PAX4 gene variations predispose to ketosis-prone diabetes

Abstract

Ketosis-prone diabetes (KPD) is a rare form of type 2 diabetes, mostly observed in subjects of west African origin (west Africans and African-Americans), characterized by fulminant and phasic insulin dependence, but lacking markers of autoimmunity observed in type 1 diabetes. PAX4 is a transcription factor essential for the development of insulin-producing pancreatic beta-cells. Recently, a missense mutation (Arg121Trp) of PAX4 has been implicated in early and insulin deficient type 2 diabetes in Japanese subjects. The phenotype similarities between KPD and Japanese carriers of Arg121Trp have prompted us to investigate the role of PAX4 in KPD. We have screened 101 KPD subjects and we have found a new variant in the PAX4 gene (Arg133Trp), specific to the population of west African ancestry, and which predisposes to KPD under a recessive model. Homozygous Arg133Trp PAX4 carriers were found in 4% of subjects with KPD but not in 355 controls or 147 subjects with common type 2 or type 1 diabetes. In vitro, the Arg133Trp variant showed a decreased transcriptional repression of target gene promoters in an alpha-TC1.6 cell line. In addition, one KPD patient was heterozygous for a rare PAX4 variant (Arg37Trp) that was not found in controls and that showed a more severe biochemical phenotype than Arg133Trp. Clinical investigation of the homozygous Arg133Trp carriers and of the Arg37Trp carrier demonstrated a more severe alteration in insulin secretory reserve, during a glucagon-stimulation test, compared to other KPD subjects. Together these data provide the first evidence that ethnic-specific gene variants may contribute to the predisposition to this particular form of diabetes and suggest that KPD, like maturity onset diabetes of the young, is a rare, phenotypically defined but genetically heterogeneous form of type 2 diabetes.

Figure 1.

(A) Structure of the PAX4 molecule and position of the R37W and R133W variants. (B) Effect of the PAX4 mutations on gene transcription activity in α-TC1.6 cells. The wild-type and mutant hPAX4 (WT, R133W, R37W) were co-transfected with an insulin promoter reporter construct (−410Rins-pfoxluc) in α-TC1.6 cells. The amount of reporter plasmid used was 2 μg/lane (lanes 1–5). The amount of co-transfected hPAX4 constructs (wt and variants) was 50 ng/lane(lanes 3–5). Data are expressed as mean ± SE. *P < 0.01 comparing R133W or R37W to WT. (C) Effect of PAX4 mutations on PAX4 binding to target genes.An EMSA using control (lane 1), wild-type PAX4 (lane 2), R37W variant (lane 3) and R133W variant (lane 4) is shown. The arrow indicates binding of PAX4 proteins to the probe. The oligonucleotide probes were the human PAX4 promoter (left) and the rat insulin promoter (right). Identical results were obtained in at least two independent experiments.

Figure 2.

β-cell insulin secretory reserve was assessed in a cohort of ketosis-prone diabetic subjects carrying the wild-type PAX4 (R/R, n = 18), the heterozygote (R/W, n = 11) and the homozygous mutation (W/W, n = 4), by measuring basal C-peptide (0 min) and C-peptide response following IV glucagon injection (8 min) after correction of hyperglycemia. *P < 0.05, R/R versus W/W.