| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Lipids, Lipoproteins, and Cardiovascular Risk Factors |
1 Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis, MN.
2 Department of Nutrition, University of Oslo, Oslo, Norway.
3 Department of Laboratory Medicine, Medical School, University of Minnesota, Minneapolis, MN.
4 Institute of Molecular Medicine, University of Texas Health Science Center at Houston, Houston, TX.
aAddress correspondence to this author at: University of Minnesota, Division of Epidemiology, School of Public Health, 1300 South 2nd St., Suite 300, Minneapolis, MN 55454. Fax 612-624-0315; e-mail jacobs{at}epi.umn.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionParticipants and MethodsResultsDiscussionReferences |
|---|
Methods: We investigated the association of the commonly observed UCP2 Ala55Val (V) polymorphism with diabetes mellitus and impaired fasting glucose (IFG) among 3684 participants in the Coronary Artery Risk Development in Young Adults (CARDIA) study.
Results: The V frequency was 
45% in blacks and 42% in whites. Those with the Val/Val (VV) genotype had a higher incidence of diabetes than those having the Ala/Ala (AA) genotype (5.8% vs 3.3%; P = 0.02). Similarly, the incidences of diabetes in participants without abdominal obesity were 2.8% and 1.0% (P = 0.03) in the VV and AA groups, and 12.4% and 8.3% (P = 0.15) in participants with abdominal obesity. The incidence of IFG was higher in VV vs AA only in those without abdominal obesity (12.9% vs 9.2%). These trends persisted in minimally and fully adjusted models, and in strata of blacks and whites and men and women. The homeostasis model assessment for insulin resistance was highest in VV in the combined group of those with IFG or untreated diabetes, but not in those with normal fasting glucose.
Conclusion: The VV genotype of the UCP2 polymorphism was positively related to diabetes. It may involve increased insulin resistance in those with impaired glucose homeostasis.
|
Introduction |
|---|
|
TopAbstractIntroductionParticipants and MethodsResultsDiscussionReferences |
|---|
Uncoupling proteins (UCPs) 1 comprise a complex of several related proteins that may significantly regulate energy utilization. UCP, as a proton transporter, uncouples the proton-motive force (diminishing the proton gradient across the inner mitochondrial membrane), decreasing formation of both ATP and reactive oxygen species (ROS) with the release of chemical energy as heat. UCP2, one of 5 known human homologs (1)(2), is located on chromosome 11 (1), a site linked to lower resting metabolic rate (3). Work in model systems and in animals suggests that increased expression of UCP2 decreases glucose-stimulated insulin secretion and thus impairs glucose homeostasis and increases the risk of diabetes mellitus (4)(5)(6). In contrast, elimination of the UCP2 gene in diabetes-prone rodents decreases the risk of diabetes (7).
Variation in UCP activity may influence the production of ATP and ROS in humans. A common UCP2 polymorphism involves a C-to-T substitution in exon 4 at position 164 of a complementary DNA (cDNA) product, changing an alanine to valine at codon 55 (Ala55Val polymorphism) (8). Experiments have found lower 24-h energy expenditure (9) or higher exercise energy efficiency in the Val/Val (VV) genotype in healthy individuals (10). Persons with the VV genotype may therefore have a lower degree of uncoupling, more efficient energy utilization, more production of ROS, and lower fat oxidation for a given unit of work than persons with the Ala/Ala (AA) genotype (9)(11)(12). The VV genotype(13) and 4 others (including the promoter region –866 variants) (14)(15)(16)(17) of the UCP2 gene have shown a modestly increased risk of type 2 diabetes in case–control or clinical studies. Thus, these relationships require more clarity.
Given the metabolic role of UCP2 in adipose tissue, muscle, and islet cells, the known activities of the UCP2 Ala55Val polymorphism, and another study suggesting a role of the polymorphism in type 2 diabetes mellitus (13), we asked in the Coronary Artery Risk Development In Young Adults (CARDIA) population whether the V allele is associated with higher amounts of central body fat and other factors relating to the metabolic syndrome and/or to the incidence diabetes mellitus. We also determined the interactive influences of UCP2 genotype and central adiposity on the development of diabetes mellitus. Finally, we carried out an exploratory analysis to determine whether an element of insulin resistance was involved in the association of the UCP2 polymorphism and diabetes status.
|
Participants and Methods |
|---|
|
TopAbstractIntroductionParticipants and MethodsResultsDiscussionReferences |
|---|
anthropometric and other risk factor measurements
At each clinic visit the following were measured: body weight, height, body mass index (BMI) as body weight/height2 (kg/m2), waist circumference (mean of 2 measurements midway between the iliac crest and the lowest lateral portion of the rib cage), and sociodemographic risk factors of cardiovascular diseases [by use of standardized questionnaires (19)].
blood collection, processing, and analyses
Blood samples were collected after an overnight fast, and the serum and plasma were separated. At several examinations, plasma glucose was measured with the hexokinase method and serum insulin by a specific RIA (20)(21). Adiponectin [by RIA (20)] was assayed in serum at the year 15 examination.
genotyping of the ucp2 polymorphism
The UCP2 Ala55Val polymorphism was genotyped by use of the TaqMan system (Applied Biosystems). The 101-bp product was amplified in a 22-µL reaction containing 0.9 µM each of the forward (5'-TTGCAGATCCAAGGAGAAAGTCA-3') and the reverse primer (5'-CCCTCAGTACGCACCATGGT-3'), 50 ng of DNA, 5.0 mM MgCl2, and 1x TaqMan Universal PCR Master Mix containing AmpliTaq Gold DNA Polymerase. After an initial step of 2 min at 50 °C and 10 min at 95 °C to activate the AmpliTaq Gold, the products were amplified by 40 cycles of 15 s at 95 °C and 1 min at 62 °C. A total of 0.2 µM of the 6-carboxyfluorescein (FAM)/6-carboxytetramethylrhodamine (TAMRA)-labeled sequence-specific probe (5'-FAM-CGCTACAGCCAGCGCCAGTACC-TAMRA-3' and 0.1 µM of the VIC-labeled sequence-specific probe 5'- VIC-CGCTACAGTCAGCGCCAGTACCG-TAMRA-3' were used in the allele differentiation assay. Allele detection and genotype calling were performed by use of the ABI 7700 and the Sequence Detection System software (Applied Biosystems).
diabetes mellitus, impaired fasting glucose, and metabolic syndrome assessment
Diabetes was defined as fasting glucose 
1260 mg/L and/or taking diabetes medication; impaired fasting glucose (IFG) was defined as a fasting glucose of 1000–1250 mg/L. The incidences of diabetes and IFG were defined as any single occurrence in the 15 years. The homeostasis model assessment (HOMA) of insulin resistance was computed as fasting plasma glucose (mmol/L) times fasting serum insulin (mIU/L) divided by 22.5.
data analysis
Data from all 6 examinations were analyzed with SAS, Ver. 8.2. Dependent variables included the incidences of diabetes and IFG, and anthropometric and biochemical measurements were included as continuous variables predicted by UCP2 genotypes. Analyses of biochemical measurements at each examination were restricted to those with a fasting duration at least 8 h, to women who were not pregnant, and to those persons not taking medications for diabetes (n = 151 observations excluded) and/or lipid-lowering (n = 57 observations excluded) medications.
Logistic regression was used to obtain odds ratios (ORs) for the incidence comparisons between genotypes. The general linear model was used to estimate mean plasma/serum measurements at year 15. To assess the interaction between central obesity and genotype with incident diabetes mellitus, abdominal obesity was considered to be present if the National Cholesterol Education Program waist circumference criteria were met at any examination (abdominal obesity: waist circumference >102 cm for men, >88 cm for women); thus, those not abdominally obese never met this threshold through 6 examinations. The correlation coefficient between waist circumference and BMI was 0.8 (20).
Minimal adjustment subsumed race, sex, age, and center. The fully adjusted model additionally included physical activity, smoking status, education (years), total energy intake (kcal), total fat intake (g/day), and alcohol intake (mL/day). No interactions were found between genotype and these factors, and results were essentially unchanged with full compared with minimal adjustment (data not shown).
|
Results |
|---|
|
TopAbstractIntroductionParticipants and MethodsResultsDiscussionReferences |
|---|
25 years for all race and sex groups (Table 1
). Blacks had lower mean duration of education than whites. Black women had a higher BMI and waist circumference than white women. Blacks also had slightly higher fasting circulating insulin concentrations and HOMA than whites, and men had lower circulating adiponectin than women. The genotype and allele frequencies did not vary by race or sex (Table 2
). The genotype frequencies were in Hardy–Weinberg equilibrium (
2 test) for both blacks (P = 0.86) and whites (P = 0.87). There were no differences in obesity status among the genotypes: 31.4% of those with the VV genotype, 31.7% of those with the AA genotype, and 32.8% of those with the AV genotype ever had central obesity (P = 0.69 for differences among the 4 race–sex categories).
|
|
Over 15 years of observation in CARDIA, 5.8% of those with the VV genotype had diabetes mellitus compared with 3.7% for the AV genotype and 3.3% for the AA genotype (Table 3
). The minimally adjusted OR for incident diabetes was 1.75 (95% confidence interval, 1.11–2.76) for VV vs AA. When the participants were divided into those ever with abdominal obesity vs those never obese, differences in the incidence of diabetes remained statistically significant for the nonobese [for VV vs AA, OR = 2.7 (1.09–6.58)], and a similar trend was present in the centrally obese (Table 3
). Furthermore, the incidence of diabetes was significantly higher in the VV genotype than among all other participants [VV vs combined AA/AV, 5.8% vs 3.5%; OR = 1.64 (1.12–2.39)] and among those who did not have abdominal obesity [VV vs combined AA/AV, 2.8% vs 0.9%; OR = 2.83 (1.38–5.81)]. Among those with abdominal obesity, consistent but nonsignificant findings existed [VV vs combined AA/AV, 12.4% vs 8.9%; OR = 1.48 (0.88–2.23)]. Similar findings were found within race and sex strata. No evidence of a race-by-genotype interaction was observed for any of the associations with diabetes or abdominal obesity examined here (data not shown). Furthermore, adjustment for hypertension, previously reported to be associated with UCP2 (22), was not associated with UCP2 Ala55Val in these data and did not explain the associations seen in Table 3
.
|
Given the excess incident diabetes found in the VV genotype, we assessed the distribution of IFG among genotypes, omitting persons with diabetes from the denominators in Table 3
for these 2 outcomes. Among all persons, IFG was not significantly different in persons with the VV genotype vs those with the AA genotype. However, the incidence of IFG was significantly increased in VV individuals without abdominal obesity compared with a nonsignificant decrease in incidence in obese individuals with the VV genotype (P for interaction = 0.01).
In exploratory analyses (Table 4
) to explore a possible role of insulin resistance in the association of the UCP2 polymorphism with diabetes incidence, we examined insulin resistance, using HOMA among those with normal fasting glucose (NFG; n = 2563), IFG (n = 413), and untreated diabetes mellitus (n = 55). Slightly fewer participants were included in Table 4
than in Table 3
because of missing fasting insulin values. Given the small numbers of participants with untreated diabetes, their data are combined with those participants with IFG. The year 15 HOMA was significantly higher among participants with IFG or untreated diabetes than among those with NFG. Furthermore, HOMA did not differ greatly across genotypes among people with NFG, despite a significantly lower HOMA in participants with the AV genotype than those with the AA genotype. However, among people with IFG or untreated diabetes, HOMA was 0.58 higher among those with the VV genotype than in those with the AA genotype (P = 0.007) and 0.74 higher than in those with the AV genotype (P = 0.0002). The P value for interaction was 0.002 in predicting HOMA from genotype and fasting glucose status.
|
There were no differences for fasting adiponectin among UCP2 genotypes at year 15 (data not shown). In each examination, the patterns found in Tables 3
and 4
and for HOMA and adiponectin persisted after adjustment for concurrent physical activity, smoking status, alcohol intake, and mean of baseline and year 7 estimates of total energy and fat intakes (data not shown).
|
Discussion |
|---|
|
TopAbstractIntroductionParticipants and MethodsResultsDiscussionReferences |
|---|
45% of blacks and 42% of whites carried the V allele. In our study of a UCP2 polymorphism affecting the coding sequence, those with the homozygous VV genotype were more likely to be diagnosed with diabetes mellitus. The existence of diabetes in 5.8% of participants homozygous for the V allele was statistically significantly greater than the 3.3% and 3.7% in the AA and AV genotypes. These results extend the findings from studies of Ala55Val or other UCP2 polymorphisms involved in the development of type 2 diabetes (13)(14)(15)(16)(17)(18). Among the participants of our study, the relative risk of diabetes for individuals with the VV genotype compared with those with the AA or AV genotypes was similar for consistently thin persons and for those with central obesity. The findings were similar in women vs men and in African-American vs white participants. There are many candidate pathways that could mediate the association between the UCP2 polymorphism and diabetes. We examined insulin resistance as measured by HOMA as one possibility. Those with the VV genotype had higher HOMA indexes than those with the AA genotype after adjustment for abdominal obesity in persons who already had an abnormally increased fasting glucose (i.e., those with IFG or untreated diabetes mellitus). Those with the AV genotype had slightly lower HOMA indexes than those with the AA genotype, consistent with an underdominance model (i.e., the values for the heterozygote are lower than those for either homozygote). Our findings are consistent with the role of mitochondrial function in the etiology of diabetes. In normal energy metabolism within all cells, oxidative phosphorylation in mitochondria captures energy from the resulting proton gradient, with efficient formation of ATP. UCPs diminish the proton gradient, leading to less ATP formation and release of energy as heat (1)(2). Uncoupling serves an important physiologic function by dissipating excess energy in the presence of a positive energy balance. In humans, the extent of uncoupling is partially regulated by the genetic polymorphism Ala55Val, in which the VV genotype uncouples at a lower rate than the AA genotype (1).
Changes in UCP activity have been related to diabetes mellitus by mechanisms causing a decrease in insulin secretion and/or by an increase in insulin resistance (26). The change in insulin secretion may occur by at least 2 different pathophysiologic mechanisms. In one scenario, enhanced uncoupling decreases the production of ATP and/or lowers ATP/ADP ratios in islet beta cells with consequent lowering of insulin secretion (4)(5)(7)(27). For example, islets from UCP-deficient mice had higher concentrations of intracellular ATP and greater insulin secretion with glucose stimulation (7)(28). In addition, UCP RNA and protein concentrations were shown to be increased in islets from ob/ob mice, a model of type 2 diabetes mellitus. In contrast, UCP-deficient ob/ob mice remained obese but demonstrated enhanced insulin secretion and action and improved glucose tolerance (7).
In an alternative mechanism, the reduction in uncoupling not only enhances the production of ATP but also ROS. In this case, the greater amounts of ROS damage or destroy the beta cells, with a corresponding decrease in insulin secretion (6). Alternatively, tighter coupling through the release of ROS may increase oxidation of lipids, proteins, and/or DNA, which in turn may damage the metabolic machinery of the cell and decrease glucose metabolism (26). For example, the enhanced production of ROS in muscle and other cells responsible for metabolism of glucose may contribute to insulin resistance, again through damage to the cellular constituents enabling the metabolic pathways.
In our study, increased insulin resistance in persons with IFG/untreated diabetes and the VV genotype suggested that insulin resistance played a role in the relationship between UCP2 and diabetes. In addition, depending on the location of increased ROS production, increased insulin resistance and/or reduced beta-cell function may occur simultaneously. Therefore, decreased UCP2 uncoupling activity in the VV participants in CARDIA might sustain or increase intracellular ATP concentrations and increase ROS production in muscle cells and/or islet beta cells.
Although the hypotheses put forth here are consistent with observations and theories based on cell and animal models, we have completed an association study examining a single polymorphism. It is difficult in any association study to make any definite conclusion as to the underlying cause of the reported association. In addition, there is possible linkage dysequilibrium with other variants, for example, UCP3, that reside on the same chromosome as UCP2. The HapMap database (29) identifies strong linkage dysequilibrium between the UCP2 Val55 polymorphism (dbSNP id = rs660339) and polymorphisms located in the UCP3 gene (rs826079 and rs826082) in whites. No corresponding information was available for African Americans. We cannot rule out the existence of other important genes and polymorphisms in linkage dysequilibrium with the UCP2 gene. By adjustment and stratification by race, we attempted to rule out population stratification attributable to race as a source of the associations presented here; however, residual confounding is possible in all observational epidemiologic studies. In addition, we included those "taking diabetes medication" as indicating the diagnosis of diabetes because those taking diabetes medications were indeed established patients with diabetes mellitus (and verified in self-reports). Analysis using a definition of diabetes based solely on glucose concentrations showed a similar positive relationship between VV polymorphism and diabetes (data not shown). Nevertheless, we excluded treated diabetic patients from analyses of HOMA, permitting an analysis of insulin resistance with no intervening effects of medications. Although we could not definitely distinguish type 1 vs type 2 diabetes mellitus in CARDIA, very few participants were tentatively diagnosed with type 1 diabetes (diagnosis before age 30, solely treated with insulin), preventing subanalyses based on type of diabetes.
To our knowledge, this is the first prospective population study to document a positive association between the UCP2 Ala55Val polymorphism and diabetes in both thinner and centrally obese people. Although our study could not determine the mechanisms, exploratory analyses point to a possible role of insulin resistance. The strong association of the VV genotype with diabetes in centrally thin persons may at the same time indicate a relationship with declining beta-cell function. This and other speculations compel additional studies to confirm the association between UCP2 and diabetes and to examine possible mechanisms linking UCP2 and diabetes mellitus in humans.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionParticipants and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  E. Chevillotte, M. Giralt, B. Miroux, D. Ricquier, and F. Villarroya Uncoupling Protein-2 Controls Adiponectin Gene Expression in Adipose Tissue Through the Modulation of Reactive Oxygen Species Production Diabetes, April 1, 2007; 56(4): 1042 - 1050. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
