The Role of ONECUT1 Variants in Monogenic and Type 2 Diabetes Mellitus Free

Pancreatic transcription factors mediate the process of cellular differentiation from pluripotent stem cells through to mature pancreatic cells. Genetic variants that disrupt pancreatic transcription factors’ function result in a wide spectrum of pancreatic phenotypes, ranging from diabetes in the first 6 months of life (neonatal diabetes mellitus [NDM]) to increased risk for type 2 diabetes (1–3). Pathogenic variants in a subset of pancreatic transcription factors are common causes of maturity-onset diabetes of the young (MODY), with variants affecting the hepatocyte nuclear factor (HNF) genes HNF1A, HNF1B, and HNF4A accounting for more than 50% of cases of the disease (4).

The latest HNF gene to have been shown to have a role in monogenic diabetes is ONECUT1 (HNF6), with two unrelated cases of NDM caused by biallelic ONECUT1 pathogenic variants reported (5). These individuals, both of whom had pancreatic hypoplasia, were homozygous for different coding variants in the ONECUT1 gene: one missense (p.Glu231Asp), the other stop-gain (p.Glu231*). The authors also identified a significant enrichment of heterozygous rare ONECUT1 variants (minor allele frequency [MAF] <0.005) in a cohort of 2,165 individuals with type 2 diabetes when compared with healthy controls, suggesting a role of ONECUT1 in adult-onset diabetes. Furthermore, the relatively young age profile and low BMI of ONECUT1 variant carriers in the type 2 diabetes cohort was consistent with a role of certain pathogenic ONECUT1 variants in autosomal dominant monogenic diabetes (5).

ONECUT1 has long been considered a candidate gene for monogenic diabetes, as it plays a vital role in pancreatic endoderm and pancreatic progenitor cell differentiation (6). Through co-operative binding with other pancreatic transcription factors, including PDX1, GATA6, and FOXA2, it regulates gene expression during the early stages of pancreatic development and facilitates the maturation of precursor cells into functional adult pancreatic cell types. In addition to this early developmental role, ONECUT1 is also important for development and maintenance of β-cell identity (6).

In this study, we build upon recent findings by further exploring the role of ONECUT1 variants in monogenic and type 2 diabetes using a large cohort of individuals with suspected monogenic β-cell disorders and >400,000 individuals from UK Biobank (7).

We studied an international cohort of 697 subjects referred to the Exeter Genomics Laboratory for monogenic diabetes testing, in whom all the known genetic causes of monogenic diabetes had been excluded as previously described (8). Two hundred thirteen unrelated individuals were diagnosed with NDM (defined as diabetes with onset at <6 months old), and 484 unrelated individuals were clinically suspected of having MODY.

Whole-genome sequencing was performed on 333 individuals (172 NDM, 161 MODY) as previously described (9), with the remaining 323 individuals suspected of MODY and 41 individuals with NDM tested on our targeted next-generation sequencing panel (10), which had been updated to include baits for ONECUT1. The study was conducted in accordance with the Declaration of Helsinki, and all subjects or their parents gave informed consent for genetic testing with ethical approval received from the Genetic Βeta-cell Research Bank, Exeter, U.K.

We interrogated the sequencing data of the 231 individuals with NDM without a pathogenic variant in a known gene to identify homozygous or compound heterozygous ONECUT1 pathogenic variants. Only biallelic coding or splice site variants (±2 base pairs of intron/exon boundary) with an MAF <0.0001 in gnomAD v2.1.1 (11) affecting the canonical transcript (NM_004498) were considered as potentially pathogenic.

We first analyzed the sequencing data of 484 individuals clinically suspected of MODY without a genetic diagnosis to detect any null (here defined as frameshift, stop-gain, start-loss, canonical splice site variants, and single-exon/multiexon deletions) ONECUT1 variant affecting the canonical transcript. We performed Fisher’s exact tests to investigate enrichment of ONECUT1 variants in this cohort when compared with 16,465 individuals included in the gnomAD v3.1.2 control population (11). For the analysis to be comparable to that conducted by Philippi et al. (5), which detected an enrichment of missense variants in the young-onset type 2 diabetes cohort, the same gnomAD v2.1.1 MAF cutoff (0.005) was used to filter variants in the case and control cohorts. We investigated synonymous variant enrichment in both case and control cohorts, which acted as a negative control to ensure that any differences in the sequencing technologies used did not influence variant frequency.

We performed burden testing for association between type 2 diabetes and ONECUT1 variants using the UK Biobank exomes data set (7). Separately, we tested the European (n = 419,850, of which 32,899 had type 2 diabetes), African (n = 7,393, of which 1,356 had type 2 diabetes), and South Asian (n = 9,539, of which 2,431 had type 2 diabetes) UK Biobank cohorts. The definition of type 2 diabetes used here was a report of non-insulin-dependent diabetes from any source or a HbA_1c level ≥48 mmol/mol. We excluded individuals receiving insulin or with an additional conflicting diagnosis of type 1 diabetes.

We created three subsets of ONECUT1 variants: null variants (as previously defined in this study), in silico–predicted deleterious missense variants (Combined Annotation Dependent Depletion [CADD] phred score ≥25) and nondeleterious missense variants (CADD phred score <25). Only variants with an MAF <0.001 that affect the canonical ONECUT1 transcript were included in the subsets.

The variant subsets were tested for association with type 2 diabetes using the UK Biobank analysis pipeline developed by Regeneron (12). Because of the lack of ONECUT1 null variants in the African and South Asian cohorts, the null subset could only be tested in the European cohort. The two missense variant subsets were tested in all three cohorts. Age, sex, testing center, exome batch, and principal components 1–40 were all included as control variables. When testing for association with binary traits including type 2 diabetes, the pipeline performs a Firth-corrected logistic regression statistical test.

The monogenic diabetes patient data used in this study are available from the corresponding author upon reasonable request. The type 2 diabetes data used in this study are available via UK Biobank (https://www.ukbiobank.ac.uk/), subject to necessary approvals. For the purpose of open access, the author has applied a CC BY public copyright license to any author accepted manuscript version arising from this submission.

Within our NDM cohort of 231 individuals, we identified one novel biallelic variant in the ONECUT1 gene. The homozygous frameshift variant (p.Met289Argfs*80, Chr15(GRCh37):g.53081217_53081218dup) was detected in a single proband referred from India. The p.Met289Argfs*80 variant is predicted to result in the translation of a truncated ONECUT1 protein lacking the CUT and HOX domains (Fig. 1).

The female proband presented with intrauterine growth restriction (birth weight was 1,590 g at 40 weeks gestation; −6.42 SD) and was diagnosed with diabetes at 3 days of age. She was treated with insulin (0.15–0.2 units/kg/day) from diagnosis but displayed postnatal growth failure and died at 2 months of age. Exocrine pancreatic function was not tested. There was no known family history of diabetes. In addition to diabetes, the proband had multiple extrapancreatic features including facial dysmorphism, skeletal dysplasia, camptodactyly, epilepsy, hypotonia, and anemia.

ONECUT1 missense variants were not significantly enriched in our cohort of 484 individuals with suspected MODY when compared with population controls (Fisher exact P = 0.644). There were 9 individuals in the Exeter MODY cohort with coding variants in ONECUT1 (all heterozygous missense, listed in Supplementary Table 1), compared with 384 individuals in the gnomAD control cohort (381 heterozygous, 3 homozygous). There were no null variants detected among the suspected MODY cases (10 individuals with heterozygous null variants were present in the controls). No significant difference in the frequency of synonymous variants was found between the MODY and gnomAD cohorts (Fisher exact P = 0.822), confirming that the cohorts are reliably comparable and not significantly affected by technical differences in sequencing platforms.

We identified 12 individuals heterozygous for ONECUT1 null variants in UK Biobank (Fig. 2). Four of these cases (33%) had type 2 diabetes, with a median age of diagnosis of 58 years (interquartile range = 57.5–63.5), and BMI of 30.4 (interquartile range = 29.1–33.3). A significant association was found between null variants and type 2 diabetes in the UK Biobank European cohort (odds ratio [OR] = 27.90, 95% CI = 2.56–304.35, P = 0.00633). The South Asian and African cohorts were not tested in the null variant analysis because of the lack of null variants in these UK Biobank subsets.

No association with type 2 diabetes was found for either CADD-predicted deleterious missense variants or CADD-predicted nondeleterious missense variants in the European, African, or South Asian cohorts (Table 1). There were nine individuals (three in the European cohort and six in the African cohort) homozygous for ONECUT1 missense variants in the UK Biobank (1/9 had a CADD-predicted deleterious variant); none of these individuals had a clinical diagnosis of diabetes or an HBA_1c ≥48 mmol/L.

In this study, we explored the role of ONECUT1 variants in the etiology of monogenic and type 2 diabetes. Our assessment of 697 individuals referred for NDM or MODY testing identified a third case of neonatal diabetes caused by a homozygous ONECUT1 variant. Furthermore, using UK Biobank, we detected an association between heterozygous null ONECUT1 variants and increased risk of adult-onset diabetes.

With the report of the third case of NDM caused by a homozygous ONECUT1 variant, the requirement of three unrelated cases to confirm a gene-disease association, as is required by Genomics England for a gene to be added to their disease panels, has now been met (13). As a result, we recommend that ONECUT1 be added to gene panels for NDM.

The identification of a third NDM case provides additional details on the phenotype associated with this rare NDM subtype. The previously described individual (5) homozygous for the ONECUT1 missense variant had a relatively milder phenotype (diabetes onset after 6 months of age and no extrapancreatic features) when compared with both the previously described individual with the homozygous stop-gain variant and the individual in our cohort with a homozygous frameshift variant. This could possibly indicate that some function is preserved with the missense variant, while there is complete loss with homozygous null variants. Both individuals with homozygous null variants had extremely low birthweights (both <−5 SD), suggesting reduced insulin secretion in utero, similar to what has been reported in other forms of NDM characterized by absence of fetal insulin (14). The birth weight of the individual with the homozygous missense variant was less extreme at −1.4 SD. Both individuals with homozygous null variants had additional morphological and neurological features, while the phenotype of the individual with the homozygous missense variant was limited to pancreatic hypoplasia and NDM. The presence of extrapancreatic features is consistent with the phenotype of mouse knockout models, which display impaired development of liver, gallbladder, and pancreas (Table 2) (15–19). The possibility of some retained function in the missense variant protein is supported by the fact that, unlike the two null variants, the missense variant does not directly disrupt the known functional domains of the ONECUT1 protein. These observations are suggestive of a possible genotype-phenotype relationship in ONECUT1-NDM, with null variants causing a multisystem syndromic disease while variants which do not result in complete loss of protein function cause a pancreatic-only phenotype. The identification of additional individuals with biallelic pathogenic ONECUT1 variants is needed to further define the clinical features of this rare NDM subtype.

We show that null variants in ONECUT1 are likely to be a risk factor contributing to the development of type 2 diabetes in adulthood. The OR of the observed UK Biobank association is high compared with other genes where rare variants have been associated with type 2 diabetes, such as GIGYF1, in which null variants were associated with type 2 diabetes with an OR of 4.5 (CI = 2.71–6.37) (20). The magnitude of the association is, in fact, similar to that observed for heterozygous null RFX6 variants (21), which are associated with MODY with reduced penetrance. However, the small number of variants (n = 9) and wide CI of the association between ONECUT1 null variants and type 2 diabetes mean that caution must be taken in the interpretation of the result.

Screening in our large cohort of individuals with clinically suspected MODY did not reveal an enrichment of ONECUT1 missense variants, and no individuals within the cohort had a null variant in the gene. Similarly, Prudente et al. (22) did not find any predicted pathogenic ONECUT1 variants in their cohort of familial diabetes cases. Taken together, this suggests that either monoallelic variants in ONECUT1 are not a cause of monogenic diabetes or they represent an extremely rare subtype of the disease. Additional studies of larger MODY cohorts are needed to determine the role of ONECUT1 variants in monogenic adult-onset diabetes.

In summary, our study provides further evidence showing the role of biallelic ONECUT1 variants in NDM. Furthermore, we identified a possible role for heterozygous ONECUT1 null variants as a risk factor for adult-onset type 2 diabetes. Our results further define the role of ONECUT1 variants in both neonatal and adult-onset diabetes, confirming the important role of this transcription factor in β-cell development and function.

Acknowledgments. This research has been conducted using the UK Biobank Resource under application numbers 9072, 9055, and 49847. The authors are grateful to the family of the patient with NDM who participated in this study.

Funding. This work is funded by Diabetes UK (19/0005994 and 21/0006335), Medical Research Council (MR/T00200X/1), Wellcome Trust (224600/Z/21/Z), and Wellcome Trust Institutional Strategic Support Fund awarded to University of Exeter. E.D.F. is a Diabetes UK RD Lawrence Fellow (19/005971). K.A.P. is funded by the Wellcome Trust (219606/Z/19/Z). A.T.H. is employed as a core member of staff within the National Institute for Health Research (NIHR)–funded NIHR Exeter Clinical Research Facility and is an NIHR Emeritus Senior Investigator. S.E.F. has a Wellcome Trust Senior Research Fellowship (223187/Z/21/Z). T.W.L. has a lectureship funded by a Research England Expanding Excellence in England award. M.B.J. and M.N.We. are recipients of an Exeter Diabetes Centre of Excellence Independent Fellowship funded by the Research England Expanding Excellence in England fund. M.B.J. has been the recipient of a European Association for the Study of Diabetes Rising Star fellowship during this study. The work is supported by the NIHR Exeter Biomedical Research Centre, Exeter, U.K.

The Wellcome Trust, Medical Research Council, and NIHR had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The views expressed are those of the author(s) and not necessarily those of the Wellcome Trust, Department of Health, National Health Service, or NIHR.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.R.-S., S.E.F., M.N.We., and E.D.F. participated in study conception and design. K.A.P., A.T.H., and E.D.F. recruited patients and performed clinical phenotyping. P.P.P. provided clinical information. J.R.-S., K.A.P., T.W.L., M.B.J., M.N.Wa., and E.D.F. analyzed the MODY and NDM cohort sequencing data. J.R.-S., G.H., and M.N.We. analyzed the UK Biobank exome sequencing data. J.R.-S. carried out statistical analysis. J.R.-S. wrote the first draft of the manuscript. K.A.P., T.W.L., G.H., M.B.J., M.N.Wa., A.T.H., S.E.F., M.N.We., and E.D.F. participated in manuscript improvement. All authors reviewed the manuscript. E.D.F. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.