Introduction

MODY is a genetically and clinically heterogeneous form of non-autoimmune diabetes that is characterised by a highly penetrant autosomal-dominant mode of inheritance, an early age of onset (usually before 25 years) and a primary dysfunction of the pancreatic beta cells [1, 2].

So far, 13 MODY genes have been identified [3]. Among these, BLK (MODY11) encodes a non-receptor tyrosine-kinase of the SRC family of proto-oncogenes, which is present in many tissues and cells including pancreatic beta cells [4]. Five rare mutations were shown to segregate with diabetes in three MODY families: four mutations are located in non-coding regions (at the end of the non-coding 3′ untranslated region or outside the gene) and only one non-synonymous mutation (p.A71T) was found in the fourth exon of BLK [4]. The p.A71T mutation is part of a haplotype including two of the non-coding mutations [4]. It was shown that BLK overexpression in the MIN6 pancreatic beta cell line enhanced insulin content and insulin secretion in response to glucose [4]. These actions were greatly attenuated by the BLK-p.A71T mutation [4]. Since this report, no study has attempted to replicate these data, in particular to confirm BLK firmly as a MODY susceptibility gene in other cohorts.

In the present study, we investigated the role of BLK in MODY in two European cohorts of unelucidated MODY patients (MODY-X). Furthermore, we assessed the contribution of the BLK-p.A71T mutation to type 2 diabetes risk and to the variation of related traits, as we found that this mutation is not rare in the general European population.

Methods

Patients selected for BLK sequencing

We studied 14 probands from MODY-X families recruited at the CNRS-UMR-8199 unit in Lille (France), who were diagnosed with diabetes before 36 years of age (10 of whom were diagnosed before 25 years of age). As previously published, these families are of French European origin, except one family from Mauritius [5]. We also studied 39 probands from Danish MODY-X families collected at the Steno Diabetes Center (Copenhagen, Denmark) (age at diagnosis 9–25 years) [6] and 11 additional patients from families with a vertical transmission of diabetes (at least one family member presented with diabetes diagnosed before 25 years of age). The clinical features of these patients are reported in Table 1 of the electronic supplementary material (ESM).

Sequencing protocol

BLK is located on human chromosome 8p23-p22 and encodes a 505-amino-acid protein (NM_001715.2; NP_001706.2 [NCBI36]). Genomic DNA was amplified by PCR with primers designed to cover the 13 exons and flanking intron–exon boundaries of BLK. A standard sequencing protocol was subsequently used [7].

Study cohorts

The case–control study included 4,901 French type 2 diabetic individuals and 4,280 French normoglycaemic controls (age at examination ≥45 years) (ESM Table 2). Type 2 diabetic individuals were recruited by the CNRS-UMR-8199 unit (n = 335), Corbeil-Essonnes Hospital (n = 2,182), Diab2-Néphrogène study (n = 1,997), DESIR (Data from an Epidemiological Study on the Insulin Resistance Syndrome) (n = 304) and the SUVIMAX (Supplementation en Vitamines et Mineraux Antioxydants) study (n = 83), as previously described [7]. Control individuals were recruited by the CNRS-UMR-8199 unit (n = 249), the DESIR (n = 3,118) and SUVIMAX (n = 913) cohorts [7].

All participants with type 2 diabetes or normoglycaemia were defined as such according to the 2003 American Diabetes Association criteria.

Non-diabetic participants in the DESIR (n = 4,760) and SUVIMAX (n = 1,758) cohorts were also analysed for metabolic quantitative traits (ESM Table 2).

All studies were approved by local ethics committees and were performed according to the principles of the Helsinki Declaration II. Written informed consent was obtained from all participants.

Genotyping of the p.A71T variant

Genotyping was performed using the high-resolution melting method on a LightCycler 480 PCR System (Roche Diagnostics, Meylan, France), as previously described [7]. Positive signals were confirmed by sequencing, with a concordance rate of 99% between genotyping and sequencing results. A genotype call rate of at least 96% was obtained in each cohort. No deviation from Hardy–Weinberg equilibrium (p > 0.05) was detected in any of the studied populations.

Statistical analyses

The association between the p.A71T variant and the risk of type 2 diabetes or the variation of related metabolic traits was assessed using logistic or linear regression models, respectively, which were adjusted for age, sex and BMI (when appropriate) under a dominant model. By applying a Bonferroni correction, a p value below 5.6 × 10−3 = 0.05/(6[metabolic traits] + 3[type 2 diabetes case–controls]) was considered significant, and between 0.05 and 5.6 × 10−3 was considered nominal (trend of association).

We assessed the power of our study using QUANTO software. The statistical power to detect an OR of 1.45 was 86% in the overall case–control analysis (all cases, n = 4,901) and 80% when the cases were stratified according to BMI <30 or ≥30 kg/m2. In the non-diabetic participants in the DESIR (n = 4,760) and SUVIMAX (n = 1,758) cohorts, the statistical power to detect an effect on BMI variation between 2 and 5 kg/m2 was higher than 99%. All statistical analyses were performed with SPSS (version 14.0,) and QUANTO (version 1.2.4, http://hydra.usc.edu/GxE/) software.

Homeostasis model assessment of pancreatic beta cell function (HOMA-B) and insulin resistance (HOMA-IR) were calculated as previously described [8].

Results

No rare non-synonymous BLK mutations were found in French and Danish MODY patients (n = 64); in particular, the c.211G>A/p.A71T mutation was not identified. Of note, the previously identified extragenic or non-coding mutations were not screened in these patients, because, first, two of these variants were reported to be unequivocally transmitted together with the missense p.A71T mutation; and, second, the effect of these mutations on pancreatic beta cell function was not investigated, contrary to the p.A71T mutation [4]. The 1000 Genomes Project and the NHLBI Exome Sequencing Project listed the BLK c.211G>A/p.A71T mutation as a low-frequency variant in Europeans (rs55758736; minor allele frequency 0.012–0.013). We genotyped this variant in the DESIR French general population (n = 5,064) and found 58 carriers of the A allele, including 52 normoglycaemic carriers (age at examination 31–65 years). We therefore concluded that the p.A71T variant was unlikely to cause MODY, which is a highly penetrant disorder.

The c.211G>A/p.A71T variant was further genotyped in French type 2 diabetic individuals (n = 4,901) and control samples (n = 4,280). We found a nominal effect of the A allele on increased type 2 diabetes risk (OR 1.47 [95% CI 1.03, 2.11], p = 0.035; Table 1). As most of the carriers of the BLK c.211G>A/p.A71T mutation in the MODY family reported by Borowiec et al were overweight or obese [4], we stratified the diabetic individuals according to BMI (BMI <30 kg/m2, n = 2,456; BMI ≥30 kg/m2, n = 2,445). We found an enrichment of the A allele in obese diabetic participants only (OR 2.44 [95% CI 1.32, 4.49], p = 4.29 × 10−3; Table 1), in contrast to non-obese diabetic participants (p = 0.145; Table 1).

Table 1 Effect of the BLK c.211G>A/p.A71T variant on type 2 diabetes risk
Full size table

Despite high statistical power, no nominal or significant association was found between the c.211G>A/p.A71T variant and BMI variation in 6,518 non-diabetic participants (p = 0.576; Table 2). Surprisingly, in the 4,760 non-diabetic participants from DESIR, the reported MODY A allele was associated with decreased HbA1c levels (β [SE] = −0.139[0.050]%, p = 5.27 × 10−3; Table 2) and an increase in beta cell function modelled by HOMA-B (β = 0.185[0.066], p = 5.37 × 10−3; Table 2), which does not fit with a MODY phenotype. No association was found for fasting glucose levels, fasting insulin levels or HOMA-IR (Table 2).

Table 2 Effect of the BLK c.211G>A/p.A71T variant on the variation of metabolic traits in non-diabetic participants in the DESIR and SUVIMAX studies
Full size table

Discussion

In this study, we identified numerous elderly normoglycaemic carriers of the BLK-p.A71T mutation, which does not fit with either rarity and high penetrance of a MODY gene mutation or the usual clinical features [3, 9, 10]. Furthermore, we did not find any other coding mutations in two European cohorts that included 64 MODY patients.

We observed a weak effect of the loss-of-function BLK-p.A71T mutation on increased type 2 diabetes risk, with an enrichment of the variant in obese diabetic cases, although no association with BMI was found. The BLK-p.A71T variant may therefore be ‘diabetogenic’ through obesity-related mechanisms, as previously described for some more frequent genetic variants such as the ENPP1-K121Q, PPARG-P12A or ADIPOQ polymorphisms [11–14]. Twelve out of 13 carriers of the p.A71T variant in the MODY family reported by Borowiec et al were overweight, of whom eight were obese [4]. The only carrier presenting with a normal BMI was non-diabetic at age 71 years [4]. This is in line with our present results and may reflect a modulation of the p.A71T variant effect by adiposity on type 2 diabetes risk.

Of note, the meta-analyses of European genome-wide association studies performed by the DIAbetes Genetics Replication and Meta-analysis (DIAGRAM) Consortium, the Meta-Analyses of Glucose and Insulin-related traits Consortium (MAGIC) or the Genetic Investigation of ANthropometric Traits (GIANT) Consortium, did not show any evidence of associations between variants at the BLK locus and risk of type 2 diabetes or variation of metabolic traits including BMI, fasting glucose, fasting insulin, HbA1c or HOMA-B (ESM Figure 1). Our present results on the contribution of the BLK-p.A71T variant to the risk of type 2 diabetes or the variation of HbA1c or HOMA-B are therefore likely to be nominal associations only.

In conclusion, we demonstrated that the loss-of-function BLK-p.A71T mutation is very unlikely to cause MODY. Instead, it may modestly influence type 2 diabetes risk through an interaction with obesity, although this will require further validation in additional studies.