Novel De Novo Mutation in Sulfonylurea Receptor 1 Presenting as Hyperinsulinism in Infancy Followed by Overt Diabetes in Early Adolescence

A female aged 10.5 years presented with fasting hyperglycemia (12.2 mmol/l), obesity (BMI 30.2 kg/m²), and aggressive behavior. She was the last born of nonconsanguineous Palestinian parents and has seven older, healthy siblings. Both parents developed type 2 diabetes during their 5th decade of life. Her birth weight was significantly large (4,950 g) despite a normal gestational oral glucose tolerance test. From the 2nd day of life until age 3 years, she had recurrent hypoglycemic convulsions (blood glucose 1.9–2.7 mmol/l) that were initially treated by intravenous glucose infusions and later by administration of sweet drinks at home. Hypoglycemia was not documented after age 3 years, but convulsions were never adequately controlled despite antiepileptic treatment (carbamazepine, hydantoin). Electroencephalogram at age 1 year revealed diffuse epileptic activity, while brain computed tomography scan was normal. Despite normal developmental milestones, her current school performance is poor, and she behaves aggressively and uses insulting words. Her menarche was at age 10 years. Hyperphagia and obesity have been a concern since early childhood.

Except for few pink striae on the abdomen, her physical exam was unremarkable for a fully pubertal adolescent. Laboratory investigations showed a blood glucose level of 12.2 mmol/l, moderate glucosuria, and undetectable anti-GAD (<1.0 μ/ml) and anti-insulin autoantibodies. Thyroid function tests and prolactin were normal, and a mildly elevated urinary free cortisol (299 nmol/24 h; normal <250 nmol/24 h) was easily suppressed by 90% to 37.5 nmol/24 h after 1 day of 0.5 mg dexamethasone every 6 h. Neurological evaluation, electroencephalogram, and brain magnetic resonance imaging were normal. Hydantoin replacement by topiramate achieved better epileptic control.

Before any antidiabetes treatment, a continuous glucose monitoring system (CGMS; Minimed, Sylmar, CA) was applied for 48 h, during which the patient had a low-carbohydrate diet. An intravenous glucose tolerance test (IVGTT) using 25% dextrose solution (0.5 g/kg) was performed following a 10-h fast. Glucose and insulin levels were determined at −5, 0, 1, 3, 5, and 10 min.

DNA was extracted from leukocytes of the patient, her parents, and one brother. For the patient, all coding exons and exon/intron boundaries of KCNJ11 and ABCC8, coding for Kir6.2 and SUR1 components of K_ATP channel, respectively, were PCR amplified and sequenced as previously described (13,14). Genetic markers were used to confirm parenthood, and the relevant DNA segment was sequenced in both parents. DNA from 1 brother and 65 unrelated Jerusalem Palestinians was analyzed by a specifically designed primer mismatch assay, which creates a new Sdu1 restriction site on the mutant allele.

The R370S point mutation of SUR1 was introduced into hamster SUR1 cDNA in the pECE plasmid using the QuikChange site-directed mutagenesis kit (Stratagene). The mutation was confirmed by DNA sequencing (15). Mutant clones from at least two independent PCRs were analyzed to avoid false results from undesired mutations introduced by PCR. Rat Kir6.2 cDNA is in the pCDNAI vector.

Cell surface expression level was determined by a chemiluminescence assay using SUR1 containing a flag epitope (DYKDDDDK) at the NH₂ terminal (fSUR), as previously described (16). Briefly, COSm6 cells grown on 35-mm dishes were transfected with 0.6 μg fSUR1 and 0.4 μg Kir6.2 cDNA per 3 μg Fugene6 at ∼70% confluency. At 48–72 h post-transfection, cells were fixed with 2% paraformaldehyde, incubated in M2 anti-FLAG antibody (10 μg/ml), then horseradish peroxidase–conjugated anti-mouse antibody (1:1,000 dilution; Jackson Immuno Research), followed by luminol. Luminescence signal was quantified using a Turner 20/20 luminometer.

COSm6 transiently expressing K_ATP channel subunits were loaded with ⁸⁶RbCl (1 μCi/ml) overnight. Before measurement of ⁸⁶Rb⁺ efflux, cells were incubated for 30 min at room temperature in Krebs-Ringer solution with metabolic inhibitors (2.5 μg/ml oligomycin and 1 mmol/l 2-deoxy-d-glucose). At selected time points, the solution was aspirated from the cells and replaced with fresh solution. At the end of a 40-min period, cells were lysed. The ⁸⁶Rb⁺ in the aspirated solution and the cell lysate was counted. The percentage efflux at each time point was calculated as the cumulative counts in the aspirated solution divided by the total counts from the solutions and the cell lysates.

Inside-out patch-clamp recording of COSm6 cells transiently expressing channel subunits was performed as previously described (4,16,17) using Axopatch 1D amplifier and Clampex 9.2 software. Recordings were made at −50 mV membrane potential in symmetrical K-INT solution containing (in mmol/l) 140 KCl, 10 K-HEPES, and 1 K-EGTA (pH 7.3 with KOH). For measuring MgADP, diazoxide, and glibenclamide responses, MgCl₂ was added for a free concentration of ∼1 mmol/l. Data were analyzed using pCLAMP software. Statistical analysis was performed with Origin 6.1 using an independent two-population two-tailed Student's t test.

Following the in vitro study results, oral hypoglycemic agents repaglinide (NovoNorm, 1 mg before meals for 1 day), and glibenclamide (5–10 mg b.i.d. for 4 consecutive days) were administered, without insulin. Fasting and postprandial glucose levels were recorded and compared with pretreatment values.

As CHI was evident in infancy, continuous blood glucose monitoring was performed to exclude occult hypoglycemic episodes. The CGMS revealed 48 h of continuous hyperglycemia (glucose 9.2–20 mmol/l) with no hypoglycemic events (Fig. 1A). Following a 12-h fast, while glucose was consistently high (12.2 mmol/l), insulin was 285 pmol/l, cortisol 338 nmol/l, growth hormone 81.4 pmol/l, and ammonia 44 μmol/l (reference range 10–60). β-OH-butyrate, acetoacetate, and free fatty acids were elevated at 232, 101, and 460 μmol/l, respectively. IVGTT revealed some incremental insulin secretion, up to 582 pmol/l, that failed to restore euglycemia (Fig. 1B) but suggested a possible role for insulin secretagogues as a treatment modality.

The increased birth weight, recurrent hypoglycemia, and consequent neurological damage suggested neonatal hyperinsulinism despite the current reduced insulin secretion. A novel heterozygous missense mutation was detected in exon 7 of ABCC8 (AGG→AGC at the 1,111 nucleotide position) (Fig. 2A), predicting an arginine to serine mutation at the 370th amino acid of the SUR1 protein (R370S). Exon 7 was sequenced in both parents’ DNA and was found to be entirely normal with no evidence of mosaicism. One sibling and 130 alleles from control subjects of the same ethnic background were also negative for the mutation (Fig. 2B). Parenthood was confirmed by genetic markers.

To confirm the functionality of our patient's mutation, several expression studies were performed. Cotransfection of Kir6.2 and mutant SUR1 (SUR1-R370S) in COSm6 cells demonstrated normal channel expression on the cell membrane (Fig. 3A). The effect of metabolic inhibition on different channel states was studied by ⁸⁶Rb⁺ efflux assays. Upon metabolic inhibition, wild-type (WT) channels were rapidly activated, giving rise to over 90% cumulative efflux at the end of the 40-min incubation period (Fig. 3B). On the other hand, the mutant channel only gave rise to <10% efflux. Inside-out patch clamp recordings revealed that the homozygote mutant channels completely lacked response to MgADP, in contrast to the stimulatory effect of MgADP on WT channels (Fig. 3C). When cells were made heterozygous by expressing both wild-type and mutant (WT:R370S) proteins to imitate our clinical setting, the metabolic inhibition–induced channel activity observed was greatly reduced and MgADP-induced current was reduced by 50% of the WT (Fig. 3D). These results corresponded with in vivo membrane depolarization in the fasting state, causing hyperinsulinemic hypoglycemia. Similar responses were seen when applying the K_ATP channel opener diazoxide, which could reverse the inhibitory effect of 0.1 mmol/l ATP on WT channels but failed to stimulate the homozygous mutated channels (Fig. 3E). In the heterozygous cells (WT:R370S), the channel was partially stimulated by diazoxide (Fig. 3F). The sulfonylurea glibenclamide, given at 10 and 100 nmol/l, blocked the WT channels but also had a similar blocking effect on the mutant channels, in homozygous and heterozygous states (Fig. 3G and H).

Based on some retained insulin response to IVGTT and the response of the mutated channel to glibenclamide, we studied our patient's response to insulin secretagogues. Repaglinide (1 mg), which closes K_ATP channels and binds to a site distinct from sulfonylureas (18), was given before meals; in two consecutive trials, it resulted in a decrease of 60-min postprandial glucose levels from 9.7–10.5 mmol/l to 5–8.3 mmol/l (Fig. 1C). Similarly, 5 mg glibenclamide administered twice daily also lowered postprandial glucose levels from 8.3–11 mmol/l to 7.2–7.8 mmol/l (Fig. 1D), and a further increment in the glibenclamide dose seemed even more effective.

Here, we report a patient with a de novo heterozygous mutation in the SUR1 gene (R370S) resulting in congenital hyperinsulinism followed by overt diabetes becoming evident before age 10.5 years. ABCC8 gene mutations have been known to cause autosomal recessive and dominant hyperinsulinism (7,11,19). Although the amino acid location of the R370S mutation could not predict its dominant effect, we demonstrated normal membrane expression, a finding common to previously reported dominant SUR1 gene mutations (16,19). As the K_ATP channel is a hetero-octamer containing four SUR1 subunits, a normally processed and translocated heterozygous mutant SUR1 subunit such as R370S would be expected to be present in 15 of 16 of the channels on the membrane and thus may alter their function. Conversely, mutated SUR1 proteins that do not reach the membrane will not result in abnormal channel activity in the heterozygous state because all channels reaching the membrane would contain four normal SUR1 subunits.

Activating SUR1 mutations were recently found to cause neonatal diabetes without antecedent hyperinsulinemic hypoglycemia (10). Huopio and coworkers (11,12) reported a dominant inactivating ABCC8 mutation, E1506K, that caused hyperinsulinemic hypoglycemia followed by hypoinsulinemic diabetes; however, when compared with our patient, nonpancreatectomized E1506K patients in whom overt diabetes presented later, during middle age, had mild neonatal hypoglycemia. Studies of predominantly Ashkenazi Jews with recessive ABCC8 mutations suggest that the natural history of the disease is one of clinical remission despite pancreas preservation. In some cases, the remission progressed to impairment of glucose tolerance (20).

While the molecular mechanisms causing ABCC8 mutation-mediated CHI and neonatal diabetes have been clarified, the molecular mechanisms resulting in the progression from hyper- to hypoinsulinism are not known. A slow, progressive loss of β-cell mass due to increased β-cell apoptosis may be responsible for both the remission in CHI and for the progression to diabetes (20–22). Interestingly, the heterozygous HNF4A mutation causing maturity-onset diabetes of the young (MODY1) was recently found to cause a phenotype similar to that described here, with transient neonatal hyperinsulinemic hypoglycemia followed by diabetes later in life (23), implying the possibility of similar underlying pathophysiology.

Our data in COS cells suggest equal biogenesis efficiency (data not shown) and surface expression of the WT and mutant R370S proteins at the plasma membrane. Thus, it is unlikely that the decrease in functional β-cell mass is related to apoptosis induced by misfolding of the mutant SUR1 protein, as has been suggested for neonatal diabetes–associated insulin gene mutations (24). Kassem et al. (21) have previously shown increased β-cell apoptosis in patients with recessive ABCC8 mutations. This apoptosis could be precipitated by increased metabolic stress due to continuous stimulation of insulin secretion, resulting in oxidative stress–mediated apoptosis. Alternatively, apoptosis could be initiated by chronically elevated cytoplasmic calcium levels, as have been seen in β-cells of patients with CHI (25). Studies of Kir6.2 and SUR1 knockout mouse models yielded variable phenotypes of glucose tolerance ranging from normoglycemia to diabetes, but this experimental model also appears to be associated with increased β-cell apoptosis and progression from hypoglycemia to diabetes, albeit at a different rate (26–29).

Regardless of the precise mechanism, enhanced apoptosis appears be a generic response to inactive K_ATP channels and not related to the specific mutation. Why then did our patient develop diabetes at such a young age (10.5 years)? It is likely that the obesity and the early puberty present in our case played an important role. Both factors contribute to insulin resistance that may have unmasked an insulin secretion defect caused by the ABCC8 mutation. Her constant intake of sweet drinks during childhood may have caused β-cell exhaustion. Furthermore, both parents developed type 2 diabetes during their 5th decade, suggesting the presence of a relatively heavy genetic burden for this disease. However, none of her seven older siblings developed adolescent or early adult diabetes, and adolescent-onset type 2 diabetes is rare in Palestinians despite a high prevalence of severe obesity, making it likely that the ABCC8 mutation contributed significantly to both the obesity and the diabetic phenotype.

Our in vitro findings had significant implications for clinical management. Initially, our patient was treated with insulin. However, the surprising response of the mutated R370S K_ATP channel to sulfonylureas in vitro suggested the use of these medications for glycemic control. The good response to glibenclamide enabled our patient to be treated with oral medications rather than insulin injections, a significant advantage given her poor compliance. In addition, the partial response of the WT:R370S heterozygous channel to diazoxide suggests that this treatment modality might have been effective in our patient, perhaps preventing hypoglycemia and improving the neurological outcome. In retrospect, adequate treatment with diazoxide might have also prevented the need for constant glucose loading and decreased the risk of obesity, perhaps preventing or delaying the onset of diabetes.

In conclusion, we describe a new de novo mutation in the ABCC8 gene in a patient with CHI in infancy followed by overt diabetes in childhood. This clinical phenotype gives another perspective to the crucial role of the SUR1 protein in β-cell function. The expression studies performed on the mutated SUR1 confirmed channel dysfunction and indicated a possible mechanism for dominant CHI. The clinical consideration of atypical diabetes at this young age and the in vitro studies had an important impact on patient management, suggesting the use of oral sulfonylurea treatment which resulted in improved glycemic control. The similarity between the phenotype described here and the phenotype related to HNF4A mutations (MODY1) emphasizes the complex interrelationship between hyperinsulinemic hypoglycemia and early-onset hypoinsulinemic diabetes. Our findings suggest a critical role for the ABCC8 R370S mutation in the early presentation of diabetes in this patient.