PPARGC1A Variation Associated With DNA Damage, Diabetes, and Cardiovascular Diseases
The Boston Puerto Rican Health Study
OBJECTIVE—Individuals with type 2 diabetes exhibit higher DNA damage and increased risk of cardiovascular disease (CVD). However, mechanisms underlying the association between DNA damage and development of type 2 diabetes and CVD are not understood. We sought to link peroxisome proliferator–activated receptor-γ coactivator-1 α (PPARGC1A), a master transcriptional regulator of mitochondrial oxidative phosphorylation and cellular energy metabolism, with DNA damage, type 2 diabetes, and CVD.
RESEARCH DESIGN AND METHODS—We measured DNA damage as urinary 8-hydroxydeoxyguanosine (8-OHdG) concentration and examined the relationship between nine PPARGC1A genetic variants, DNA damage, type 2 diabetes, and self-reported CVD in 959 participants of the Boston Puerto Rican Health Study.
RESULTS—With respect to urinary 8-OHdG, PPARGC1A variants showed significant association, and PPARGC1A haplotypes exhibited significant association after correction for multiple testing. Two independent PPARGC1A variants associated significantly with type 2 diabetes (odds ratios [ORs] 1.35 and 2.46; P = 0.045 and <0.001). Carriers of minor alleles of two other PPARGC1A variants, both in strong linkage disequilibrium and associated with lower DNA damage, showed lower prevalence of CVD (ORs 0.53 and 0.65; P = 0.030 and 0.175). Moreover, we found that physical activity correlated negatively with DNA damage.
CONCLUSIONS—It is plausible that low physical activity combined with risk haplotyes contribute to the high prevalence of type 2 diabetes in this population. We propose that PPARGC1A influences development of type 2 diabetes and CVD via DNA damage. Increasing physical activity, which induces PPARGC1A expression, is a potential strategy to slow DNA damage, thereby decreasing the risk of CVD for individuals with type 2 diabetes.
- 8-OHdG, 8-hydroxydeoxyguanosine
- 8-OHG, 8-hydroxyguanosine
- AGE, advanced glycation end product
- CRP, C-reactive protein
- CVD, cardiovascular disease
- GPX1, glutathione peroxidase 1
- HNF4A, hepatocyte nuclear factor 4α
- mtDNA, mitochondrial DNA
- NIH, National Institutes of Health
- PPARGC1A, peroxisome proliferator-activated receptor-γ coactivator-1 α
- ROS, reactive oxygen species
- SNP, single nucleotide polymorphism
- SOD2, superoxide dismutase 2
Type 2 diabetes is a major risk factor for cardiovascular disease (CVD) and other age-related ailments and affects >200 million people worldwide (1). The prevalence of type 2 diabetes differs across regions and ethnicities, being higher in African-American, Asian, Native-American, and Hispanic populations. In addition to the classical disease biomarkers, type 2 diabetes patients exhibit significantly elevated oxidative DNA damage, as measured by concentrations of 8-hydroxydeoxyguanosine (8-OHdG) or 8-hydroxyguanosine (8-OHG) in leukocytes (2) or urine (3)—such that their use as biomarkers in the diagnosis of the disease has been considered (3). Mitochondria control both energy metabolism and reactive oxygen species (ROS) production (4–6). Thus, mitochondrial dysfunction may contribute to the development of type 2 diabetes (4). Furthermore, diabetic hamsters treated with inhibitors of advanced glycation end products (AGEs) showed reduced oxidative stress and restored pancreatic β-cell function (7). However, the mechanism underlying the development of type 2 diabetes, how that mechanism relates to DNA damage, and how type 2 diabetes increases the risk of CVD are not well understood.
Peroxisome proliferator–activated receptor-γ coactivator-1 α (PPARGC1A) has been identified as a master transcriptional regulator of a series of nuclear receptors, including hepatocyte nuclear factor 4α (HNF4A), PPARA, and PPARG, through which PPARGC1A controls mitochondrial function, oxidative phosphorylation, and cellular energy metabolism (8–10). PPARGC1A is expressed in a broad range of tissues and, together with PPARGC1B, exhibits enhanced expression in tissues with high oxidative capacity, such as heart, skeletal muscle, brown adipocyte (8–10), brain, and kidney. PPARGC1A expression can be induced by cold exposure in brown fat tissue (11,12), by exercise in skeletal muscle (13–15), and by fasting in heart and liver (8,10). PPARGC1A and PPARGC1B expression is upregulated in response to oxidative stress (13). PPARGC1A is required for the regulation of the ROS defense system, including ROS-detoxifying enzymes glutathione peroxidase 1 (GPX1) and superoxide dismutase 2 (SOD2) (16). Upregulation of PPARGC1A expression can suppress the production of ROS. Individuals with type 2 diabetes exhibit decreased expression of PPARGC1A in skeletal muscle (17,18). Furthermore, decreased PPARGC1A expression has been linked to reduced oxidative gene expression in this disease (18). Elderly adults show significantly lower PPARGC1A expression in skeletal muscle when compared with young adults (19). Several SNPs in PPARGC1A have been investigated for their relationship with diabetes (20–27). SNP G482S appears to be functional with 482S carriers having ∼60% less PPARGC1A expression in skeletal muscle than noncarriers (19). Furthermore, this variant has been associated with type 2 diabetes (20,24,27).
Adult Puerto Ricans who live in the U.S. mainland have been identified as a vulnerable group at increased risk for age-related chronic diseases. Health disparities affecting a high percentage of this population include diabetes, depression, hypertension, and physical impairment (28,29). Thus, this population provides the opportunity to explore the correlation between PPARGC1A genetic variation and risk of age-related diseases. Therefore, we examined the relationship between PPARGC1A variants, DNA damage, type 2 diabetes, and self-reported CVD in adults, aged 45–75 years, participating in the Boston Puerto Rican Health Study.
RESEARCH DESIGN AND METHODS
The study population comprises 273 men and 686 women who were self-identified Puerto Ricans living in the greater Boston metropolitan area and have full data records for demographic and biochemical characteristics, genotypes, and DNA damage. These subjects, aged 45–75 years, were recruited by investigators from the Boston Puerto Rican Center for Population Health and Health Disparities to participate in the Boston Puerto Rican Health Study, a longitudinal cohort study on stress, nutrition, health, and aging (online appendix [available at http://dx.doi.org/10.2337/db07-1238]) (28; http://hnrcwww.hnrc.tufts.edu/departments/labs/prchd/).
Data collection and variable definition.
Information on sociodemographics, health status, history, and behavior was collected by home interview administered by bilingual interviewers. CVD was defined as a positive response to the question “Have you ever been told by a physician that you had heart disease” or to similar questions on heart attack or stroke. Anthropometric and blood pressure measurements were collected using standard methods. Smoking and alcohol use were determined by questionnaire and defined for this analysis as current, past, or never smokers or drinkers.
Participants were instructed to provide a 12-h urine sample, which was retrieved at the home the following morning. Fasting blood samples were also drawn during this visit by a certified phlebotomist. Aliquots were saved and stored at −80°C until processed. Fasting plasma glucose was analyzed using standard procedures. C-reactive protein (CRP) was analyzed in serum, using an immunoturbidimetric reaction in a Cobas Fara II Centrifugal Analyzer with DiaSorin CRP SPQ test system antibody reagent set II (AM-0039; Atlantic Antibodies, Stillwater, MN). Using American Diabetes Association criteria, subjects were classified as having type 2 diabetes if fasting plasma glucose concentration was ≥126 mg/dl or use of insulin or diabetes medication was reported (30). Physical activity was estimated as physical activity score based on the Paffenbarger questionnaire of the Harvard Alumni Activity Survey (31). The physical activity score is constructed by weighting time spent in various activities by factors that parallel increasing oxygen consumption rates associated with physical activity intensity and is categorized as follows: 0–29, sedentary; 30–39, light activity; 40–49, moderate activity; and >50, heavy activity.
Urinary 8-OHdG DNA damage.
Oxidative DNA damage and the whole-body repair of DNA were estimated (4) by measuring 8-OHdG in urine samples with a monoclonal antibody ELISA kit from Assays Designs (Ann Arbor, MI). Briefly, ∼10 μl urine, which was collected from each subject after a 12-h overnight period, was thawed after storage at −80°C and diluted 20-fold before analysis. Diluted urine samples were measured in duplicate with a standard provided by the vendor in a 96-well microplate format. Concentrations of urinary 8-OHdG were calculated by multiplying the measured concentration by the total volume of 12-h urine and then normalized against the total amount of creatinine in the urine.
PPARGC1A and SNP selection.
The PPARGC1A gene maps to 4p15.1 on chromosome 4 and encompasses ∼110 Kb. The overall rationale for SNP selection was to balance coverage and cost with putative functional consequence of one allele against another. The nine SNPs selected for genotyping (Table 1) fall within different haplotype blocks using HapMap (http://www.hapmap.org) genotype data and TAGGER (32) with a pairwise linkage disequilibrium correlation coefficient of r2 = 0.80 and the minor allele frequency ≥0.05 in the CEU Utah residents with ancestry from northern and western Europe population. Additional information factored into selection of certain SNPs. SNP m9722 is a tag SNP for one of the largest haplotype blocks of the upstream region. The minor allele of SNP m1668 is predicted by PATCH (33) to abolish a PXR/RXR (NR1I2/RXRA) binding site. SNP i5378 is just 65 bp downstream of S74L, and both likely reside in the same haplotype block. SNPs G482S and T612M alter protein sequence and may alter structure/function of PPARGC1A protein. 3′-untranslated region SNP 3U4898 has the potential to alter mRNA structure and stability.
DNA isolation and genotyping.
Genomic DNA was isolated from buffy coats of the peripheral blood using QIAamp DNA Blood mini kit (Qiagen, Hilden, Germany) according to the vender’s recommended protocol. SNPs were genotyped with Applied Biosystems TaqMan SNP genotyping system (34). For all genotyping, blinded no-template controls and replicates of DNA samples were incorporated in each of the DNA sample plates, which were routinely checked by laboratory personnel. Based on our internal quality control and that estimated independently by external laboratories, the genotyping error rate was <1%.
Statistical analyses were performed using SAS 9.1. (Cary, NC). Continuous dependent variables, such as 8-OHdG and plasma glucose concentrations, that were not normally distributed were Box-Cox transformed (35) to achieve normality before fitting statistical models. We assessed the relationship between PPARGC1A variants, urinary 8-OHdG, and plasma glucose concentrations by covariance analysis. For type 2 diabetes and CVD, we used logistic regression. With a rare minor allele, homozygotes and heterozygotes were combined to increase statistical power. In these analyses, the dependent variables were DNA damage, plasma glucose concentration, type 2 diabetes, and CVD status. Independent variables were genotypes of the individual PPARGC1A SNPs. Analyses were adjusted for potential confounders (age, sex, BMI, smoking, alcohol intake, physical activity, and medication use) using a linear or logistic regression model. All analyses were further adjusted for population admixture estimated using the program STRUCTURE 2.2 (see below). Men and women were analyzed together, as well as separately, to examine sex-specific effects. P values ≤0.05 were considered statistically significant.
Linkage disequilibrium and haplotype analysis.
Pairwise linkage disequilibria among all nine SNPs were estimated as correlation coefficients using the HelixTree program (GOLDEN Helix, Bozeman, MN). For haplotype analysis, we estimated haplotype frequencies using the expectation-maximization algorithm (36) for a subset of SNPs selected on the basis of individual association with a given trait. To determine the association between haplotypes and phenotypes, we used haplotype trend regression analysis with the option of composite haplotype estimation implemented in Helix Tree (37). Analyses were adjusted for potential confounders and population admixture (see below). P values were further adjusted for multiple tests by a permutation test. A stepwise regression analysis was conducted to further identify the most prominent haplotype associated with phenotypes.
Population admixture was calculated using STRUCTURE 2.2, an extension of an earlier method, based on Bayesian clustering, using a Markov chain Monte Carlo algorithm (38). This new method enables estimation of population admixture using linked markers. We estimated population admixture based on 96 SNPs, with minor allele frequencies >0.05 or <0.95, representing 35 genes located on 15 different chromosomes (for details, see the online appendix).
Characteristics of participants.
The percentage of individuals who reported smoking or drinking alcohol was significantly higher in men than women (P < 0.001 for both; Table 2). In contrast, BMI (or the percentage of participants who were obese [BMI ≥30]) and CRP levels were significantly higher in women than men (P < 0.001 for both). Other demographic and biochemical characteristics did not differ significantly by sex.
Minor allele frequencies and pairwise linkage disequilibrium in correlation coefficients (r2) of all nine SNPs are presented in Table 1 and Fig. 1. The frequencies of minor alleles of the selected nine SNPs ranged from 0.04 to 0.43. No significant difference in genotype frequency was observed between men and women. All SNPs at PPARGC1A were in Hardy-Weinberg equilibrium. SNPs T612M and 3U4898 were in strong linkage disequilibrium (r2 = 0.82), whereas i55301 and G482S (r2 = 0.49), i5378 and i15867 (r = 0.35), i55301 and T612M (r2 = 0.33), m9722 and m1668 (r2 = 0.28), i5378 and i27289 (r2 = 0.27), and m1668 and i15867 (r2 = 0.25) exhibited intermediate linkage disequilibrium. Other SNPs showed weak or no linkage disequilibrium (r2 < 0.1). Because all pairwise linkage disequilibria were <0.85, all nine SNPs were retained for further analysis.
Association between DNA damage and nongenetic factors and disease states.
Women had significantly higher levels of DNA damage than men (156 vs. 138 ng/μg, P = 0.009). Current smokers tended to have higher DNA damage than past smokers or nonsmokers (156 vs. 139 ng/μg; P = 0.004; Fig. 2A). In addition, higher physical activity (score >30) was associated with less DNA damage (142 vs. 154 ng/μg; P = 0.019; Fig. 2B) after adjustment for diabetes, hypertension, BMI, CVD, and depression medication. This was further supported by a negative correlation between physical activity and DNA damage (r = −0.12, P < 0.001). However, age and alcohol use were not correlated with 8-OHdG concentrations (data not shown).
We next examined whether subjects of different disease status showed altered concentrations of 8-OHdG. Subjects who reported CVD, including stroke, heart attack, and other heart disease, had significantly higher concentrations of DNA damage (156 vs. 144 ng/μg; P = 0.035; data not shown). Individuals with type 2 diabetes tended to have more DNA damage than those without diabetes (152 vs. 143 ng/μg), although this was not statistically significant (P = 0.100). Likewise, DNA damage correlated significantly with plasma glucose concentration (r = 0.09, P = 0.004, n = 959), especially in subjects with diabetes but not on diabetes medication (r = 0.50, P = 0.003, n = 31). In addition, taking depression medication is significantly associated with increased DNA damage (data not shown).
Association between PPARGC1A variation and DNA damage.
SNP m1668, located in the promoter region, showed significant association with 8-OHdG concentration (P = 0.010; Table 3). Carriers of the m1668A allele had significantly more DNA damage than noncarriers (172 vs. 159 ng/μg). In addition, homozygous subjects (AA) of the i55301A allele also had significantly higher urinary 8-OHdG concentration (P = 0.012) than G carriers (AG and GG). Furthermore, SNP 3U4898 showed significant association with DNA damage: carriers (AA+AG) of the minor allele 3U4898A had low 8-OHdG concentration (P = 0.030). In strong linkage disequilibrium with SNP 3U4898 and 19.6 kb upstream, SNP T612M exhibited a similar association with 8-OHdG concentration; i.e., carriers (AA+AG) of the minor T612M A allele tended to have lower 8-OHdG concentrations (156 vs. 168 ng/μg) than homozygotes (GG), although this was not statistically significant (P = 0.114). However, there were no significant interactions between genotypes and type 2 diabetes or CVD on DNA damage.
Association between PPARGC1A variation and plasma glucose and glycosylated hemoglobin concentrations.
Plasma glucose was highly correlated with DNA damage, especially among subjects with untreated type 2 diabetes. After adjustment for smoking, age, BMI, sex, physical activity, depression medications, and population admixture, only SNP i5378 showed a significant association with plasma glucose concentration (P = 0.023). However, the fasting plasma glucose of subjects with type 2 diabetes and using diabetes medications may not reflect biological concentrations. Thus, we analyzed the data by dividing subjects into two groups: diabetes medication users and nonusers. Among nonmedication users, we found that four SNPs (m9722, m1668, i15867, and i55301) showed significant association with glucose concentration (Supplemental Table 1). For glycosylated hemoglobin (A1C; Supplemental Table 2), SNPs m9722 and i15867 showed significant association (P = 0.002 and 0.021, respectively) with A1C among subjects not using diabetes medication.
Correlation between PPARGC1A variation and type 2 diabetes and CVD.
Homozygotes of the minor allele i5378G had a ∼2.5-fold increased likelihood of type 2 diabetes (OR 2.46; P < 0.001) than noncarriers (TT), whereas heterozygotes had a 69% greater likelihood relative to noncarriers (1.69). In addition, homozygous subjects (AA) of the major allele i27289A also had significantly greater likelihood of having type 2 diabetes (1.35; P = 0.045) than those with genotype AG or GG. Other SNPs showed no significant association with type 2 diabetes (Supplemental Table 3).
Because CVD was associated with DNA damage, we examined whether subjects with different PPARGC1A genotypes exhibited differential CVD prevalence. Carriers of the minor alleles for either T612M or 3U4898 (both in strong linkage disequilibrium) showed 50% lower likelihood of self-reported CVD (OR 0.53 and 0.65, P = 0.030 and 0.175, respectively), whereas no significant association was observed for other SNPs (Supplemental Table 4).
To understand the combined effects of multiple variants at PPARGC1A, we conducted haplotype analysis using a subset of PPARGC1A SNPs. Although the best method among proposed strategies to select SNPs for haplotype analysis is debatable (39), the major goal was to explore the interaction among variants and to increase the power to detect associations between genotypes and phenotypes. In this regard, we selected SNPs according to their association with the phenotypes as individual variants. Based on the above analyses and simulation of haplotype association, we only included SNPs for haplotype analysis that showed significant or marginally significant (P < 0.2) association with phenotypes to ensure reasonable statistical power.
For DNA damage, six of the nine SNPs (P < 0.2) met this criterion: m1668, i15867, i55301, G482S, T612M, and 3U4898 (Table 3). There are 12 haplotypes with frequencies ranging from 1 to 21%, accounting for 99% of all haplotypes in this population (Table 4). After adjustment for covariates, haplotype trend regression analysis showed that PPARGC1A haplotypes were significantly associated with DNA damage (global significance, P = 0.007). Based on a permutation test, the probability for observing such an association is 0.009 under a null hypothesis. For individual haplotypes, carriers of the haplotype G-T-A-C-G-G showed significantly less DNA damage (β = −2.54, P = 0.027), whereas carriers of A-C-A-C-G-G, A-C-G-T-G-G, and G-T-A-T-G-G exhibited significantly more DNA damage (β = 1.91, 5.17, and 2.36 and P = 0.032, 0.042, 0.020, respectively; Table 4). We further conducted stepwise regression analysis to identify the most prominent haplotypes associated with DNA damage. Haplotype G-T-A-C-G-G exhibited the lowest level of DNA damage (β = −2.66, P = 0.005).
Two SNPs, i5378 and i27289, with respective P values of <0.001 and 0.045 for individual association with type 2 diabetes, were selected for haplotype analysis. The four haplotypes had frequencies ranging from 16 to 41% (Table 5). These haplotypes showed significant association with type 2 diabetes after permutation correction (global significance, P = 0.002). Carriers of the haplotype G-A showed the greatest association with type 2 diabetes, with a ∼2.5-fold greater likelihood relative to noncarriers (P = 0.001).
We selected four SNPs—m9722, m1688, T612M, and 3U4898—for haplotype analysis in relation to CVD. There are three major haplotypes (C-G-G-G, C-A-G-G, and A-G-G-G; data not shown) with frequencies of 0.52, 0.24, and 0.18, respectively, accounting for 94% of all haplotypes in this population. Three other haplotypes (A-A-G-G, C-G-A-A, and C-G-A-G; data not shown), each with a frequency of 0.02, represent 6% of all haplotypes. All six haplotypes showed a globally significant association with CVD after permutation correction (P = 0.038; data not shown). The haplotype A-G-G-G, representing 18% of the population, correlated significantly with increased risk of CVD (OR 8.30; P = 0.006). This haplotype represents the most prominent haplotype associated with increased risk for CVD (2.16; P = 0.014) based on stepwise regression analysis.
Adult Puerto Ricans who live in the Boston metropolitan area disproportionately bear a number of health problems, including diabetes, obesity, hypertension, stress, and decline in cognitive function (28,29). Such disparities can be attributed to lifestyle, dietary intake, and genetic variation. Here, we measured urinary 8-OHdG concentration as a biomarker of DNA damage. Results showed that DNA damage was substantial in this population and was influenced by smoking, physical activity, sex, and disease status. Current smokers exhibited greater DNA damage than nonsmokers or past smokers. Physical activity was negatively correlated with DNA damage after adjustment for disease status. Average physical activity (31) was low in this population (mean physical score = 31.6 ± 5). Low physical activity could be an important environmental factor contributing to health disparities in this population.
An examination of the relationships between genetic variants at the PPARGC1A locus and DNA damage, type 2 diabetes, and self-reported CVD was undertaken to ascertain biological plausibility. We identified three SNPs at the PPARGC1A locus showing significant association with DNA damage, although none individually passed the Bonferroni correction test (P = 0.006). However, haplotypes consisting of six SNPs at this locus did exhibit significant association with DNA damage after permutation test correction for multiple testing. Considering the limited population size (n = 959) and the overconservative nature of the Bonferroni correction, we believe correction for multiple testing in the haplotype analysis by permutation is appropriate. In addition, PPARGC1A SNPs and haplotypes were significantly associated with type 2 diabetes and CVD. It is well established that a hyperglycemic state can lead to an increase in superoxide production in mitochondria due to attenuation of electron transfer within complex III as the voltage gradient reaches its threshold (40,41). Overproduction of superoxide results in mitochondrial DNA (mtDNA) damage. Thus, DNA damage may be the by-product of type 2 diabetes. Conversely, we speculate that DNA damage could stimulate the development of type 2 diabetes (Fig. 3).
PPARGC1A has a dual role in regulating mitochondrial oxidative phosphorylation (16). First, PPARGC1A regulates glucose metabolism and mitochondrial electron transport, providing energy to the cell while generating ROS. Second, PPARGC1A is required to activate ROS-defending enzymes, such as SOD1, GPX1, uncoupling protein 2, and catalase, for the clearance of ROS (16). Balance between these two roles of PPARGC1A ensures energy homeostasis without oxidative damage to the mitochondria. An imbalance, however, which could be triggered by environmental factors, including dietary intake (i.e., hyperglycemia) and lack of physical activity (i.e., reduced PPARGC1A expression), can damage mtDNA, leading to mitochondrial malfunction followed by cell apoptosis (4,8,41). This process can occur both in skeletal muscle and in pancreatic β-cells. As individuals age, those who carry the dysfunctional variant of PPARGC1A would not maintain metabolic homeostasis of glucose. This hypothesis is supported by the observation that mitochondrial malfunction or damage leads to apoptosis of pancreatic β-cells (4,41), which, in turn, affects insulin regulation and homeostasis. Individuals then enter a cycle of increasing damage to mtDNA in skeletal muscle cells and pancreatic β-cells, which is augmented by low-performing variants of PPARGC1A. mtDNA damage and dysfunction in muscle and pancreatic β-cells ultimately leads to insulin resistance and eventual development of type 2 diabetes (4,41). Thus, in this population, low physical activity combined with genetic risk factors present one possible explanation for the high prevalence of type 2 diabetes. The cycle by which low activity of the PPARGC1A gene or protein promotes DNA damage and ultimately type 2 diabetes is modeled in Fig. 3.
A remaining question is, why are SNPs i5378 and i27289 strongly associated with type 2 diabetes but not with DNA damage (Table 3)? Type 2 diabetes is prevalent in this population (42%), and most individuals with diabetes were under treatment to control plasma glucose. Therapy may disrupt the correlation between DNA damage, measured by 8-OHdG, and type 2 diabetes. This is supported by evidence that insulin treatment normalized plasma and tissue 8-OHdG concentrations in streptozotocin-induced diabetic APA rats (42). Here, we report a moderate correlation (r = 0.09, P = 0.004) between glucose and DNA damage in the whole population. However, among untreated individuals with type 2 diabetes, the correlation was strong (r = 0.50, P = 0.003) relative to those undergoing treatment (r = 0.08, P = 0.112). Furthermore, no significant difference in plasma glucose concentration was observed between genotypes of SNPs i5378 and i27289 after adjustment for diabetes medication, suggesting that drug treatment weakens the correlation between DNA damage and i5378 and i27289 genotypes in this population as the associated phenotypes were targeted for treatment. Alternatively, urinary measures of 8-OHdG reflect the repair of ROS-induced DNA damage, and the DNA damage associated with SNPs i5378 and i27289 may not be detected in urine but could be detected in tissues (e.g., pancreatic β-cells).
An alternate hypothesis is that each SNP alters PPARGC1A gene or protein activity in a manner that affects only a subset of the multiple processes this protein regulates, namely, oxidative phosphorylation, cellular energy metabolism, and the ROS defense system. Given this notion, the consequences of a SNP are observed only for certain functions of PPARGC1A and not for others. This assumes that there are function-specific domains to the PPARGC1A gene and that certain SNPs that alter transcription factor binding sites (affecting gene expression), exon-intron splicing (alternative mRNA splicing), or protein function reside in such domains. Thus, variants i5378 and i27289, which are associated with glucose homeostasis and increased likelihood of type 2 diabetes, would do so through different processes that are independent of DNA damage.
Subjects with type 2 diabetes have increased CVD risk (43,44), but our understanding of the underlying mechanism is incomplete. In this report, we observed that carriers of the minor alleles of T612M and 3U4898, which are in strong linkage disequilibrium, tend to have lower levels of both DNA damage and CRP (P = 0.012 and 0.175, respectively; data not shown). These carriers also had lower prevalence of CVD. Conversely, subjects carrying the haplotype (A-G-G-G) representing the major alleles of T612M and 3U4898 showed significantly higher prevalence of CVD. This observation suggests that type 2 diabetes may contribute to CVD through the process of DNA damage (see Fig. 3). Hyperglycemia can cause DNA damage because of overproduction of superoxide by the mitochondrial electron transport chain (40,41). The imbalanced superoxide can also activate intracellular production of AGE precursors (41), leading to inflammation and increased plasma CRP. In addition, DNA damage may lead to injury in the endothelium and induce inflammation and, subsequently, the development of atherosclerosis (45). Monocytes from subjects with diabetes have elevated oxidative stress, inflammation response (46), and binding to endothelium compared with cells from individuals without diabetes (47,48). Furthermore, human atherosclerotic plaques showed high levels of DNA damage (49). In this study, among the subjects not using diabetes medication, both carriers and noncarriers of the minor alleles of T612M and 3U4898 have a normal range of glucose and A1C levels. However, among type 2 diabetic patients using diabetes medication, the carriers have significantly lower glucose and A1C levels than the noncarriers (Supplemental Tables 1 and 2). This observation suggests that the carriers of the minor alleles of T612M and 3U4898 have better control of glucose concentrations when in a hyperglycemic state, with concomitant reduction in DNA damage, leading to less inflammation (lower CRP) and thus lower prevalence of CVD. Exercise can increase expression of PPARGC1A in skeletal muscle of subjects with type 2 diabetes (14,50). We found that physical activity correlated with lower DNA damage. Although the current study is limited in size and has only self-reported CVD, the pattern of results suggests that increasing expression of PPARGC1A by exercise may be a strategy to reduce the risk of CVD in individuals with type 2 diabetes.
This study has received the National Institutes of Health (NIH), National Institute on Aging Grant 5P01-AG-023394-02, NIH/National Heart, Lung, and Blood Institute Grant HL54776, and U.S. Department of Agriculture Research Service Contracts 53-K06-5-10 and 58-1950-9-001.
We thank three anonymous reviewers for corrections and suggestions to improve the manuscript.
Published ahead of print at http://diabetes.diabetesjournals.org on 30 January 2008. DOI: 10.2337/db07-1238.
Additional information for this article can be found in an online appendix at http://dx.doi.org/10.2337/db07-1238.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Received August 31, 2007.
- Accepted December 19, 2007.