<emph type="ital">HMGA1</emph>, A Novel Locus for Type 2 Diabetes Mellitus

Abstract

BOTH GENETIC AND ENVIRONMENTAL FACTORS CONtribute to the pathogenesis of type 2 diabetes mellitus (DM). Linkage analysis and genome-wide association studies have revealed 34 common variants (also called single-nucleotide polymorphisms) associated with type 2 DM. Most of the loci are associated with either abnormal insulin processing or secretion, suggesting that most of the risk of type 2 DM in the population is due to beta cell dysfunction. However, some type 2 DM loci, such as peroxisome proliferator-activated receptor gamma (PPARG) and Kruppel-like factor 14 (KLF14), indicate defective insulin action or insulin resistance as a contributor. In addition, some genes (eg, glucokinase [hexokinase 4] regulator, insulin-like growth factor 1 [IGF1], and the fat mass and obesity associated gene) are associated with fasting insulin, insulin resistance, and obesity and may also contribute to type 2 DM. Most of the identified variants have modest effect sizes (1.1-1.3), and all of the type 2 DM–associated variants can explain only 10% to 15% of the heritability. Thus, there is a need to identify additional novel loci for type 2 DM. One approach is to conduct large-scale meta-analyses or association analyses with common variants. Another is to find rare variants in candidate genes selected for their known role in biological processes related to the disease. In this issue of JAMA, Chiefari and colleagues used the second approach to test for association between rare variants in high-mobility group A1 (HMGA1) gene and type 2 DM by sequencing the promoter region, all exons and splice sites, and the 3 untranslated region in 3278 patients with type 2 DM and 3328 control study participants. Patients with type 2 DM had increased prevalence compared with controls of having rare heterozygous variants: c.136-14_136-13insC (7.23% vs 0.43%, respectively), c.310G T (p.E104X) (0.43% vs 0%), c.*82A G (1.4% vs 0.2%), and c.*369delG (0.73% vs 0%). The authors confirmed the increased prevalence of the c.13614_136-13insC variant in 2 smaller cohorts of patients with type 2 DM compared with controls from the United States (7.7% vs 4.7%, respectively) and France (7.6% vs 0%). Thus, Chiefari et al add another locus to the list of previously known loci, which includes a highly homologous gene, HMGA2. However, there are some concerns regarding the conclusions reached by the authors. First, despite one of the alleles (c.136-14_136-13insC) consistently having a frequency of 7.2% to 7.7% in the 3 cohorts with type 2 DM, there was more than 10-fold variation in the frequency of this variant among the control populations (0.43%-4.7%). While the authors provide possible explanations for this discrepancy based on different inclusion criteria for selecting controls from Italy, the United States, and France, it makes it difficult to ascertain the precise effect size associated with this variant for type 2 DM. Second, the replication cohorts were not large enough. In the US cohorts, there were 970 patients with type 2 DM and 958 controls. In the French cohorts, there were 354 patients with type 2 DM and 50 controls. Thus, this observation must be confirmed in large case-control studies. Third, despite the increased prevalence of the rare variants in patients with type 2 DM, none was reported to harbor the homozygous variant or compound heterozygous variants. Thus, the dose effect of these variants on phenotype or on functional assays could not be ascertained. HMGA1 encodes mainly 2 isoforms: a full-length form with 107 amino acids and a truncated protein of 96 amino acids with an internal deletion of 11 amino acids. HMGA1 belongs to the high-mobility group family of proteins, which are the second most abundant nuclear proteins after histones. HMGA1 influences many biological processes including cell differentiation, growth, proliferation, and death and is also involved in oncogenic transformation. HMGA1 binds AT-rich DNA sequences in the promoter region of many genes, including insulin receptor (INSR) and IGF-1 receptor (IGF1R). In support of their observations, Chiefari et al showed reduced