Co-occurrence of Loss-of-Function GCK and ABCC8 Variants in a Pedigree With a Spectrum of Dysglycemia

Abstract

Glucokinase and ATP-sensitive potassium (K_ATP) channels in pancreatic β-cells control insulin secretion in response to glucose stimulation to maintain glucose homeostasis. It is well established that loss-of-function (LOF) variants in GCK, which encodes glucokinase, are diabetogenic, whereas LOF variants in K_ATP channel genes lead to congenital hyperinsulinism (HI) and hypoglycemia. However, how the co-occurrence of GCK and K_ATP channel variants manifests in glycemic phenotypes is unknown. This study presents a multiplex pedigree with a heterozygous GCK deletion along with a heterozygous ABCC8 variant that results in the E1209K missense variant in the regulatory subunit of the K_ATP channel sulfonylurea receptor 1 (SUR1). The three-generation pedigree exhibits a complex spectrum of dysglycemia depending on genotype and age at referral. Individuals harboring the heterozygous GCK deletion alone present with maturity-onset diabetes; those harboring the heterozygous ABCC8 E1209K variant alone exhibit HI resulting from the LOF of K_ATP channels, with some developing diabetes later in life; and those harboring both GCK and ABCC8 variants escape HI but not diabetes. This unique pedigree offers insights into the complex interplay between LOF genetic variants of glucokinase and K_ATP channels in age-dependent dysglycemia.

Article Highlights

• Genetic variants causing loss of function (LOF) of glucokinase or ATP-sensitive potassium (K_ATP) channels underlie diabetes or congenital hyperinsulinism, respectively, but how the co-occurrence of such variants affects glucose control is unknown.

• This study presents genotypes and clinical phenotypes in a pedigree with LOF variants in both GCK and the pancreatic K_ATP channel.

• Heterozygous glucokinase deletion masked infantile hyperinsulinemia and hypoglycemia caused by a heterozygous LOF K_ATP channel variant.

• Carriers of the LOF K_ATP variant exhibited a shift from hypoglycemia to diabetes, as has been reported previously in some carriers of K_ATP LOF variants.

Introduction

Glucose-stimulated insulin secretion from pancreatic β-cells is essential for glucose homeostasis. Glucokinase and ATP-sensitive potassium (K_ATP) channels are two key signaling molecules in glucose-stimulated insulin secretion. Glucokinase acts as a glucose sensor by catalyzing glucose phosphorylation, the first rate-limiting step of glycolysis (1). K_ATP channels act as gatekeepers of insulin secretion by coupling glucose metabolism with β-cell electrical activity (2). Glucose catabolism on a rise in blood glucose elevates the cytosolic ATP-to-ADP ratio, leading to K_ATP channel closure, β-cell membrane depolarization, and insulin secretion.

Loss-of-function (LOF) variants of GCK encoding glucokinase cause monogenic diabetes subtype GCK (formerly known as maturity-onset diabetes of the young [MODY] 2), an autosomal dominant form of diabetes characterized by mild, early-onset hyperglycemia and defective β-cell insulin secretion (3,4). In contrast, LOF variants in ABCC8 and KCNJ11 encoding the sulfonylurea receptor 1 (SUR1) and Kir6.2 subunits of the K_ATP channel, respectively, cause congenital hyperinsulinism (HI), an infantile or childhood disease characterized by excessive insulin secretion despite profound hypoglycemia, which can be autosomal dominant or recessive depending on the variant (5,6). Paradoxically, HI can evolve into diabetes later in life in some patients, particularly when additional diabetes risk factors are present (7–11). Adding more complexity, gain-of-function (GOF) K_ATP variants are linked to neonatal diabetes, MODY, or increased risk of type 2 diabetes (12–14).

Although LOF variants in GCK or K_ATP channel genes alone are known to cause abnormal insulin secretion and glucose dysregulation, how the co-occurrence of variants in both genes manifests clinically is unknown. We report a three-generation pedigree with LOF variants in GCK and ABCC8 presenting diverse glycemic phenotypes, ranging from overt diabetes to mild hyperglycemia to hypoglycemia. Genotype–phenotype correlation examined family members harboring either of the heterozygous variants alone and those harboring both. The study reveals an intricate interplay between GCK and ABCC8 variants in the longitudinal glycemia phenotypes.

Research Design and Methods

Genetic Studies

A three-generation family from Morocco was referred to the Paris Department of Genetics for genetic screening for young-onset diabetes. Informed written consent to genetic testing was obtained for all patients.

Two approaches were used for the proband II-2. The first was targeted next-generation sequencing based on a multiplex PCR assay (MODY-MASTR Assay; Agilent Technologies) analyzing GCK, HNF1A, HNF4A, HNF1B, ABCC8, KCNJ11, and INS. Multiplex libraries were run on MiSeq (Illumina). SEQNEXT software (version 4.2.2; JSI Medical Systems) was used for alignments, variant calling, and annotations. Second, a search for large genomic deletions was performed in GCK, HNF1A, HNF4A, and HNF1B using a multiplex ligation-dependent probe assay (SALSA MLPA P241 MODY; MRC Holland). Variants were classified following the Monogenic Diabetes Expert Panel (ClinVar) rules, adapted from the American College of Medical Genetics and Genomics guidelines (15). Segregation analysis was subsequently conducted by Sanger sequencing for the ABCC8 (NM_001287174.3) variant and by MLPA (see above) for the GCK (NM_000162.5) deletion.

Functional Studies

COSm6 cells transfected with human wild-type (WT) or ABCC8 E1209K cDNA (in pCMV6b) and human WT KCNJ11 (in pcDNA3.1) using FuGENE 6 were used for functional studies 48 to 72 h later. Untransfected cells served as negative controls. Western blots were used to assess SUR1 maturation and trafficking, and Rb⁺ efflux and electrophysiology experiments were used to assess K_ATP channel activity in intact cells and isolated membrane patches, respectively, as described previously (16). Statistical analysis was performed in GraphPad Prism using two-tailed unpaired or paired t tests, with statistical significance set at P < 0.05.

Data and Resource Availability

All data generated or analyzed during this study are included in the published article and its online Supplementary Material.

Results

The study examines a large family with affected members over three generations (Fig. 1). A female in the first generation, heterozygous for the ABCC8 c.3625G>A p.(E1209K) (referred to as ABCC8 E1209K hereinafter) variant, was initially diagnosed with gestational diabetes at age 20 years while lean. She started insulin treatment at age 24 years and later switched to oral hypoglycemic agents (OHAs) at age 55 years at a BMI of 29 kg/m². She had no microvascular complications and developed neuropathy at age 55 years. Her husband, who carried a heterozygous GCK deletion of exons 2 to 9, had diabetes after age 40 years and was overweight. Initially treated with OHAs, he was later switched to insulin at age 70 years.

Figure 1.

Genotype/phenotype correlation depicting the range of phenotypic presentations in the pedigree and underlying molecular backgrounds. GCK⁺ indicates patients with heterozygous deletion of GCK exons 2 through 9; ABCC8⁺ indicates patients were heterozygous for ABCC8 c.3625G>A p.(E1209K) variant. Symbols are black for patients with hyperglycemia and red for hypoglycemia. GCK⁻ and ABCC8⁻ genotypes indicate patients who do not carry these variants. Proband is indicated by the arrow. Birth weight in percentile is corrected for gestational age. Age is indicated in years unless specified otherwise. BMI is given in kg/m²; mo, months old; NA, not available; Metf, metformin; Repag, repaglinide.

Their offspring inherited varied combinations of their molecular events (Fig. 1 and Supplementary Fig. 1). Two daughters (II-2 and II-6) inherited both the GCK deletion and the ABCC8 variant, whereas one son (II-4) inherited only the latter. No genetic information was available for II-3, who died in utero, or for II-5, who was born at 38 weeks weighing 3,200 g (56.7th percentile) and was not known to have diabetes. II-2 was born at 37 weeks weighing 2,700 g (25.6th percentile) and spent time in the neonatal intensive care unit (no details available) when she was 2 months old. She later presented with diabetes at age 12 years, which was managed with diet alone. She received insulin treatment during pregnancy with both her daughters. II-6, born at 40 weeks weighing 3,700 g (62.5th percentile), was diagnosed with diabetes at age 8 years, when her BMI was at 25.5 kg/m², which is considered overweight; she was initially not treated, but OHA and insulin treatment was initiated when she was 12 years old and her BMI had reached 38 kg/m². Their brother (II-4) had a high birth weight (>97.5th percentile), indicative of hyperinsulinemia, and experienced hypoglycemia for a few days after birth, which was treated with glucagon. He later developed diabetes at age 15 years, when his BMI was 23.5 kg/m², and started OHA and insulin treatment. At last evaluation, he was age 32 years, had a BMI of 29.9 kg/m², had severe dysglycemia (HbA_1c level had reached 14% or 130 mmol/mol before therapy adjustment), had experienced numerous diabetes complications, and was receiving dialysis.

In the third generation, the first daughter (III-1), carrying her mother’s ABCC8 variant only, was born at 38 weeks weighing 3,800 g (96.4th percentile); she was diagnosed with HI at birth and treated with diazoxide. At age 3 years, she still needed diazoxide therapy. The second daughter (III-2), born at 39 weeks weighing 3,000 g (19.3th percentile), inherited both the GCK deletion and the ABCC8 variant and remained normoglycemic at the time of this report (aged 2 years). The third daughter (III-3), born at 38 weeks weighing 2,905 g (25.9th percentile), inherited only the GCK deletion and was hyperglycemic in her first month of life.

Variants of GCK and K_ATP genes are known to have an impact on birth weight (17). The World Health Organization fetal growth calculator confirmed that babies in this family with the ABCC8 variant alone were large for gestational age (>96th percentile for both). Neonates carrying the GCK variant in addition to ABCC8 c.3625G>A had lower birth weights, varying between the 19.3th and 62.5th percentiles. It should be noted that the birth weight of III-3, carrying only the GCK variant, was in the 25.9th percentile because her own mother carried the GCK variant herself, thus rebalancing the in utero insulin secretion (18).

Whereas GCK deletions have already been studied, and haploinsufficiency is a known mechanism for GCK-MODY (19), the pathogenicity of the ABCC8 missense variant is not self-evident. ABCC8 E1209K was previously found to cosegregate with autosomal dominant hypoglycemia in a three-generation family, but no functional characterization studies have been performed (5). In this study, the two patients carrying the ABCC8 c.3625G>A variant alone for whom neonatal charts were available (II-4 and III-1) were diagnosed with HI at birth, whereas the two older patients (I-2 and II-4) developed early-onset diabetes at ages 20 and 15 years, respectively, with normal BMIs at diagnosis. Because both GOF and LOF K_ATP variants can result in early-onset diabetes (20), we investigated the functional impact of the ABCC8 E1209K variant in order to understand the molecular etiology of the varied glycemic phenotypes in this family.

COSm6 cells transiently transfected with WT or ABCC8 E1209K–mutated channel plasmids were used for biochemical and functional studies to assess the impact of ABCC8 E1209K (Fig. 2A). We first determined the effect of the variant on SUR1 protein maturation and trafficking. Western blots showed no difference in the intensity of mature SUR1 between WT and the E1209K variant (Fig. 2B), indicating the variant does not impede channel biogenesis or trafficking to the plasma membrane. Next, we performed inside-out patch clamp recordings to evaluate how the E1209K variant affects K_ATP channel gating. Compared with WT channels, the E1209K channels exhibited a markedly reduced response to the physiological activator MgADP as well as the pharmacological K_ATP opener diazoxide (Fig. 2C and D). Lastly, Rb⁺ efflux assays, in which Rb⁺ is used as a surrogate ion for K⁺ to report channel activity in intact cells (21), showed reduced efflux in E1209K-expressing cells compared with WT-expressing cells, both in response to metabolic inhibition, which mimics hypoglycemia, and in response to diazoxide, which is a frontline treatment for HI. Together, the functional data show the ABCC8 E1209K variant reduces the function of K_ATP channels by causing defective MgADP response, without affecting channel trafficking, and supports ABCC8 E1209K as a dominant pathogenic variant of HI. Moreover, the successful treatment of HI with diazoxide in patient III-1 (Fig. 1) indicates the residual diazoxide response in the E1209K channels is sufficient to reach the therapeutic threshold (22).

Figure 2.

Effects of ABCC8 E1209K on the expression and function of K_ATP channels. A: ABCC8 E1209K variant site (highlighted in red) shown in the cryogenic electron microscopy structure of the K_ATP channel bound to MgADP, MgATP, and K_ATP opener NN414 (Protein Data Bank ID 7W4O). Only one SUR1 subunit (light green) is shown for clarity. B: Western blot of WT or E1209K SUR1 in COSm6 cells cotransfected with cDNA for SUR1 and Kir6.2 (RESEARCH DESIGN AND METHODS). Untransfected (UT) cells served as negative controls. Gray circle marks core glycosylated (immature) SUR1, and clear circle marks complex glycosylated (mature) SUR1, with molecular mass markers (kDa) shown to the right of the blots. Quantification of the upper SUR1 band to total SUR1 (upper plus lower bands) using ImageJ showed no difference between mutated and WT channels (E1209K = 99% ± 5% of WT; n = 3). C: Representative inside-out patch clamp current recordings of WT and ABCC8 E1209K channels at −50 mV. Arrows mark patch excision; upward deflections indicate inward currents; black dashed lines indicate zero baseline in the presence of 1 mmol/L MgATP. Exposure of channels to 0.5 mmol/L MgADP or 0.2 mmol/L diazoxide is denoted by bars above the recordings. D: Bar graph showing quantification of currents in MgADP and diazoxide relative to currents in K-INT at patch excision (normalized currents). Clear circles represent data points from different cells. E: Fractional Rb⁺ efflux from cells expressing WT or ABCC8 E1209K channels during 30-min incubation with metabolic inhibitors (MI; 1 mmol/L deoxyglucose and 5 µg/mL oligomycin) or 0.2 mmol/L diazoxide. Paired t test was used to account for differences in transfection efficiency in different biological repeats (clear circles). NBD, nucleotide binding domain. *P < 0.05 by paired t test (E), ***P < 0.001 by unpaired t test (D), ****P < 0.0001 by unpaired (D) or paired (E) t test.

Discussion

The multigeneration family presented here provides a rare opportunity to study how co-occurrence of LOF GCK and K_ATP gene variants affects glycemic control. To our knowledge, this is the first pedigree with such a genotype reported. The genetic and functional findings provide clues to the molecular etiology of dysglycemia in the family and suggest an interplay between GCK and K_ATP channel gene variants in clinical manifestations, with implications for diagnosis, treatment, and genetic counseling with regard to dysglycemia related to variants in GCK and the K_ATP channel genes.

Diabetes later in life in patients with LOF K_ATP variants and HI was first reported in a Finnish family by Huopio et al. (8). More recently, Scala et al. (10) reported similar findings in multiple patients with LOF K_ATP variants and proposed that LOF K_ATP variant–related diabetes be classified as a distinct diabetes type. In our study, diabetes was observed in two carriers of the ABCC8 E1209K variant without the GCK deletion at the ages of 20 and 15 years with elevated BMIs. Potentially, impaired insulin secretion response to glucose stimulation resulting from the LOF K_ATP variant could underlie the diabetes observed in these individuals, which could have been further exacerbated by additional risk factors, such as obesity. We note that the North African family studied here may have a type 2 diabetes polygenic background, as suggested by the metabolic profiles of several individuals. Although the precise mechanisms of diabetes in I-2 and II-4 remain to be determined, a third patient in the third generation (III-1) with the same genotype who was diagnosed with HI and is now 4 years old may also develop diabetes later on. Reducing risk factors that will accelerate diabetes development, such as weight management, might help in prevention and management of diabetes.

Contrary to the HI phenotype observed in individuals carrying only the ABCC8 E1209K variant, no neonatal hypoglycemia was reported in any of the three members carrying both the ABCC8 E1209K variant and the GCK deletion. Of the three, two had early-onset diabetes, whereas the third, currently aged 2 years, showed no sign of dysglycemia, suggesting that GCK LOF averts the HI-inducing effects of the LOF ABCC8 variant. This may be due to a direct effect of GCK inactivation that increases the glucose threshold for insulin secretion. Of note, there is evidence that GCK inactivation may preserve β-cell mass and function by reducing glycolytic flux (23,24). Thus, GCK deletion may also paradoxically help halt diabetes development because of LOF K_ATP. Conversely, the absence of hyperglycemia in 2-year-old III-2 raises the possibility that K_ATP LOF might also reduce the hyperglycemia caused by GCK deletion in the first years of life, because this type of variant can lead to a remission period; however, more data are needed to reach such a conclusion. Intriguingly, prior animal studies have suggested a link between GCK activity and K_ATP channel function. Knockout of the pore-forming subunit of the K_ATP channels (Kir6.2^−/−) in mice mitigated the hyperglycemia associated with GCK deletion (25). Our limited study here raises an interesting possibility that such a phenotype-modifying relationship between the two gene variants may similarly exist in humans. Future studies with more patients are needed to test this hypothesis.

This article contains supplementary material online at https://doi.org/10.2337/figshare.31860736.

Funding Statement

This work was supported by National Institutes of Health (NIH) grant R01DK066485 (S.-L.S.), Egyptian government predoctoral scholarship GM 1109, and NIH grant T32CA285265 (A.E.).