Functional Variants in the Glutathione Peroxidase-1 (GPx-1) Gene Are Associated With Increased Intima-Media Thickness of Carotid Arteries and Risk of Macrovascular Diseases in Japanese Type 2 Diabetic Patients

A total of 184 unrelated Japanese type 2 diabetic patients were recruited from patients attending the outpatient clinic of the Wakayama University of Medical Science Hospital. In all subjects, the IMT of common carotid arteries was measured with high-resolution B-mode ultrasonography. This group consisted of 102 men and 82 women. The average age was 63.7 ± 9.1 years (range 29–84), BMI was 23.7 ± 3.0 kg/m² (range 15.2–29.7), and age at diagnosis was 44.8 ± 8.5 years (range 25–60). Of the subjects, 24% were treated with diet, 41% with oral agents, and 35% with insulin. Subjects who were GAD antibody positive and/or had started insulin therapy within 3 years of the diagnosis of diabetes were excluded from this study. Diabetes was diagnosed according to the criteria of the World Health Organization. All the participants gave written informed consent before participating in the study. This study was approved by the ethics committee of the Wakayama University of Medical Science.

All exons including 5′- and 3′-untranslated regions, intron, and ∼1.8 kb of the promoter region were amplified with a PCR and sequenced in DNA samples from 20 Japanese type 2 diabetic patients. PCR was carried out using AmpliTaq Gold polymerase (Applied Biosystems, Foster City, CA) or LATaq polymerase (TaKaRa Biotechnology, Kusatu, Japan). Sequencing was carried out using a Big Dye Terminator Cycle Sequencing FS Ready Reaction Kit (Applied Biosystems) on an automated DNA capillary sequencer (model 310; Applied Biosystems). Polymorphisms identified in this screening were designated according to their location from the transcription start site described by Moscow et al. (14). Sequence information of the primers and the conditions for PCR direct sequencing are shown in Table 1.

Polymorphisms identified were genotyped in 184 Japanese type 2 diabetic subjects using PCR direct sequencing. Genotyping was performed using primers (promoter-2 forward and promoter-2 reverse) for the A to G polymorphism at −602 and the C to T polymorphism at +2, primers (exon1-1 forward and exon1-1 reverse) for the Ala⁵/Ala⁶ in exon 1, and primers (exon2-2 forward and exon2-2 reverse) for the Pro198Leu missense mutation in exon 2. We estimated the haplotypes using the EH program (http://linkage.rockefeller.edu) and calculated linkage disequilibrium coefficients (D′ and Δ²) using the Graphical Overview of Linkage Disequilibrium (GOLD) program (http://www.sph.umich.edu/csg/abecasis/GOLD).

The IMT of common carotid arteries defined by Pignoli and Longo (15) was measured with high-resolution B-mode ultrasonography (Power Vision 7000; Toshiba, Tokyo, Japan) using a 7.5-MHz electrical linear transducer. Scanning of the common carotid arteries was performed bilaterally from two different longitudinal projections (i.e., anterior-oblique, lateral) as well as a transverse projection. At each longitudinal projection, the site of the greatest thickness was sought along the arterial walls nearest the skin and farthest from the skin at the common carotid arteries. Localized thickness of >2.0 mm was excluded from the measurement as plaque lesion. Three determinations of IMT were conducted at the site of the greatest thickness and at two other points: 1 cm upstream and 1 cm downstream from this site. The average of these six determinations (three on each side) was used as the representative value for each individual.

Assessment of macrovascular disease was based on each patient’s medical records. Cardiovascular disease (myocardial infarction and angina pectoris) and cerebrovascular disease (stroke) were diagnosed according to the criteria used in the Japan Diabetes Complication Study (16). Peripheral vascular disease was diagnosed if there was intermittent claudication with a low ankle-brachial index, i.e., <0.9. Subjects with a systolic blood pressure >140 mmHg and/or with a diastolic blood pressure >90 mmHg or on antihypertensive treatment were defined as hypertensive. Subjects with total cholesterol >5.69 mmol/l (220 mg/dl), HDL cholesterol <1.03 mmol/l (40 mg/dl), and/or triglycerides >1.70 mmol/l (150 mg/dl), or on antihyperlipidemic medication were defined as hyperlipidemic. The existence of nephropathy was determined if a patient had overt proteinuria (two successive urinalyses positive by reagent strip) and/or an elevated serum creatinine level (>1.2 mg/dl). Patients with simple, preproliferative, or proliferative retinopathy were assumed to have retinopathy.

Two human GPx-1 clones (UI-E-CL1-AEL-E-04-0-UI and UI-CF-EC1-ABK-A-09-0-UI) were obtained from the EST clone collection (Open Biosystems, Huntsville, AL). UI-E-CL1-AEL-E-04-0-UI has a combination of polymorphism and Ala⁵/198Pro, and UI-CF-EC1-ABK-A-09-0-UI has a combination of polymorphism and Ala⁶/198Leu. An 800-bp XhoI-XhoI fragment, which also contained a selenocysteine insertion sequence in the 3′-untranslated region (17), was subcloned into the plasmid pcDNA3.1 (Invitrogen, Carlsbad, CA), in which expression is driven by a cytomegalovirus promoter.

Bovine aortic endothelial cells (BAECs) were obtained from Dainippon Pharmaceuticals (Osaka, Japan) and maintained in Dulbecco’s modified Eagle’s medium containing 4,500 mg/l glucose, 10% FBS, and antibiotics (100 units/ml penicillin G sodium and 100 mg/ml streptomycin sulfate). The BAECs (passages 5–10) were seeded into six-well culture plates. Two micrograms each of pcDNA3.1-GPx-1 constructs (pcDNA3.1-GPx-1-Ala⁵/198Pro and pcDNA3.1-GPx-1-Ala⁶/198Leu) or pcDNA3.1 plasmid was transiently transfected to the BAECs using a FuGENE6 transfection reagent (Roche Diagnostics, Mannheim, Germany). After 24 h, the BAECs were washed three times with ice-cold PBS and gently scraped with a lysis buffer (50 mmol/l Tris/HCl: pH 7.5, 5 mmol/l EDTA and 1 mmol/l dithiothreitol). The cell suspension was homogenized with an ultrasonicator (2 × 5 s) on ice. GPx activity was then determined from the supernatant by coupling the reduction of peroxides (t-butylperoxide) and the oxidation of glutathione using NADPH as a cofactor with a commercially available kit (Cayman, Ann Arbor, MI). The specific activity (per milligram protein) was calculated after protein determination by a Bio-Rad protein assay kit (Bio-Rad Laboratories, Richmond, CA). In preliminary experiments, a plasmid coding for green fluorescence protein had been transfected under the same condition and indicated that ∼20% of cells demonstrate green fluorescence. The data presented represent the means of five independent transfection experiments per construct. The difference in the averaged activity among three groups was analyzed by one-way ANOVA, and a Fisher’s protected least-significant difference test was used for the post hoc test. The difference in the averaged increment of activity between Ala⁵/198Pro- and Ala⁶/198Leu-transfected BAECs was analyzed by an unpaired Student’s t test.

DNA fragments corresponding to the basal promoter of GPx-1 (nucleotide −819 to + 44) were amplified by PCR using genomic DNA from two patients with different single nucleotide polymorphism combinations (−602A/G and +2C/T, −602A/G and +2C/C) as a template. The PCR products were subcloned into a pGL3-basic firefly luciferase reporter vector (Promega, Madison, WI) in the 5′-3′ orientation, and the luciferase reporter plasmids representing three genotypes (−602A/+2C, −602G/+2C, and −602G/+2T) were made.

We transiently transfected 0.25 μg of one of the constructs with 0.01 μg pRL-SV40 vector (renilla luciferase under control of SV40 promoter), as an internal control for transfection efficiency, into the BAECs (passages 5–10) using a FuGENE6 transfection reagent. After 24 h, we collected the cells and measured luciferase activity using the Dual-Luciferase Reporter Assay System (Promega). The luciferase relative activity for each construct was calculated as a fold increase over the activity for the negative control empty vector (pGL3-basic). The data presented represent the means of five independent transfection experiments per construct. The differences in the averaged activity among three groups were analyzed by one-way ANOVA, and a Fisher’s protected least-significant difference test was used for the post hoc test.

Results are presented as means ± SD or n (%). Differences between the type 2 diabetic patient groups divided on the basis of the codon 198 and −602A/G polymorphisms were assessed for statistical significance using an unpaired Student’s t test, Mann-Whitney’s U test (when the data were skewed), and a χ² test. We used a multivariable regression analysis to estimate the relationship between the GPx-1 genotype (Pro/Pro = 0, Pro/Leu = 1) and the value of IMT and a multivariable logistic regression analysis for comparing the GPx-1 genotype and the prevalence of macrovascular disease with adjustment for common vascular risk factors (age, sex, HbA_1c, hyperlipidemia, hypertension, and a smoking habit). All analyses were performed with the StatView program for Windows (version 5.01; SAS Institute, Cary, NC). P < 0.05 was considered statistically significant.

All exons, intron, and ∼1.8 kb of the promoter region were screened in 20 Japanese type 2 diabetic patients, and four polymorphisms were identified (Fig. 1). These polymorphisms include a G for A substitution at −602 (−602A/G), a T for C substitution at +2 (+2C/T), a six for five polyalanine polymorphism at codon 7-11 (Ala⁵/Ala⁶), and the leucine (CTC) for proline (CCC) amino acid substitution at codon 198 (Pro198Leu) and have been previously reported in the study for patients with lung cancer (18) (−602A/G was described as −592A/G in the article). We found a mistake in the original sequence used for the numbering and changed it to the exact number in this study. The reference single nucleotide polymorphism identification (SNP ID) numbers of −602A/G, +2C/T, and Pro198Leu are rs3811699, rs1800668, and rs1050450, respectively. We further genotyped these polymorphisms in 184 type 2 diabetic patients. Thirty-three patients had −602G, +2T, Ala⁶, and 198Leu all as the heterozygous state, and four patients had only −602G as the heterozygous state. From these results, we estimated three haplotypes (Table 2). The +2T, Ala⁶, and 198Leu were in complete linkage disequilibrium with each other (D′ = 1.00, Δ² = 1.00). The −602G was also in strong linkage disequilibrium with those polymorphisms (D′ = 1.00, Δ² = 0.881). Genotype distributions did not significantly differ from the Hardy-Weinberg equilibrium expectations. Based on these results, the Pro198Leu was selected as the genetic marker for the detailed clinical analysis.

Patients were divided on the basis of the codon 198 polymorphism into two groups (Pro/Pro and Pro/Leu). Leu/Leu genotype was not identified in this study. These groups were not significantly different with respect to age, maximum BMI, HbA_1c, age at diagnosis of diabetes, duration of diabetes, family history of diabetes, mode of treatment, hypertension, hyperlipidemia, and the prevalence of microvascular complications such as nephropathy and retinopathy (Table 3). However, the mean IMT of the common carotid arteries was significantly higher in the Pro/Leu group than in the Pro/Pro group (1.04 ± 0.23 vs. 0.91 ± 0.18 mm, P = 0.0028) (Table 4). We also found that the prevalence of cardiovascular disease and peripheral vascular disease was significantly higher in the Pro/Leu group than in the Pro/Pro group (24.2 vs. 10.6%, P = 0.035, and 15.2 vs. 4.6%, P = 0.027, respectively). The prevalence of cerebrovascular disease was higher in the Pro/Leu group than in the Pro/Pro group but was not statistically significant (15.2 vs. 7.9%, P = 0.195) (Table 4). All of the results were virtually unchanged when statistical models were adjusted for common vascular risk factors (age, sex, HbA_1c, hyperlipidemia, hypertension, and smoking habit). The regression coefficient using multivariable regression analysis for the comparison between the GPx-1 genotype (Pro/Pro = 0, Pro/Leu = 1) and the value of IMT was 0.127 (P = 0.0002). The odds ratios using multivariable logistic regression analysis for the comparison between the GPx-1 genotype (Pro/Pro = 0, Pro/Leu = 1) and the risk of macrovascular complications were as follows: coronary heart disease, 4.07 (95% CI 1.47–11.26, P = 0.0068); cerebrovascular disease, 2.59 (95% CI 0.79–8.53, P = 0.1183); and peripheral vascular disease, 4.46 (95% CI 1.17–17.05, P = 0.028).

When patients were divided into two groups on the basis of the −602A/G polymorphism (A/A and A/G; G/G genotype was not identified), the mean IMT was significantly higher in the A/G group than in the A/A group (1.02 ± 0.22 vs. 0.91 ± 0.18 mm, P = 0.0052), but the differences for the prevalence of cardiovascular (21.6 vs. 10.8%, P = 0.083), cerebrovascular (13.5 vs. 8.2%, P = 0.315), and peripheral vascular (13.5 vs. 4.8%, P = 0.053) disease were not statistically significant.

To determine whether the two polymorphisms in the coding region of the GPx-1 gene (Ala⁵/Ala⁶ and Pro198Leu) would affect its enzyme activity, we constructed two expression vectors, pcDNA3.1-GPx-1 (Ala⁵/198Pro) and pcDNA3.1-GPx-1 (Ala⁶/198Leu), and transiently transfected them into the BAECs. The GPx enzymes express endogenously in the BAECs, and cellular GPx activity in sham-transfected BAECs was 99.6 ± 8.4 units/mg protein. GPx activity significantly increased in both pcDNA3.1-GPx-1 (Ala⁵/198Pro)-transfected (132.5 ± 3.3 units/mg protein, P < 0.0001) and pcDNA3.1-GPx-1 (Ala⁶/198Leu)-transfected (119.8 ± 3.7 units/mg protein, P = 0.0004) BAECs compared with sham-transfected BAECs (Fig. 2). These levels of overexpression were similar to that observed in a similar experiment described by Weiss et al. (19). When the increment of GPx activity was compared between Ala⁵/198Pro- and Ala⁶/198Leu-transfected BAECs, that of Ala⁶/198Leu was significantly lower than that of Ala⁵/198Pro (20.1 ± 10.6 vs. 32.9 ± 9.5 units/mg protein, P = 0.0230).

To investigate whether two polymorphisms, −602A/G and +2C/T, would affect the transcriptional activity of GPx-1, we constructed three kinds of plasmids with an 819-bp basal promoter fragment containing these polymorphisms upstream of a luciferase gene transcriptional unit. In transient transfection into the BAECs, the −602A/+2C construct induced luciferase expression over 1,000-fold relative to a promoter-less pGL3-basic vector (Fig. 3). The relative luciferase activity of the −602G/+2C haplotype decreased significantly compared with that of the −602A/+2C haplotype (986 ± 47 vs. 1,096 ± 76, P = 0.0180). The relative luciferase activity of the −602G/+2T haplotype decreased significantly compared with that of the −602A/+2C haplotype (822 ± 76 vs. 1,096 ± 76, P < 0.0001) and that of the −602G/+2C haplotype (822 ± 76 vs. 986 ± 47, P < 0.0001). These results indicated that the two polymorphisms, −602A/G and +2C/T, affected transcription of GPx-1.

In this study, the mean IMT of common carotid arteries was significantly higher in the Pro/Leu group than in the Pro/Pro group in type 2 diabetic patients. It has been shown that the carotid IMT is positively associated with an increased risk of macrovascular disease (20,21). The prevalences of cardiovascular and peripheral vascular disease in our samples were also significantly higher in the Pro/Leu group than in the Pro/Pro group. We identified four polymorphisms (−602A/G, +2C/T, Ala⁵/Ala⁶, and Pro198Leu) in the human GPx-1 gene. The 198Leu was in strong linkage disequilibrium with −602G, +2T, and Ala⁶. In vitro functional analyses indicated that the combination of polymorphisms (Ala⁶/198Leu) in the coding region of the GPx-1 gene had a 40% decrease in enzyme activity, and the combination of polymorphisms (−602G/+2T) had a 25% decrease in transcriptional activity. These results suggest that functional variants in the GPx-1 gene are associated with increasing risk of atherosclerosis in type 2 diabetic patients.

GPx-1 is a cytosolic isoform of GPx and is expressed in vascular endothelial cells. GPx-1 activity decreased in atherosclerotic plaques excised from patients with carotid artery disease (22). The patients with a selenium deficiency had an increased incidence of coronary artery disease and a decrease in GPx activity, because GPx is a selenocysteine-containing enzyme and a low serum selenium level induces a decrease in GPx enzyme activity (23). In addition, hyperhomocysteinemia is one of the risk factors for atherosclerotic vascular disease. It has been reported that homocysteine inhibited the expression of GPx-1 and lead to an increase in reactive oxygen species that inactivated nitric oxide and promoted endothelial dysfunction (24). The overexpression of GPx-1 rescued this homocysteine-induced endothelial dysfunction (19). Furthermore, heterozygous GPx-1–deficient mice show endothelial dysfunction, such as vasoconstriction to acetylcholine compared with vasodilation in wild-type mice, and an increase in the plasma and aortic levels of the isoprostane iPF2α-III, a marker of oxidant stress (10). A histological section from the coronary vasculature of these mice shows increased perivascular matrix deposition, an increase in the number of advential fibroblasts, and intimal thickening (11). These findings support the hypothesis that a change in the function of Gpx-1 contributes to vascular oxidant stress and the progression of atherosclerosis.

Four polymorphisms identified in this study have been previously reported in a study for patients with lung cancer (18). Linkage disequilibrium among these polymorphisms was also observed in the screening of 17 tumor specimens. The allele frequency of Ala⁶, which was 9.0% in our study, was 27.8% in 42 American cancer patients in the study. The association between the 198Leu variant and erythrocyte GPx activity was examined in 66 subjects from the Finnish/Swedish population (25). The allele frequency of 198Leu was higher (41%) than our result, and no significant association was observed. Because the −602A/G, +2C/T, and Ala⁵/Ala⁶ polymorphisms were not examined in the population, the haplotype configuration was unknown. Although the level of erythrocyte GPx activity is higher in women than in men (26) and smoking reduces the activity (27), the information for sex and smoking was not shown. These factors may affect the result, which is inconsistent with our result. Furthermore, a lack of association between the Leu variant and risk of stroke was reported in a case-controlled study on the Swedish general population from the same group (25). The prevalence of stroke in our study was higher in the Pro/Leu group than in the Pro/Pro group, but it was not statistically significant. On the other hand, the GPx-1 gene variants were associated with increased IMT. Because it is known that carotid IMT correlates with prevalent and incident stroke (20), the association between the gene variants and risk of stroke requires further clarification in a larger series of individuals. In addition, increased polyol pathway flux in a diabetic condition exacerbates an antioxidant system by GPx. In the pathway, aldose reductase catalyzes glucose to sorbitol, using NADPH as a cofactor. The increased glucose flux in a hyperglycemic environment results in decreased NADPH. Because NADPH is required for regenerating reduced glutathione, which is also a cofactor for GPx, this could enhance the genetic dysfunction of GPx-1 in diabetic patients. It will also be necessary to further evaluate the role of the genetic variants in nondiabetic subjects.

In conclusion, functional variants in the GPx-1 gene are associated with increased IMT of carotid arteries and risk of cardiovascular disease and peripheral vascular disease in diabetic patients. This information might be helpful in picking up the patients with high genetic risk for macrovascular complications and in preventing them by intensive treatment for other risk factors of atherosclerosis.

This work was supported by a grant-in-aid for scientific research on priority areas “Medical Genome Science”; a grant-in-aid for scientific research (13204074) from the Ministry of Education, Culture, Sports, Science and Technology of Japan; and a grant-in-aid for scientific work for diabetes from Wakayama University of Medical Science.

We thank Dr. Issei Yoshiuchi for his support of genetic analysis. We also thank Kanako Fujiuchi and Machiko Deguchi for technical assistance.