Mosaic Paternal Uniparental Isodisomy and an ABCC8 Gene Mutation in a Patient With Permanent Neonatal Diabetes and Hemihypertrophy

A male child weighing 3.87 kg at 40 weeks gestation (75th percentile) was born to nonconsanguineous, Somali parents. There had been two previous uneventful pregnancies. The proband was discharged home on the day of birth but returned to hospital aged 36 days, failing to thrive, severely dehydrated, and acidotic (pH 6.99 [normal range {NR} 7.3–7.4]). The initial blood glucose measured was 50 mmol/l with a bicarbonate of 5 mmol/l (NR 21–34) and urinary ketones of >160 mg/dl. A diagnosis of neonatal diabetes was made, and subsequent to resuscitation the child was stabilized on twice-daily subcutaneous insulin. Both parents had normal fasting glucose levels (father 5.2 mmol/l, age 36 years; mother 4.6 mmol/l, age 31 years) and no prior history of diabetes.

The child had left hemihypertrophy (Fig. 1) but no macroglossia, anterior abdominal wall defects, ear creases, or peri-auricular pits as seen in BWS. Echocardiography of the heart revealed branch pulmonary artery stenosis but no cardiomegaly, while serial abdominal ultrasounds identified increasing left renal hypertrophy compared with a normal right kidney.

The young age of the patient at presentation suggested a diagnosis of neonatal diabetes. Blood was collected for DNA analysis of chromosome 6 anomalies causing transient neonatal diabetes (19) and for KCNJ11 mutation analysis (which at this time was the only channelopathy described as causing permanent neonatal diabetes). These tests were both normal. GAD, insulin, and islet cell autoantibodies were all negative. The child's karyotype was normal: 46 XY. However, microsatellite analysis of chromosome 11 revealed the child to be mosaic for paternal uniparental isodisomy (UPD) of 11pter-11p14 (Table 1 and Fig. 2A). As neonatal diabetes is not characteristic of either BWS or isolated hemihypertrophy, this prompted a review of potential candidate genes for diabetes that were encompassed within this region of UPD. The SUR1 gene (ABCC8), which lies within this region, was a logical candidate, as it coassembles with Kir6.2 (KCNJ11) to form the K_ATP channel, and activating mutations in KCNJ11 are known to cause neonatal diabetes (9–11).

The ABCC8 gene was sequenced as described previously (14). The father was identified as being heterozygous for a missense mutation, N72S (c.215A>G), while the proband's leukocyte DNA showed mosaicism for the N72S mutation (Fig. 2B). The N72S mutation was not found in 250 normal chromosomes, and the affected residue was conserved across species. Quantification by real-time PCR (TaqMan assay) demonstrated that the N72S mutation was present at ∼70% in leukocyte DNA and 50% in buccal cells. Briefly, genomic DNA from leukocytes and buccal cells was amplified using a mutation-specific TaqMan approach. Genomic DNA from the father, who is heterozygous for the N72S mutation, was used in serial dilution to produce standard curves to determine linear range and accuracy of quantification (primer and probe sequences are available on request). Reactions contained 5 μl TaqMan Fast Universal PCR Mastermix without Amperase, 0.5 μl Assays-by-Design (Applied Biosystems, Warrington, U.K.) probe and primer mix (corresponding to 36 μl of each primer and 8 μl of each probe), 2.5 μl water, and 2 μl DNA at a concentration of 100 ng/μl. Amplification conditions consisted of a single cycle of 95°C for 20 s followed by 40 cycles of 95°C for 1 s and 60°C for 20 s. The relative level of mutant transcript relative to wild-type transcript was determined by the ΔΔCT method (20). Each test was carried out in triplicate.

Functional studies in Xenopus oocytes (methods in 9,10) demonstrated that the N72S mutation results in a reduced sensitivity to inhibition by MgATP. Half-maximal block of homozygous N72S channels was produced by 23 ± 1 μmol/l ATP (n = 10) compared with 15 ± 1 μmol/l (n = 6) for wild-type channels. Two-electrode voltage-clamp studies showed a concomitant small increase in the whole-cell K_ATP channel resting current (14) (Fig. 3). Homozygous N72S channels were substantially blocked by the sulfonylurea tolbutamide in vitro (Fig. 3).

Identification of an ABCC8 mutation, together with the patient's relatively poor control on twice-daily insulin (GHb 8.3% NR 4.0–6.0%), the known efficacy of sulfonylureas in neonatal diabetes caused by K_ATP channel mutations (21), and the tolbutamide block of mutant N72S channels observed in functional studies, prompted a trial of the sulfonylurea glyburide. Over a 2-week period, under intensive monitoring in the hospital, regular subcutaneous insulin was withdrawn and glyburide gradually increased from a starting dose of 0.1 mg · kg⁻¹ · day⁻¹ to a maximum dose of 1 mg · kg⁻¹ · day⁻¹.

The therapeutic trial was terminated at 2 weeks due to the complete absence of a clinical response, with high blood glucose levels (up to 40 mmol/l) requiring the repeated administration of short-acting insulin to improve glycemia and prevent metabolic decompensation. The C-peptide level before initiating glyburide was <94 pmol/l and remained at <94 pmol/l on 1 mg · kg⁻¹ · day⁻¹ glyburide. The child is now over 18 months of age and is treated with a basal bolus insulin regimen of four injections daily (∼1 unit · kg⁻¹ · day⁻¹) to maintain growth and health (recent A1C 6.6%). Although his neurological development at this stage seems normal, he remains under regular surveillance both for neurocognitive development and for increased tumor risk.

Uniparental isodisomy is a rare abnormality that provides insights into the genetic basis of human disease. Its identification in two cases of transient neonatal diabetes (22) led to identification of imprinting anomalies associated with the transient form of this condition. UPD can also unmask genetic disorders with an autosomal recessive pattern of inheritance (23). In this case, the child's hemihypertrophy and relatively high birth weight (75th centile) did not fit with a diagnosis of permanent neonatal diabetes as, if anything, these features would be expected to co-segregate with hypoglycemia secondary to hyperinsulinemia (as seen in BWS). Since UPD can unmask recessive disorders, SUR1 (ABCC8) was an obvious candidate, due to its position within the disomic region of interest, its importance in insulin secretion, and the fact that focal adenomatous hyperplasia of islet cells with congenital hyperinsulinemia has been described in which loss of maternal 11p15 is associated with homozygosity of an ABCC8 mutation inherited from a heterozygous father (24).

Recent studies have identified activating mutations in ABCC8 as causing both transient and permanent neonatal diabetes (3–5). Several cases of permanent neonatal diabetes responded to glyburide at doses ranging from 0.22 to 0.59 mg · kg⁻¹ · day⁻¹ (3), including one child with a dominant ABCC8 mutation (15) and at least four children with recessively inherited ABCC8 mutations (A.T. Hattersley, S.E.F., S.E., unpublished observations). Unfortunately, the child reported here was unresponsive to glyburide at a dose of as much as 1 mg · kg⁻¹ · day⁻¹. The complete absence of any clinical response in our infant prompted our withdrawal of glyburide after a 2-week trial, although a further trial when the child is older may be warranted given the in vitro response. The lack of a therapeutic response to sulfonyureas in our patient is puzzling, as the mutant N72S channels were blocked by tolbutamide in in vitro studies. It is possible that other imprinted genes involved in the paternal UPD might confer a reduced response.

In conclusion, we report a patient with UPD of chromosome 11 who presented with neonatal diabetes as opposed to the classical phenotype of hypoglycemia due to hyperinsulinemia. Paternal UPD of chromosome 11pter-11p14, encompassing the K_ATP channel genes and the BWS locus, unmasked a recessively acting gain-of-function ABCC8 gene mutation, thus causing diabetes in addition to features of BWS. As has previously been described, the heterozygous father has no history of glucose intolerance or evidence of diabetes in adult life, as this is a recessively acting mutation (14). The level of mosaicism for the N72S mutation varied between leukocytes (∼70% mutation) and buccal cells (∼50% mutation). As the child is alive and well, the mutation load within the pancreas remains unknown. Given the evidence that children with UPD11 can have highly variable levels of mosaicism in different organs, we hypothesize that the pancreas must exhibit a high level of paternal UPD cells to present with this phenotype, the somatic recombination causing the paternal UPD occurring very early during embryogenesis (25).

We have recently identified nine other patients with recessively inherited ABCC8 mutations causing neonatal diabetes (14).

We acknowledge funding from the Sir Graham Wilkins studentship to S.E.F. and support from the Research & Development Directorate at the Royal Devon & Exeter NHS Foundation Trust to S.E.