inhibited (phosphorylated GSK3 ␤ is inactive) to the same extent in the livers of chloroquine-treated wild-type and p53-null mice ( Fig. 4E, F ). There were no signifi cant differences in glucose tolerance testing or insulin tolerance testing between salineinjected p53-null mice compared with their wild-type counterparts [see supplemental Fig. IV(A), (B )].

p53 is required for chloroquine-induced atheroprotection
Both ATM-and p53-null mice have been shown to have increased atherosclerosis ( 9,10,22,23 ), and activation of olites (cholesterol, triglycerides, glucose, and FFA) at baseline (just before initiation of the diet and chloroquine injections, Time 0 of Fig. 3 ). As shown in Fig. 3A , Westerndiet feeding increased fasting cholesterol levels to >1,000 mg/dl in all cohorts. Fasting levels of triglycerides, glucose, and FFA also increased with diet intervention and resultant weight gain ( Fig. 3B-D ). There was no p53 genotype effect and no chloroquine effect on serum metabolites at baseline or with the high-fat diet. Fasting glucose levels tended to be lower with chloroquine treatment for both wild-type and p53-null cohorts, but these results were not statistically signifi cant ( Fig. 3C ). This is an expected observation in mice as, in the absence of severe defects in insulin sensitivity (e.g., ob/ob mice), chloroquine effects on fasting glucose are modest while effects on glucose tolerance and insulin tolerance are more pronounced ( 9 ).

p53 is not required for chloroquine-induced insulin sensitization
Chloroquine is an ATM-dependent insulin sensitizer in mice as the insulin-sensitizing response in glucose tolerance tests is abrogated in ATM-null animals ( 9 ). To determine if p53 is involved in this response, glucose tolerance and insulin tolerance testing was performed after 8 weeks ter 8 weeks of high-fat feeding. Plaque burden expressed as a percentage of intimal involvement is shown for the total aorta ( Fig. 5A , left panels) and for each of three, separate aortic regions ( Fig. 5A , right panels). The degree of athero-ATM with chloroquine decreases atherosclerotic burden in mice ( 9 ). To determine if the latter effect is p53-dependent, we quantifi ed atherosclerosis in whole mount aortas as well as the aortic roots of our wild-type and p53-null cohorts af-  -473), total-Akt, phospho-GSK3 ␤ (Ser-9), and total-GSK3 ␤ from liver tissue homogenates of WT and KO mice after 2 months of chloroquine versus saline injections. * P < 0.05 for AUC data. All results are presented as mean ± SEM. AUC, area under the curve; CHQ, chloroquine; KO, knockout; WT, wild type. efi ts may not accrue in other contexts. The opposite intervention, p53 inhibition, limits ischemia/reperfusion injury in heart and brain ( 32,33 ). Given the broad spectrum of downstream targets of p53, identifying potentially useful distal arms of the genotoxic stress response, such as MAP kinase pathways, could be more likely to have therapeutic utility. In nonmacrophage tissues, p53 is thought to suppress MAP kinase activity by inducing certain phosphatases ( 34,35 ). Our results in macrophages suggest that JNK signaling might at least partly explain the increased vascular disease in both ATM-and p53-defi cient settings.
Showing that chloroquine was insulin sensitizing in the absence of p53 was somewhat unexpected. p53 activates key mediators of glucose metabolism and insulin signaling, such as AMP kinase and the Akt-mTOR pathways ( 36 ), as well as several proteins, such as 14-3-3, known to interact with the phosphatidylinositol 3-kinase/Akt metabolic arms of insulin signaling ( 37 ). p53 is also important for orchestrating complex effects on metabolism that include the maintenance of mitochondrial respiration, mediated in part by induction of synthesis of cytochrome C oxidase 2 (SCO2) ( 38 ), and the limitation of glycolysis, mediated in part by TP53-induced glycolysis and apoptosis regulator (TIGAR) (39)(40)(41). While this article was in preparation, a study was published addressing the potential role of adipose tissue p53 in metabolism ( 42 ). Using adipose tissue-specifi c knockout and transgenic mice, the authors concluded that p53 promotes adipose tissue senescence, leading to an insulin resistance phenotype ( 42 ). Despite these reports linking p53 and glucose metabolism/insulin sclerosis present in the aortic root of the same cohorts is also shown ( Fig. 5B , top panels), along with representative aortic root sections ( Fig. 5B , bottom panels). Chloroquine signifi cantly decreased total aortic and aortic root atherosclerosis in wild-type mice but not in p53-null mice.

DISCUSSION
By virtue of its rapid activation after genomic insults and its ability to regulate numerous downstream targets, ATM plays a central role in mediating the genotoxic stress response ( 24 ). In addition to suppressing cancer susceptibility, ATM and the genotoxic stress response affect signaling networks relevant to metabolism and vascular disease ( 9,10,25 ). Chloroquine, an antimalarial compound with poorly understood anti-infl ammatory effects, activates ATM signaling ( 26 ) and has been shown to have both insulin-sensitizing and antiatherogenic effects in vivo ( 9 ). The current work confi rms that low-dose chloroquine treatment decreases atherosclerosis and improves glucose tolerance in high-fat-fed mice. It extends these observations by demonstrating that benefi cial effects of chloroquine on atherosclerosis require the presence of p53, but the effects on glucose metabolism do not. These data, representing an initial dissection of ATM and its effectors in the context of insulin resistance and atherosclerosis, indicate that the ATM/p53 axis is important for suppressing atherosclerotic plaque formation, but that other components of the genotoxic stress response downstream of ATM must be critical for salutary effects on glucose metabolism ( Fig. 6 ).
Demonstrating the importance of ATM-p53 signaling induced by chloroquine in the suppression of atherosclerosis is novel but consistent with previous work with mouse models ( 9,10,22,23 ). Functional p53 signaling in macrophages, as shown in bone marrow transplantation experiments, seems to be important for atheroma suppression ( 27,28 ). p53 has amazingly diverse effects that can be categorized as stimulating either of two complex responses depending on the biological context: (1) apoptosis and cellular senescence (generally following high-level stress or damage), or (2) cell cycle arrest, metabolic regulation, and DNA repair (generally after low-level stress) ( 29 ). Both types of responses could be involved in atheroprotection ( 22,23,27,28 ). The precise downstream mediators of the p53 atheroprotective effect are unknown. One obvious putative mediator is p21, which prevents phosphorylation of the retinoblastoma protein (phosphorylated Rb is inactive), an effect implicated in decreasing vascular disease ( 30 ). However, p21 defi ciency appears to be atheroprotective in mice, perhaps due to effects on macrophage development ( 31 ). Further, while the current data suggest that p53 activation is benefi cial in atherosclerosis, similar ben-  signaling, our results show that p53 is not required for chloroquine-induced insulin sensitization in apoE-null mice fed a high-fat diet. Chloroquine stimulates ATM signaling, which is complex. Nearly 700 proteins, many of which are not limited to the DNA damage response, have been shown to be directly phosphorylated by ATM ( 43 ). These include several proteins involved in cellular metabolism and insulin signaling, such as members of the insulin receptor/phosphatidylinositol 3-kinase/Akt cascade ( 43 ). Since ATM has also been shown to be required for full insulin-mediated activation of Akt ( 43 ), it is likely that a critical ATM target, independent of p53, regulates insulin sensitivity in vivo .
The notion of modulating the genotoxic stress response through the use of chloroquine to activate ATM ( 26 ) has been proposed as a therapeutic strategy in some forms of cancer ( 44,45 ). Such an approach is logical because cancer is a disease characterized by abnormal DNA. Curiously, DNA strand breaks are also common in atherosclerotic lesions, and one of the pleiotropic effects of statin drugs limiting clinical events may be the induction of ATM in the vasculature ( 25 ). Ongoing clinical trials will determine whether induction of ATM by low-dose chloroquine will have utility in humans (ClinicalTrials.gov identifi ers NCT00455403 and NCT00455325). Regardless of the results of these trials, insulin resistance and atherosclerosis are likely to dominate clinical practice for decades, and further dissection of ATM-dependent pathways might lead to new treatments for diabetes and vascular disease.
The authors thank Trey Coleman for technical support.