Variants Within the Calpain-10 Gene on Chromosome 2q37 (NIDDM1) and Relationships to Type 2 Diabetes, Insulin Resistance, and Impaired Acute Insulin Secretion Among Scandinavian Caucasians
Variations in the calpain-10 gene (CAPN10) have been identified among Mexican-Americans, and an at-risk haplotype combination (112/121) defined by three polymorphisms, UCSNP-43, -19, and -63, confers increased risk of type 2 diabetes. Here we examine the three polymorphisms in 1,594 Scandinavian subjects, including 409 type 2 diabetic patients, 200 glucose-tolerant control subjects, 322 young healthy subjects, 206 glucose-tolerant offspring of diabetic patients, and 457 glucose-tolerant 70-year-old men. The frequency of the 112/121 combination was not significantly different in 409 type 2 diabetic subjects compared with 200 glucose-tolerant control subjects (0.06 vs. 0.05; odds ratio 1.32 [95% CI 0.58–3.30]). In glucose-tolerant subjects, neither the single-nucleotide polymorphisms individually nor the 112/121 combination were associated with alterations in plasma glucose, serum insulin, or serum C-peptide levels at fasting or during an oral glucose tolerance test, estimates of insulin sensitivity, or glucose-induced insulin secretion. In conclusion, the frequency of the 112/121 at-risk haplotype of CAPN10 is low among Scandinavians and we were unable to demonstrate significant associations between the CAPN10 variants and type 2 diabetes, insulin resistance, or impaired insulin secretion.
Familial aggregation and concordance rates of type 2 diabetes in monozygotic and dizygotic twins strongly argue for a genetic component involved in the pathogenesis of this disorder. Several genome-wide scans have been performed to identify susceptibility loci for type 2 diabetes, and several ethnic-specific susceptibility loci have been published (1–5). The first reported susceptibility locus for type 2 diabetes was assigned to chromosome 2q37 by analysis of 330 Mexican-American affected sibpairs, and the locus was designated as NIDDM1 (1). Subsequently, the NIDDM1 locus was shown to interact with a locus on chromosome 15q21 contributing to type 2 diabetes in Mexican-Americans (6), thereby supporting animal studies indicating that type 2 diabetes may result from epistatic interactions between genes (7–10).
Recently, fine mapping within the NIDDM1 region revealed a G/A polymorphism (UCSNP-43) located in the third intron of the calpain-10 gene (CAPN10) that was associated with type 2 diabetes (11). UCSNP-43 alone could not explain the previously reported linkage with the NIDDM1 locus in affected sibpairs, and, subsequently, various haplotypes in the CAPN10 region were examined (11). Three polymorphisms, UCSNP-43 (G→A within intron 3; allele 1 = G and allele 2 = A), UCSNP-19 (two repeats of 32-bp sequence, three repeats of 32-bp sequence within intron 6; allele 1 = 2 repeats and allele 2 = 3 repeats), and UCSNP-63 (C → T within intron 13; allele 1 = C and allele 2 = T) were identified to define the at-risk combined haplotype in Mexican-Americans. Heterozygosity for two common haplotypes (112 and 121), defined by UCSNP-43, -19, and -63, was associated with a 2.8-fold increased risk of type 2 diabetes among Mexican-Americans. Surprisingly, homozygosity for the 112 or the 121 haplotype was not associated with an increased risk for type 2 diabetes (11). Baier et al. (12) examined the effect of UCSNP-43 among Pima Indians and did not find an association with type 2 diabetes. However, among glucose-tolerant Pima Indians, homozygous carriers of the G-allele (G/G) had higher fasting plasma glucose and lower glucose turnover during a low-insulin euglycemic clamp than carriers of both G/A and A/A (12). Further support for UCSNP-43 being involved in type 2 diabetes was recently shown in an African-American study population in which an association between homozygosity of the G-allele of UCSNP-43 and type 2 diabetes was detected (13).
Calpains, consisting of an isoform-specific large subunit and an invariant small subunit, are a superfamily of related proteins, some of which have been shown to function as calcium-dependent cysteine proteases. In humans, 13 large subunit calpain genes and 1 small subunit calpain gene have been reported (14). Mutations in the muscle-specific CAPN3 cause limb girdle muscular dystrophy type 2A (15). In relation to type 2 diabetes, it is interesting that insulin-induced downregulation of IRS-1 in 3T3-L1 adipocytes is abolished by calpastatin, a calpain-inhibitor protein (16). Additional evidence for calpain playing a role in insulin signaling and also insulin secretion was recently reported by Sreenan et al. (17). Treatment with different calpain inhibitors was shown to result in enhanced glucose-induced insulin secretion in pancreatic islets and reduced insulin-stimulated glucose uptake into muscles and adipocytes (17). These results provide a tentative molecular mechanism to explain the association between variability in CAPN10 and type 2 diabetes. Here we report the analyses of three polymorphisms (UCSNP-43, -19, and -63) localized in CAPN10 in Scandinavian study populations. The objective of the present study was to test whether UCSNP-43 alone or the suggested at-risk haplotype combination (112/121) were associated with type 2 diabetes or diabetes-related phenotypes among glucose-tolerant subjects.
RESEARCH DESIGN AND METHODS
A total of 409 type 2 diabetic patients recruited from the outpatient clinic at Steno Diabetes Center and 200 age- and sex-matched glucose-tolerant control subjects living in the same area of Copenhagen as the patients recruited through the Danish Central Population Register were enrolled in the association study. Diabetes was diagnosed according to 1985 World Health Organization criteria. All control subjects underwent a 2-h oral glucose tolerance test (OGTT).
The control subjects had a mean age of 52 years (range 30–88) and a mean BMI of 25.4 kg/m2 (range 17.5–43.3). The patients had a mean age of 61 years (range 22–86), BMI of 29.0 kg/m2 (range 16.5–45.3), and reported average duration of diabetes of 6 years (range 0–28). Twenty-six percent of the patients were treated with diet alone, 57% with oral hypoglycemic agents, and 17% with insulin. All participants in the case-control study were Danish whites by self-report.
Further phenotype studies were performed in two Danish study populations and one Swedish Caucasian study population. One Danish study population comprised a population-based sample of 322 young healthy subjects. Physiological characterization of this population has been previously presented (18). The other Danish study population consisted of 206 glucose-tolerant offspring of type 2 diabetic probands from 62 families recruited from the Danish family resource bank at the Department of Human Genetics, University of Copenhagen, or from the outpatient clinic at Steno Diabetes Center. Physiological characterization of this population has been reported previously (19). The participants in both Danish study populations underwent a tolbutamide-modified intravenous glucose tolerance test (IVGTT) for measurements of the acute insulin and C-peptide responses and the insulin sensitivity index (Si) (18). Disposition index was calculated as the product of the Si and the acute insulin response during an IVGTT. All IVGTTs were performed in the morning in the fasting state.
The Swedish cohort, consisting of 457 70-year-old glucose-tolerant individuals, was a subpopulation of a previously described longitudinal study examining a population-based sample of 1,221 Swedish men (20). The participants underwent a 120-min euglycemic-hyperinsulinemic clamp study (56 mU · min−1 · m−2) at 70 years of age for measurements of insulin sensitivity (21). Homeostasis model assessment (HOMA) of insulin resistance was calculated as the product of fasting serum insulin and fasting plasma glucose.
In the Danish study populations, plasma glucose was analyzed by a glucose oxidase method, serum-specific insulin [excluding des(31,32) and intact proinsulin] was measured by enzyme-linked immunosorbent assay, and serum C-peptide was determined by radioimmunoassay using Steno Diabetes Center routine methods. In the Swedish study population, plasma glucose was analyzed by a glucose dehydrogenase method and plasma insulin was measured by an enzymatic-immunological assay. Fasting samples were taken in the morning after an overnight fast.
Informed consent was obtained from all the studied subjects before participating in the study. The study was approved by the ethics committees of Copenhagen and the Medical Faculty of Uppsala University and was in accordance with the principles of the Declaration of Helsinki II.
Genotyping was carried out by applying PCR on genomic DNA isolated from human leukocytes. PCR of DNA segments containing SNP-19, -43, and -63 were carried out with the following primers: SNP-19, forward primer 5′-GTT TGG TTC TCT TCA GCG TGG AG-3′ and reverse primer 5′-CAT GAA CCC TGG CAG GGT CTA AG-3′; SNP-43, forward primer 5′-CGC TCA CGC TTG CTG CGA AGT AAT G and reverse primer 5′-ACA TTT TGC TGC CGG TCA TGC TCG TAG GAT GCA TGG ACC CTA G-3′; SNP-63, forward primer 5′-AGG GCC TGA CGG GGG TGG CG-3′ and reverse primer 5′-ACA TTT TGC TGC CGG TCA AGC GCT CCC AGGC TCC TGA TC-3′. SNP-19 is a 32-bp deletion polymorphism. SNP-43 (NsiI) and SNP-63 (Hha1) were genotyped by restriction fragment-length polymorphism. Restriction site-generating PCR was carried out for genotyping of SNP-43 and -63. Mismatched bases are indicated by bold letters in the forward primers of SNP-43 and -63. PCR segments, including SNP-43 and -63, contain internal control restriction enzyme sites to verify successful digestion. These sites were introduced in the reverse primers (in italic in the SNP-43 and -63 reverse primers).
Fisher’s exact test was applied to examine for differences in allele and genotype frequencies. Differences between genotype groups among control subjects, young healthy subjects, and 70-year-old Swedish men were tested with one-way ANOVA, including sex and genotype as fixed factors and BMI and age as covariate factors. For analysis of data obtained from the offspring of diabetic probands, a variance component model was used as described (19). When analyzing plasma glucose, serum insulin, or serum C-peptide at 60 min during an OGTT, the corresponding basal levels were included as covariates in the model. Similarly, basal and 60-min values were included in the model for analysis of the corresponding 120-min levels during an OGTT.
The odds ratio (OR) (and 95% CIs) for the Danish, German, British/Irish, and Finnish datasets was computed separately using hypergeometric distribution. Logistic regression models were fitted to assess whether the ORs were equal between populations and to estimate a common value, controlling for population.
The UCSNP-43, -19, and -63 were examined in an association study consisting of 409 Danish type 2 diabetic patients and 200 age- and sex-matched glucose-tolerant control subjects (Table 1). Clinical and biochemical data of the two populations included in the association study are available in an online appendix (http://diabetes.diabetesjournals.org). The UCSNP-43, -19, and -63 were in Hardy-Weinberg equilibrium (Table 1). The allele frequencies of the examined polymorphisms were 0.27 (95% CI 0.24–0.30) and 0.28 (0.24–0.32) for UCSNP-43, P = 0.79; 0.62 (0.59–0.65) and 0.61 (0.56–0.66) for UCSNP-19, P = 0.53; and 0.07 (0.05–0.09) and 0.07 (0.04–0.10) for UCSNP-63, P = 0.92, for type 2 diabetic patients and control subjects, respectively.
The three polymorphisms were, as expected from previously reported studies, in strong linkage disequilibrium (control subjects: UCSNP-43 and -19, χ2 = 118, P < 0.0001; UCSNP-43 and -63, χ2 = 32, P < 0.0001; and UCSNP-19 and -63, χ2 = 56, P < 0.0001). To identify common haplotypes in Danish Caucasians, we genotyped the polymorphisms in 62 families ascertained through one type 2 diabetic proband with four or more nondiabetic offspring. In these families we identified, in agreement with previous findings (11), four haplotypes: 111 (referring to UCSNP-43, -19, and -63), 121, 112, and 221 (data not shown). Four haplotypes can maximally result in 10 different genotypes, which indeed was the upper limit of genotype combinations we observed in all study populations, indicating that other haplotypes are rare among Scandinavian Caucasians. The estimated haplotype frequencies were 0.31 (95% CI 0.28–0.34) and 0.32 (0.27–0.39) for 111, P = 0.59; 0.35 (0.32–0.38) and 0.33 (0.28–0.38) for 121, P = 0.46; 0.07 (0.05–0.09) and 0.07 (0.04–0.10) for 112, P = 0.91; and 0.27 (0.30–0.33) and 0.28 (0.24–0.32) for 221, P = 0.80, among type 2 diabetic patients and control subjects, respectively.
Examination of all three polymorphisms in the association study resulted in 10 different combined genotypes (Table 2). The frequencies of the suggested at-risk combined genotype 11 12 12 (composed of the 112 and 121 haplotypes) were 0.06 among type 2 diabetic patients and 0.05 among control subjects; therefore, the at-risk haplotype combination (112/121) was not associated with increased risk of diabetes (OR 1.32 [95% CI 0.58–3.30]). In general, no genotype combination was significantly associated with type 2 diabetes (Table 2). However, combining the present dataset with previously published datasets from other European study populations (11,22), identified through a literature search on PubMed, a public service of the National Library of Medicine, revealed a significant association of the at-risk haplotype combination (112/121) with type 2 diabetes (OR 1.62 [95% CI 1.01–2.58], P = 0.04) (Table 3).
The effects of UCSNP-43, -19, and -63 on diabetes-related phenotypes were examined in Scandinavian individuals, including three Danish study populations: 200 glucose-tolerant control subjects (Table 4), 322 young healthy subjects (Table 5) and 206 glucose-tolerant offspring of 62 diabetic patients (Table 6), and one Swedish study population of 457 glucose-tolerant 70-year-old men (Table 7). Based on results from Pima Indians and Mexican-Americans, we decided to stratify the examined study populations according to UCSNP-43 (G/G versus combined G/A and A/A) and the at-risk combined haplotype of UCSNP-43, -19, and -63 (121/112 versus all other combinations). Clinical and biochemical data of the study populations, stratified according to the individual polymorphisms (UCSNP-43, -19, and -63) as well as all estimated haplotype combinations, are available in the online appendix.
We found no effect of UCSNP-43 (G/G versus combined G/A and A/A) or the at-risk haplotype combination (112/121 versus all other haplotype combinations) on fasting levels of plasma glucose, serum insulin or serum C-peptide, BMI, or waist-to-hip ratio (Tables 4–7). The CAPN10 variation also had no significant effect on insulin sensitivity, as estimated by HOMA (23) in the 200 control subjects, by the Si (24) in 322 young healthy subjects and 206 glucose-tolerant offspring of type 2 diabetic patients, or by the euglycemic-hyperinsulinemic clamp (25) in 457 70-year-old men (Tables 4–7). Likewise, among 322 young healthy subjects and 206 glucose-tolerant offspring of type 2 diabetic patients, no associations of UCSNP-43 or the haplotype combination 112/121 with altered acute insulin and C-peptide responses during an IVGTT were observed. This was also the case for the disposition index.
Among the 200 glucose-tolerant control subjects and the 206 glucose-tolerant offspring of type 2 diabetic patients, no associations of UCSNP-43 with plasma glucose, serum insulin, and serum C-peptide levels during an OGTT were observed. In regard to the at-risk haplotype combination (112/121), we found, among glucose-tolerant control subjects, that carriers of the 112/121 haplotype combination had significantly increased serum insulin levels at 60 min during the OGTT compared with carriers of all other haplotype combinations (P = 0.009) (Table 4). However, this finding was not replicated among glucose-tolerant offspring (P = 0.58) (Table 6). Finally, individual analyses of each polymorphism did not reveal any consistent association to the examined variable in the Scandinavian study populations (data available in online appendix).
Recent studies in Mexican-Americans (11) have identified genetic variability located in the NIDDM1 region that shows association with type 2 diabetes. Three polymorphisms located within CAPN10 in the NIDDM1 region, UCSNP-43, -19, and -63, constituted an at-risk haplotype combination (112/121) among Mexican-Americans (11). Other studies among British/Irish (22) and Samoan (26) study populations failed to show association between the at-risk haplotype combination and type 2 diabetes. However, the British/Irish study showed increased transmission of a rare allele (UCSNP-44), which is in linkage disequilibrium with an amino acid polymorphism at CAPN10, to affected offspring in parent-offspring trios (22), indicating that variation in CAPN10 affects susceptibility to type 2 diabetes. We did not detect any association of the 112/121 haplotype combination with diabetes in Danish Caucasians, but it must be emphasized that because of the low frequency of the 112/121 haplotype combination, the power of the present association study to detect mild effects is low. For example, the power to detect a difference of 0.022 in genotype frequency is 20%, and if a more acceptable power of >85% is desirable, the difference in genotype frequency has to be >0.055. To increase the power, we combined the previously reported European studies comprising Finnish, German, and British/Irish datasets with the Danish dataset, and we actually identified a significant association to type 2 diabetes (OR 1.62 [95% CI 1.01–2.58], P = 0.04). The frequency of the at-risk haplotype combination (121/112) in a random sample of Mexican-Americans was previously reported to be 0.08 (0.03–0.13), and we found a frequency among glucose-tolerant control subjects of 0.05 (0.02–0.08). However, the 112 allele is much more frequent in Mexican-Americans than Danish Caucasians (0.23 [0.17–0.29] and 0.07 [0.04–0.10], respectively). This difference in allele frequency may offer a tentative explanation of the apparent higher impact of the 112/121 haplotype combination among Mexican-Americans.
The impact of CAPN10 variation on diabetes-related phenotypes among nondiabetic subjects has also been investigated (12,27). Baier et al. (12) examined the effect of UCSNP-43 and found, among Pima Indians, a significantly reduced glucose disposal for G/G carriers compared with combined G/A and A/A carriers, and, in the British study (27), it was shown that subjects with the at-risk haplotype combination had increased fasting and 2-h plasma glucose levels during an OGTT compared with the rest of the study population. We examined the 112/121 haplotype combination in 322 young healthy subjects, 200 glucose-tolerant control subjects, 206 glucose-tolerant offspring of type 2 diabetic patients, and 457 70-year-old Swedish men but failed to show any significant association with insulin resistance or impaired insulin secretion. Similarly, we also failed to find any effects of the UCSNP-43 on estimates of insulin resistance as reported by Baier et al. (12). Especially the results from the study of 457 glucose-tolerant Swedish men (Table 7), characterized by an euglycemic-hyperinsulinemic clamp (comparable to the low insulin clamp in the Pima Indian study), emphasized that we were unable to replicate the results from Pima Indians. Four genome-wide scans in Finnish (2,28), Utah Caucasian (4), and French (29) families failed to identify type 2 diabetes susceptibility loci on chromosome 2q37. Furthermore, several studies examining various DNA microsattelites in the vicinity of the NIDDM1 locus in affected sibpairs of British (30,31), Sardinian (32), French (33), and Finnish (28) origin were unable to show evidence for linkage of this region to type 2 diabetes. These studies of the NIDDM1 locus point to the challenges in replicating positive linkages obtained in various ethnic study populations.
In conclusion, among the examined glucose-tolerant study populations, we did not find significant associations between the previously reported variants of CAPN10 with estimates of insulin resistance or glucose-induced insulin secretion. The frequency of the diabetes at-risk haplotype combination is low among Scandinavians, and we were unable to demonstrate association between the at-risk haplotype combination and type 2 diabetes. However, the combined results of the four European studies suggest that despite the low prevalence of the at-risk genetic variants in CAPN10, these alleles may add to the risk of type 2 diabetes among North Europeans.
The study was supported by grants from the Danish Medical Research Council, the Danish Research Academy, the Danish Diabetes Association, the Velux Foundation, and the European Economic Community (BMH4-CT98-3084).
The authors thank Helle Fjordvang, Annemette Forman, Lene Aabo, Bente Mottlau, Susanne Kjelberg, Jane Brønnum, Lis Ølholm, Maja Lis Halkjær, and Quan Truong for technical assistance; Bendix Carstensen for statistical assistance; and Grete Lademann for secretarial support.
Address correspondence and reprint requests to Søren K. Rasmussen, PhD, Symphogen A/S, Elektrovej, Building 375, DK-2800 Lyngby, Denmark. E-mail:.
Received for publication 13 March 2002 and accepted in revised form 4 September 2002.
Additional information for this article can be found in an online appendix at http://diabetes.diabetesjournals.org.
HOMA, homeostasis model assessment; IVGTT, intravenous glucose tolerance test; OGTT, oral glucose tolerance test; Si, insulin sensitivity index.