To the Editor: Although the physiopathological bases of Type 2 diabetes mellitus are well established, with an important role of generalized insulin resistance hitting key organs for glucose homeostasis—like muscle, liver and beta-cell—little is known about its molecular determinants and about the genetic factors involved in the transition between obesity and Type 2 diabetes. A considerable amount of evidence has suggested that a polymorphism in fat-expressed PPAR-γ, the major thiazolidinedione target, could contribute, even modestly, to the genetic risk for Type 2 diabetes [1]. In contrast, genome wide scans presented evidence of linkage between a region of chromosome 3q27, the metabolic syndrome and Type 2 diabetes in American and French Caucasian subjects [2], and also with coronary heart disease in Indo-Mauritians [3]. We recently replicated the linkage with diabetes in Japanese Type 2 diabetic families [4]. A strong positional candidate gene located on 3q27 is APM1, which encodes for the adipocyte complement related protein (Acrp30, also named adiponectin) which is specifically expressed in differentiated adipocytes. It was recently shown that its proteolytically-cleaved product (gAcrp30) lowers glucose concentrations through fatty acid oxidation without variations of insulin and glucagon concentrations, and prevents the development of high-fat diet induced obesity [5]. Furthermore, decreased expression of adiponectin correlates with insulin resistance in mouse models of obesity or lipoatrophy, and physiological doses of adiponectin decrease insulin resistance in these mice by lowering triglyceride content in muscle and liver [5]. In human, hypoadiponectinaemia was reported in obese and Type 2 diabetic subjects, and adiponectin concentrations are closely associated with insulin sensitivity in different ethnic groups [6]. We have recently shown that sequence variations in APM1 gene are associated with circulating adiponectin concentrations and modulate the risk for insulin resistance and diabetes in French Caucasians [7]. A previous study has also suggested a large heritability for plasma adiponectin concentrations in humans [8].

In the Japanese population, we have done a case control study and reported an association between two single nucleotide polymorphisms (SNP) in the APM1 gene and Type 2 diabetes [9]. To confirm such relevant associations in another Japanese population, we studied 359 diabetic Japanese subjects from the 164 sib-ships where indication of linkage at 3q27 was found [4]. One affected subject without obesity was randomly chosen from each family (average BMI: 22.7±2.8 and 23.2±3.6 kg/m2 for men and women, respectively), and 183 non-diabetic Japanese subjects was the control group (average BMI: 22.3±3.1 and 22.5±3.4 kg/m2 for men and women, respectively). The 11 most frequent polymorphisms of the APM1 gene were studied by direct sequencing and Light Cycler assay (Roche, Manheim, Germany). The genotype distribution of the SNP and insertion or deletion variants is included in Table 1. No significant deviation of the Hardy-Weinberg equilibrium was observed for any of the SNPs studied.

Table 1. Distribution of genotypes for the eleven common SNP of APM1 and results of the association study with Type 2 diabetes
Full size table

A significant association between the C/C genotype at position −11377 in the APM1 promoter and Type 2 diabetes (p=0.002) was observed (Table 1). This promoter variant, under a recessive model of inheritance, increases the risk for Type 2 diabetes (OR=2, IC=1.30–3.08) in the population tested. A similar, although not significant, trend for association with Type 2 diabetes was reported in the previous Japanese case control study (p=0.25) [9]. A trend for association (p=0.08) was also found for the +276 G>T SNP in intron 2, which was shown to be associated with insulin sensitivity, adiponectin concentrations and Type 2 diabetes in the Japanese population [9]. The difference observed in the strength of the genetic association between the two populations can be explained by the fact that diabetic subjects of Hara's population were randomly recruited from patients attending the outpatient clinic, whereas in our study nuclear Japanese families with at least two Type 2 diabetic siblings were analyzed. However, clinical characteristics of these two populations seem to be comparable. In addition, we observed trends for association between Type 2 diabetes and the C>T SNP at −11043 in the promoter and A>G SNP at −3971 in intron 1 (Table 1).

We have calculated standardised linkage disequilibrium D' between each pair of SNPs using the PM+EH+ software. We observed a block of six closely associated variants showing D' values ranging from 0.9698 to 1.00 and comprising the –11377 promoter variant. Of note, this LD block includes four promoter and two nearby intron 1 variants, and that SNP +276 is not part of this group. The variant alleles of the other SNPs (−4041, −3971 and −11426) are very rare and not surprisingly fail to detect the same association as −11377 SNP.

An haplotype including two SNPs at positions −11391 and –11377, both located in the promoter sequence of APM1, was shown to be strongly associated with adiponectin concentrations and Type 2 diabetes in the French population [7]. Thus, we have tested if any haplotype prevalence might display a greater difference between cases and control subjects. No haplotype showed a better result than that obtained for the SNP at −11377 by itself. In the Japanese population, in contrast to French Caucasian subjects, the −11391 SNP is rather uncommon (frequency ≈0.02) [9], and we did not detect this variant in our study. Thus, even if –11391 SNP is a functional variant, because of its low frequency, it is unlikely to be involved in the genetic risk for Type 2 diabetes in Japan.

We also investigated the influence of the genotypes at the –11377 SNP on the modest linkage observed at the 3q27 locus in Japanese sib-pairs (maximum logarithm of odds score (MLS) =1.38, [4]). The pairs of sibs Identical by State (IBS) for genotypes 11, 12 and 22 were tested for linkage separately. Although the lod-score of pairs IBS for the 11 genotype reached a value of 2.0, it did not increase. Thus, we could not obtain evidence that the 11 genotype by itself contributes, even partially, to the observed linkage at the 3q27 locus.

Our results strengthen the contention that variations in the APM1 gene contribute to the genetic risk for Type 2 diabetes in the Japanese population. As the C/G SNP at –11377 is frequent, we calculated an attributable risk of 31%, i.e. 1/3 of the genetic risk for diabetes in the Japanese subjects could be related to this variant. All together, our data strongly support the hypothesis that adiponectin plays a pivotal role in the pathogenesis of Type 2 diabetes. Finally, adiponectin might be considered as a molecular link between the metabolic syndrome, obesity and Type 2 diabetes.