Decreased Fasting and Oral Glucose Stimulated C-peptide in Nondiabetic Subjects With Sequence Variants in the Sulfonylurea Receptor 1 Gene

Type 2 diabetes has been shown to have a strong hereditary basis (1,2); however, the major genes for the common late-onset type have not yet been clearly identified (3). Because impaired insulin secretion plays a critical role in the pathogenesis of the disease (4), genes that encode major elements involved in insulin secretion are good candidates for evaluating genetic susceptibility to type 2 diabetes (3).

A central component in glucose-induced insulin secretion is the ATP-sensitive potassium (K_ATP) channel of the β-cell. Increasing ATP generated by intracellular glucose metabolism leads to the closing of these channels, membrane depolarization, the opening of voltage-gated calcium channels with influx of calcium, and exocytosis of insulin (5). The K_ATP channel is composed of the high-affinity sulfonylurea receptor (SUR1) and Kir6.2, a member of the inwardly rectifying ion channel family (6). Pathological regulation of insulin secretion has been demonstrated in studies of persistent hyperinsulinemic hypoglycemia of infancy, a rare autosomal recessive disease in which mutations in the nucleotide-binding fold-2 region of the SUR1 gene have been found (7).

In type 2 diabetes, association studies in diabetic populations have been positive, with the IVS15-3c→t and exon 18 Thr759(ACC→ACT) variants of the SUR1 gene being found more frequently in Danish (8), U.S., U.K. (9), and French Caucasian (10) diabetic individuals, and the IVS15-3c→t variant alone found more frequently in Dutch Caucasian diabetic subjects (11). However, results in Japanese populations have been negative (12). Few studies have reported on the physiological impact of SUR1 gene sequence variants. A Danish group demonstrated decreased tolbutamide-stimulated insulin secretion in normal glucose-tolerant carriers of the above intron 15 and exon 18 variants (8). Another group recently reported an association of the IVS15-3c→t variant with lower second-phase insulin secretion in Dutch nondiabetic (ND) individuals (13).

The present study was thus undertaken to examine the role of the SUR1 IVS15-3c→t and exon 18 Thr759(ACC→ACT) polymorphisms and their haplotypes on glucose metabolism in French-Canadian Caucasians of the Québec Family Study (QFS).

There were only 43 diabetic subjects in QFS (Table 1). Unadjusted baseline data for the 685 ND subjects are presented in Table 2. In unrelated individuals from the parental generation (n = 259), the IVS15-3c→t polymorphism frequencies were 56% for the c allele and 44% for the t variant; and for the exon 18 Thr759(ACC→ACT) polymorphism, the frequencies were 95.8% for the C allele and 4.2% for the T variant. These frequencies are similar to those reported in ND Caucasians (8,9,11). No carriers of the rarer exon 18 Thr759(ACC→ACT) T allele (T⁺) were found among diabetic subjects, and IVS15-3c→t polymorphism frequencies were similar in diabetic and ND subjects (data not shown). Genotypes were in Hardy-Weinberg equilibrium and the two polymorphisms were in strong linkage disequilibrium (χ² = 28.4, P < 0.0001).

In ND subjects, the T⁺ group had significantly lower values than noncarriers (T^–) with regard to fasting plasma C-peptide (CPEP) and acute oral glucose tolerance test (OGTT) C-peptide response, defined as the area above basal in the first 30 minutes (CP30) (Fig. 1). The cT/tC double heterozygotes had even lower values than all cT^– haplotypes with regard to CPEP, CP30, and total OGTT C-peptide response (area above basal over 180 min) (CPtot). Trends were seen for lower values in the T⁺ group than in the T^– group with regard to acute OGTT insulin response (IN30), the area above basal in the first 30 min (P = 0.063, −14.4%). Results were independent of age, sex, and adiposity.

In women, CPEP and CP30 were significantly decreased in the T⁺ group compared with the T^– group and in the cT/tC group compared with the cT^– group (Table 3). CPtot followed the same pattern in cT/tC carriers. In men, fasting plasma glucose (GLU) was higher (P = 0.034, +4.2%), and IN30 tended to be lower (P = 0.074, −19.6%) in the T⁺ group than in T^– subjects.

In younger subjects (≤45 years old), CPEP was lower in the T⁺ group than in the T^– group (P = 0.026, −16.1%) and in the cT/tC group compared with the cT^– group (P = 0.045, −22.4%). IN30 tended to be lower in the T⁺ group than in T^– subjects (P = 0.080, −17.6%). CP30 was lower in T⁺ subjects than in T^– subjects (P = 0.058, −15.6%), and CPtot was decreased in the cT/tC carriers when compared with the cT^– group (P = 0.034, −25.0%).

With higher BMI, differences between genotype groups increased (Fig. 2). In those with BMI >25 kg/m², the T⁺ subjects had higher GLU (+4.6%) and lower CPEP (−18.1%), IN30 (−26.6%), and CP30 (−19.2%) than T^– subjects. The cT/tC group had even higher GLU (+5.9%) and lower CPEP (−24.8%), IN30 (−33.9%), and CP30 (−28.6%) values than the cT^– groups.

No significant differences between IVS15-3c→t genotypes were seen, and fasting plasma insulin (INS), OGTT glucose responses, and total insulin response were similar among the different genotypes (data not shown).

After taking into account independent family effects by adding a covariate for family membership in the analysis of variance for the above phenotypes, we found that the results were only modestly modified. For the overall cohort, P values for differences between T⁺ and T^– subjects were 0.072 for CP30, and for differences between cT/tC and non-cT haplotypes, the values were 0.119 for CP30 and 0.063 for CPtot. For all CPEP and glucose data and for all results in sex and BMI subgroups, differences between T⁺ and T^– and between cT/tC and cT^– groups remained significant at the same level (results not shown).

In this study, we have shown that the T allele of the exon 18 Thr759(ACC→ACT) polymorphism of the SUR1 gene was associated with lower CPEP and acute OGTT C-peptide responses in ND Canadians of French descent. Even lower CPEP and OGTT C-peptide responses were seen with the cT/tC haplotype, obtained from both the IVS15-3c→t and exon 18 Thr759(ACC→ACT) polymorphisms. These greater differences in the haplotype comparisons suggest that using two variants in the SUR1 gene more clearly defines the allele associated with the phenotypes mentioned. This supports findings in Danish Caucasians (8) of lower insulin and C-peptide secretion after intravenous tolbutamide in ND carriers of both the exon 18 C/T or T/T and the IVS15-3c→t c/t or t/t genotypes. However, in that study haplotypes could not be clearly identified; in the QFS they could be determined because family structures were known. Moreover, we could adjust metabolic parameters for important covariates such as waist circumference (WC) and the sum of six skinfold thicknesses (SF6).

Although we focused our study on ND subjects, for which interpretation of OGTT data are straightforward, analysis was also performed on all subjects combined. Similar results were obtained for C-peptide and insulin parameters (data not shown). Therefore, we do not believe that restricting analysis to ND subjects biased the results. There were no T⁺ subjects among the diabetic subjects. This should not be unexpected because QFS was not specifically designed to study type 2 diabetes and the numbers of diabetic subjects was small.

Lower OGTT responses could result either from decreased β-cell secretion or higher insulin sensitivity. Several arguments point to a relationship with secretion. First, SUR1 has been described mainly in pancreatic β-cells (6). Second, INS levels, which are variably correlated with insulin resistance, were unaffected by genotype status, yet CPEP levels were significantly changed. C-peptide is secreted in an equimolar fashion with insulin, has a longer plasma half-life, displays smaller oscillations in plasma concentration, and undergoes minimal hepatic extraction; therefore, it has been suggested to be a better reflection of insulin secretion overall (14). Third, OGTT glucose responses were not significantly different between the genotypes, whereas we would have expected a significant decrease in GLU and OGTT glucose responses in the case of greater insulin sensitivity with lower insulin requirements. Still, we cannot rule out a primary or compensatory increase in insulin sensitivity. Interestingly, in the Hansen et al. (8) study, the subjects displaying decreased insulin and C-peptide responses also showed a 30% increase in glucose effectiveness. Also, in a recent K_ATP channel–knockout mouse model (15), not only was insulin secretion decreased, but insulin action was enhanced, as measured by an insulin tolerance test, due either to a direct peripheral effect of the K_ATP channel deficiency or to an unknown compensatory mechanism.

The main genotype differences were found in C-peptide measures, although OGTT insulin responses also tended to be lower overall and were significantly lower in overweight subjects. This apparent discrepancy could be explained by the insulin assay used here, which crossreacts with proinsulin (PI). It has been shown that in deficient insulin secretion states, such as maturity-onset diabetes of the young, OGTT responses for C-peptide are more suppressed than those for immunoreactive insulin (IRI) (16). Moreover, in type 2 diabetes, PI is proportionately increased with respect to IRI and specific insulin (17). Hence, our results would suggest that the SUR1 variant carriers are also characterized by a deficient insulin secretion state. However, measures of PI and specific insulin would be required to clarify this point.

Significant associations were only seen in younger subjects. It is possible that in older individuals, age and accumulated environmental influences might predominantly modulate insulin secretion, overshadowing any underlying genetic factor and clouding differences between genotypes. Also, with increasing BMI, differences between T⁺ and T^– and between cT/tC and non-cT genotype groups became more pronounced for GLU, CPEP, IN30, and CP30. This pattern of higher GLU and lower OGTT secretory responses suggests that BMI directly influences these relationships. A possible explanation is the association between obesity and increased insulin resistance that leads to greater demand on β-cells, allowing differences in metabolic parameters to become more evident in SUR1 variant carriers, who would respond insufficiently because of limited secretion.

The exon 18 Thr759(ACC→ACT) polymorphism is silent; thus, it cannot alone explain our results. The IVS15-3c→t polymorphism is located by the intron-exon 16 splice junction, and thus an effect on mRNA splicing is possible but unproven. The most likely explanation is that these variants are in linkage disequilibrium with a nearby unidentified functional mutation, either in the SUR1 gene or in a gene close by. In the latter case, the Kir6.2 gene located only 4.5-kb distant on 11p15.1 is a good candidate.

In conclusion, the present study has shown for the first time that the SUR1 gene exon 18 Thr759(ACC→ACT) T allele and the cT haplotype, resulting from the combination of this polymorphism and the IVS15-3c→t polymorphism, are significantly associated with lower CPEP and OGTT C-peptide responses and possibly higher GLU and lower OGTT insulin responses in a large Caucasian population of French descent. Sex, age, and particularly BMI status modulated these associations. Because differences were greater with the cT haplotype, using two SUR1 variants may more clearly define the allele associated with altered glucose metabolism. However, these variants have not been shown to have direct functional consequences. Additional studies are needed to define the role of SUR1 or some nearby gene in insulin secretion and to identify the mutations that fully explain the results reported here.