Role of rare HNF1A variants function in monogenic and type 2 diabetes

Abstract

Variants in the transcription factor gene HNF1A have been identified in subjects with maturity-onset diabetes of the young (MODY) type 3, type 2 diabetes, as well as in children with apparent type 1 diabetes. One of the challenges in a clinical setting is distinguishing MODY patients from those with type 2 diabetes, as there is considerable overlap in terms of clinical features. Mapping HNF1A variants to the correct clinical phenotype requires functional characterization of variants effects on hepatocyte nuclear factor – 1A (HNF-1A) function. Large whole-exome sequencing of an Mexican and American Latino population, reported in Paper I, identified a low frequency rare variant p.(E508K) in HNF1A that confers increased risk for type 2 diabetes up to 5 fold (odds ratio (OR)=5.48; P=4.4 x 10-7). Functional investigation of this p.(E508K) HNF-1A protein variant demonstrated reduced transactivation activity <50%, low protein level expression and slightly impaired nuclear localization. These findings suggest that the p.(E508K) HNF1A variant mediates a mild-loss of function of HNF-1A and represents a risk variant for type 2 diabetes in the Mexican and American Latino population. Exome sequencing of HNF1A in 4,115 well-phenotyped individuals from the general population (Framingham Heart Study cohort, the Jackson Heart Study cohort, and type 2 diabetes case and control patients from the extreme type 2 diabetes cohort) have previously shown that 1/50 individuals with diabetes harbors a missense variant in the HNF1A gene. 27 rare HNF1A missense variants were identified. In Paper II we show that after using bioinformatics prediction tools to determine predicted pathogenic effect, none of the 27 HNF1A variants that were classified as pathogenic were associated with risk for type 2 diabetes in the population cohorts (OR=2.02; 95% CI 0.73-5.60; P=0.18). We further evaluated the functional consequences of individual variants by their effect on HNF-1A transcriptional activation, DNA binding, and subcellular (cytoplasmic/nuclear) localization. Furthermore, association between type 2 diabetes and different functional assay models was assessed. A transcriptional activity with a threshold of <60% compared to wild-type HNF-1A activity was able to best predict type 2 diabetes association with carrier type 2 diabetes phenotype (OR=5.04; 95% ; P=0.0007), and indicate that 0.44% of the population carry HNF1A variants that results in substantially increased risk for developing the disease. To improve the diagnostic interpretation of the increasing number of HNF1A variants identified by next-generation sequencing, there is a future demand for robust and reliable high-throughput functional investigations of variant effects on normal HNF-1A protein function. In Paper III, we searched systematically for endogenous regulated HNF-1A transcripts as possible markers for investigating diabetes associated HNF1A variants effects. For this purpose we generated HNF1Afree liver specific cell lines (HuH7 and HepB3) by knocking out endogenous HNF1A using CRISPR/Cas9, prior to the controlled re-expression (doxycycline induced) of wild-type HNF-1A versus HNF-1A variants (MODY3, type 2 diabetes). The gene expression profile analyzed by RNA sequencing identified significantly differentially expressed genes upon overexpressing HNF-1A, of which the top 20 upregulated genes were further investigated and many found down-regulated by MODY3-causing HNF1A variants. Of these, ABCC2, FABP1 and HABP2 genes in HuH7, and HKDC1, HRG and KL genes in Hep3B cell lines, were considered as potential targets for future large-scale and high-throughput investigations of numerous HNF1A variants.