Homozygous Hypomorphic HNF1A Alleles Are a Novel Cause of Young-Onset Diabetes and Result in Sulfonylurea-Sensitive Diabetes

Abstract

OBJECTIVE

Heterozygous loss-of-function mutations in HNF1A cause maturity-onset diabetes of the young (MODY). Affected individuals can be treated with low-dose sulfonylureas. Individuals with homozygous HNF1A mutations causing MODY have not been reported.

RESEARCH DESIGN AND METHODS

We phenotyped a kindred with young-onset diabetes and performed molecular genetic testing, a mixed meal tolerance test, a sulfonylurea challenge, and in vitro assays to assess variant protein function.

RESULTS

A homozygous HNF1A variant (p.A251T) was identified in three insulin-treated family members diagnosed with diabetes before 20 years of age. Those with the homozygous variant had low hs-CRP levels (0.2–0.8 mg/L), and those tested demonstrated sensitivity to sulfonylurea given at a low dose, completely transitioning off insulin. In silico modeling predicted a variant of unknown significance; however, in vitro studies supported a modest reduction in transactivation potential (79% of that for the wild type; P < 0.05) in the absence of endogenous HNF1A.

CONCLUSIONS

Homozygous hypomorphic HNF1A variants are a cause of HNF1A-MODY. We thus expand the allelic spectrum of variants in dominant genes causing diabetes.

Introduction

Heterozygous loss-of-function mutations in the hepatocyte nuclear factor 1-α gene (HNF1A) cause maturity-onset diabetes of the young (MODY) (1). Young age at onset, autosomal dominant inheritance, and an excellent glycemic response to sulfonylurea monotherapy characterize HNF1A-MODY (2). A spectrum of heterozygous HNF1A variants have been identified—from classical MODY-causing mutations that impair insulin secretion (3,4), to milder alleles that increase the risk of type 2 diabetes. Homozygous loss-of-function germline mutations are thought to be embryonically lethal (5). Establishing the pathogenicity of missense variants is challenging (6,7): hypomorphic variants in the heterozygous state may be dismissed as benign, but when inherited recessively, they may mimic loss-of-function heterozygous mutations (8).

We describe an insulin-treated individual in whom a homozygous HNF1A missense variant, p.A251T (c.751G>A), was identified. We metabolically characterized both heterozygous carriers of the p.A251T variant and individuals homozygous for the p.A251T variant, performed in vitro functional studies to test our hypothesis that p.A251T is a hypomorphic HNF1A mutation, and assessed the transition to sulfonylurea therapy.

Research Design and Methods

In the proband, recruited to the MYDIABETES (MODY in Young-Onset Diabetes in Different Ethnicities) study (clinical trial no. NCT02082132, ClinicalTrials.gov), we sequenced HNF1A promoters, exons, and key flanking intronic sequences, along with 22 genes implicated in monogenic diabetes (9), and we assessed copy number variation using multiplex ligation-dependent probe amplification (9).

In Silico Modeling of Pathogenicity

We assessed predicted pathogenicity using established in silico tools, in line with the American College of Medical Genetics classification (10) and as previously described (11). For structural analysis, we used FoldX (12), introducing variants p.A251T and a nearby MODY-causing mutation, p.V246 L (13), into the structure of the HNF1A DNA binding domain complexed with DNA (Research Collaboratory for Structural Bioinformatics Protein Data Bank identifier 1IC8) (14).

Clinical Studies

We assessed clinical features and hs-CRP, a marker of HNF1A protein function (15), in the proband and his family. We also assessed in the proband β-cell function and glucose response to a mixed meal tolerance test (MMTT) (16), and his response to a 2.5-mg glibenclamide challenge. The proband underwent continuous glucose monitoring for 7-days before and after the switch to the sulfonylurea.

In Vitro Functional Characterization

The impact of p.A251T on HNF1A function was assessed by using a suite of in vitro assays, as described previously (11). This impact on function was compared with that of wild-type HNF1A and two DNA binding domain mutations that cause MODY.

Results

Clinical Characteristics of p.A251T Variant Carriers

Young-onset diabetes cosegregated with the homozygous, but not heterozygous, HNF1A p.A251T carriers; three homozygous individuals (proband, his identical twin, and a sister) presented with type 1 diabetes as teenagers and were treated with insulin (Supplementary Fig. 1 and Supplementary Table 1). The father fulfilled criteria for prediabetes and the mother developed gestational diabetes during her third pregnancy at age 29 years; her diabetes persisted postpartum.

In the proband, no additional mutations or variants were detected, and anti-GAD65 and IA2 antibodies were negative. Although consanguinity was not reported, both parents were from the same region in Eastern Europe.

Levels of hs-CRP were low (≤0.2–0.8 mg/L) in all those who were homozygous for p.A251T and in the heterozygous mother (see Supplementary Table 1). The heterozygous father displayed a higher value (1.4 mg/L).

During the MMTT (see Fig. 1), blood glucose plateaued at 12.5 mmol/L; this was accompanied by a rise in C-peptide. After the proband ingested glibenclamide, C-peptide increased further, beyond the levels measured during the MMTT. This increase was sustained and decreased only after glucose levels began to fall. Glucose dropped beyond fasting levels (to 2.9 mmol/L), however, causing hypoglycemia. Oral glucose was administered, resulting in another, higher C-peptide increase that restored normoglycemia. The proband was discharged with a prescription for 1.25-mg glibenclamide once daily (which was subsequently increased to twice daily); insulin was stopped. Continuous glucose monitoring performed before and after the treatment change demonstrated the resolution of hypoglycemia and maintenance of glucose levels (Supplementary Fig. 2), and HbA_1c was reduced to 6.9% (52 mmol/mol). The proband’s identical twin also transitioned from insulin to sulfonylurea therapy.

Figure 1.

Composite graph depicting results of the MMTT and the sulfonylurea challenge in the proband after morning premixed insulin was omitted. Glucose, C-peptide, and insulin during an MMTT, and the glibenclamide test dose, at indicated time points are displayed. Glucogel 25 g, an oral glucose preparation B (BI Healthcare, Bridgend, U.K.), was administered sublingually.

In Silico Characterization of the p.A251T Variant

The p.A251T variant is located in a region of HNF1A that encodes a highly conserved POU homeodomain DNA binding domain, which is crucial for transcription factor function (14). Neither p.A251T nor other rare variants at this position were present in the Genome Aggregation Database (examined July 2019). A variant of uncertain significance was predicted by in silico approaches (see Supplementary Table 2). Detailed structural analysis of p.A251T compared with that of a MODY-causing mutation (p.V246 L) at a similar location revealed the formation of a novel hydrogen bond (Supplementary Fig. 3).

In Vitro Functional Assessment

In vitro functional assays demonstrated a modest reduction in transactivation activity and DNA binding for p.A251T-HNF1A, which is less severe than bona fide mutations (p.P112L and p.R203H) that cause early-onset diabetes (Supplementary Fig. 4).

Conclusions

We report a homozygous hypomorphic HNF1A mutation in a family with young-onset diabetes. The clinical phenotype is consistent with classical heterozygous HNF1A-MODY: individuals have low hs-CRP and demonstrate sensitivity to sulfonylureas, enabling a transition off insulin. Importantly, young-onset diabetes cosegregated with homozygous individuals, not heterozygous carriers. In vitro studies support p.A251T as being a hypomorphic variant with modest transactivation defects only in the absence of endogenous HNF1A. This defect is likely driven by low DNA-binding potential. The defects observed, however, were less pronounced than those of reported bona fide heterozygous MODY-causing mutations in a similar location.

Our in silico studies provide a putative explanation for why p.A251T could have modest structural effects despite being located in the POU homeodomain. We showed that p.A251T effects are attenuated by the predicted formation of a novel hydrogen bond and that structural changes are less deleterious than those of a similarly located MODY-causing mutation. The hypomorphic status of this variant is further supported by the phenotype observed in heterozygous carriers, which is atypical for heterozygous loss-of-function HNF1A mutations.

The p.A251T HNF1A variant has been reported in two studies describing heterozygous carriers (17,18), bringing the total number of reported heterozygous carriers to eight. In the study by Thanabalasingham et al. (17), the proband, who presented with diabetes at the age of 43 years, was managed with metformin and sulfonylurea therapy for 25 years. A heterozygous relative of the proband developed diabetes later in life, at age 80 years. Although hs-CRP levels were low in the three carriers studied, the level of fucosylated glycan GP30 (low levels of which are predictive of HNF1A mutations) was above the cutoff in two of the three (18). In another study, a heterozygous proband received a diagnosis at age 39 years and had a BMI of 27 kg/m². The proband’s father, who was diagnosed at age 50 years, and a sister, diagnosed at age 53 years, had only impaired fasting glycemia when studied (11,18). Thus the age of heterozygous carriers at diagnosis has been markedly older (11) than that of individuals who are homozygous (12, 15, and 16 years). This is consistent with incomplete penetrance, and reported data suggest that a high BMI might be required to unmask the effect of the variant in the heterozygous state.

In our study, two homozygous individuals had hs-CRP levels ≤0.2 mg/L, whereas the heterozygous father had a level of 1.4 mg/L—within the range observed for individuals with type 2 diabetes and those without diabetes (15,19). This higher level contrasts data from previous reports in which hs-CRP levels were all <0.3 mg/L in the heterozygous state (17), but it could be accounted for by his age and BMI.

The most compelling evidence favoring pathogenicity of the biallelically inherited p.A251T variant is the successful transition from insulin to low-dose sulfonylurea therapy after >10 years of diabetes. It would not be expected that alternative diabetes types could be effectively managed with low-dose glibenclamide replicating the observed responses to a sulfonylurea in individuals with typical heterozygous HNF1A mutations (20).

Conclusion

We conclude that biallelic hypomorphic variants of HNF1A may present with a typical heterozygous MODY phenotype. Hypomorphic alleles causing MODY may be encountered more frequently as diverse populations, particularly those with autozygosity, are studied using whole-exome or whole-genome approaches.