Identification and characterization of candidate genes for type 2 diabetes in the GK rat

Abstract

Type 2 diabetes is a common multifactorial disorder with complex genetics, which presents with a heterogeneous clinical phenotype as a result of interaction between genetic susceptibility and environmental risk factors. Hence, the search for susceptibility genes for type 2 diabetes in man is difficult. Inbred animal models therefore provide an important means for the reduction of this complexity, both in genetic and environmental perspectives. The GK rat is a well-characterized animal model for polygenic type 2 diabetes; it displays many features of the metabolic disease usually described in humans with type 2 diabetes. The goal of this study was to identify diabetes susceptibility genes and determine their pathophysiological role.

In order to map susceptibility loci for type 2 diabetes in GK, two crosses between the GK rat and the normoglycemic F344 rat were arranged. These crosses yielded 374 second-generation progeny (F2), which were subjected to an intra-peritoneal glucose tolerance test (IPGTT), determination of weight, and genome-wide genotyping. The F2 progeny were subjected to a genome-wide search with 253 simple sequence length polymorphism (SSLP) markers in the 45 progeny with the most extreme glucose values. Following proper adjustments for sex, reciprocal cross, and age, data were analyzed for linkage. Despite the polygenic etiology of diabetes in GK, three genome-wide significant (P < 5 x 10-5) quantitative trait loci (QTL) for diabetes were identified, Niddm1, Niddm2, and Niddm3 on chromosomes 1, 2, and 10. Niddm1, which is the major diabetes-QTL with a maximum LOD score of 11.0, exerts a strong effect on postprandial glucose levels and weaker effects on body weight and insulin levels.

A breeding protocol was established to transfer the GK-allele of the Niddm1-locus onto the F344 genetic background to produce the Niddm1a homozygous congenic strain (52 cM). This strain showed significantly higher postprandial glucose levels compared to F344 during IPGTT. To sub-map and further investigate the observed phenotype associated with Niddm1, five sub-strains from Niddm1a were established. Two of these, Niddm1b and Niddm1i retained 28 and 22 cM of non-overlapping GK-derived alleles. Rats from the Niddm1b and Niddm1i strains were characterized physiologically regarding a number of diabetes-associated phenotypes. Niddm1i affects both insulin secretion in pancreatic ß-cells and insulin action in adipocytes, while Niddm1b rats showed signs of insulin resistance with age that were associated with obesity, hyperinsulinemia, and dyslipidemia. Since Niddm1b and Niddm1i have no overlap of GK-derived alleles and different profiles of diabetes-associated phenotypes, it was concluded that at least two different genes in Niddm1 affect glucose homeostasis.

The other three sub-strains (Niddm1c, Niddm1e, and Niddm1f), that together contained different parts of Niddm1b were tested for a number of phenotypes. One of the strains, Niddm1c, containing the major part of Niddm1b was indistinguishable from F344 in postprandial glucose and insulin. The two other congenic strains, Niddm1e, and Niddm1f), exhibited similar diabetes-associated phenotypes, as Niddm1b did. This demonstrates that the diabetes locus located in Niddm1b is fine mapped to the 3.7 cM-interval carried by Niddm1e.

QTL identification and positional cloning of disease susceptibility genes relay on dense and accurate genetic maps. A genetic linkage map of the rat, containing 400 SSLP markers, was constructed using data from the GK/F344 F2 intercross. The map provides markers with an average spacing of 4.4 cM, and spans a total genetic length of 1,782 cM. A number of 16 new genes and 51 new anonymous SSLPs were added to the existing rat genetic linkage map. This genetic linkage map is derived from a large set of informative meiosis, and most of the framework loci were analyzed in at least 235 animals or more. The majority of markers were also grouped with highly stringent criteria for linkage, ensuring high quality of distances and locations in the reported map.

A dense map of genes was required within the Niddm1 region for localization and positional cloning of diabetes genes in GK. Radiation hybrids were used to map ten candidate genes in the Niddm1-region. Two of the candidate genes (ADP-ribosylation factor-like protein 3 and stearoyl-CoA desaturase 1) that lie in the genetic region covered by Niddm1i were also analyzed by DNA sequencing. Sequence analysis of the complete translated parts of these genes revealed no nucleotide differences between GK and F344. Based on these data none of the structural parts of these candidate genes are likely candidates for the Niddm1 phenotype in GK rats. On the other hand, a candidate gene, encoding insulin degrading enzyme (IDE), located within the Niddm1e, region, revealed that the variant of IDE encoded by GK has reduced activity, as demonstrated by in vitro-expression. Insulin degrading enzyme is suggested to be one of the factors contributing to the diabetic phenotype in GK rats.

Finally, this study emphasizes the pathophysiological and genetic complexity of diabetes, even within an apparently single QTL, and demonstrates the potential of the GK model in transforming the multifactorial diabetes phenotype into single traits that are amenable to mapping and positional cloning of disease-pathways in type 2 diabetes.