Kir6.2 Mutations Are a Common Cause of Permanent Neonatal Diabetes in a Large Cohort of French Patients

Neonatal diabetes, defined as insulin-requiring hyperglycemia within the first months of life, is a rare entity, with an estimated incidence of 1 in 400,000 neonates (1). In about one-half of the neonates, diabetes is transient (transient neonatal diabetes [TND]), with remission within the first 6 months of life, whereas the rest of the patients have a permanent form of diabetes. In many cases of TND, abnormalities of chromosome 6 have raised the possibility that an imprinted, paternally expressed gene is involved in its pathogenesis (1,2).

By contrast, very few cases of permanent neonatal diabetes (PND) have been elucidated at the molecular level. Complete deficiency of the glycolytic enzyme glucokinase (GCK) caused by homozygous or compound heterozygous mutations was found in five case subjects (3,4). Other rare cases were also described (1), with mutations in the genes coding for insulin promoter factor-1 (IPF-1), forkhead box P3 (FOXP3), and the eukaryotic translation initiation factor 2α kinase 3 (EIF2AK3). Recently, Gloyn et al. (5) reported in an ethnically diverse patient cohort that six heterozygous activating mutations in the KCNJ11 gene, encoding the Kir6.2 subunit of the pancreatic β-cell ATP-sensitive K⁺ channel, caused PND in 10 probands and that some of these mutations are also associated with developmental delay, muscle weakness, and epilepsy. Homozygous inactivating mutations in KCNJ11 encoding Kir6.2 have also been shown (6) to cause familial persistent hyperinsulinemic hypoglycemia of infancy.

Here we present the screening for mutations of the coding sequence of KCNJ11 in 17 case subjects presenting with PND and having no GCK mutation and 7 case subjects with TND with no abnormalities of the chromosome 6q24 region. The patients were recruited through the French Network for the Study of Neonatal Diabetes (2). The 17 PND patients had a mean birth weight of 2,817 g (range 1,890–3,600) (Tables 1 and 2). None of the patients were born prematurely. Median age at diagnosis of diabetes was 60 days (range 1–260). At last follow-up visit, they were all on insulin, at a mean dose of 0.7 units · kg⁻¹ · day⁻¹ (range 0.18–1.2). The coding exon of the KCNJ11 gene was screened for mutations by direct sequencing of genomic DNA. No mutations were found in the patients with TND. On the contrary, in nine patients with PND (Table 1), we identified seven heterozygous missense mutations. Three of them (V59M, R201C, and R201H) have been previously described (5), and the four novel mutations resulted in amino acid substitutions at the following positions: F35L, G53N, E322K, and Y330C (Table 1). No Kir6.2 mutation or familial transmission was observed after testing both parents of each the nine patients, suggesting that they are carriers of de novo Kir6.2 mutations. In addition, six microsatellite markers from the chromosome 6q region have been analyzed in all nine of these families, allowing us to confirm paternity (7). None of the mutations were present in 90 nondiabetic unrelated French Caucasian subjects. The R201H mutation was found in three unrelated probands and occurred at a highly conserved codon, supporting a critical role for residue 201 in the channel function and ATP binding (5,8). All of the other mutated residues are highly conserved between mammals (man, rat, and mouse) and the Japanese Fugu fish (data not shown).

Therefore, KCNJ11 mutations are a common cause (53%) of PND in our French cohort, which allows us to look for genotype/phenotype correlations. No statistically significant phenotypic differences were noted between patients with or without a KCNJ11 mutation (Tables 1 and 2). Nevertheless, more patients with a Kir6.2 mutation were reported to have intrauterine growth retardation (<5th percentile for gestational age) compared with those without a mutation (6 of 9 vs. 2 of 8; P = 0.15); the statistically nonsignificant P value is most likely due to a lack of power because of the small number of case subjects. This suggests that mutations in KCNJ11 are associated with an in utero insulin secretory insufficiency, as previously suggested (5). However, two patients (DPN-9 and ND-C, carriers of a novel mutation, which were G53N and F35L, respectively) have a normal birth weight (Table 1). DPN-9 showed a normal weight gain in the few first postnatal weeks, suggesting that insulin was secreted in an adequate amount before an abrupt secretory failure. The patient ND-C also had a normal birth weight but was unable to gain weight in the neonatal period, suggesting inadequate insulin secretion of very early postnatal onset. Therefore the correlations between the type of mutation and the insulin secretory capacity both in utero and in the postnatal period as well as the effect of sulfonylureas need to be studied.

Two patients with a KCNJ11 mutation had diabetes with developmental or neurological features: one presenting with a severe developmental delay with leucodystrophy and no seizures and another has a history of hemiplegia. The patient with severe developmental delay had the V59M mutation, which has previously been reported (5) in an individual with developmental delay and muscle weakness. On the other hand, one subject with PND with no mutation in Kir6.2 had partial epilepsy too, which suggests that an unknown gene involved in the same pathway may be responsible for a similar but possibly milder phenotype than PND due to a Kir6.2 defect.

In this regard, although screening for KCNJ11 mutations should elucidate >50% of all PND cases in Caucasians, it does not identify a phenotypically specific form of disease. In addition, in our cohort, Kir6.2 deficiency is not involved in TND, which should be considered an etiologically different disease from PND.