Origin of de Novo KCNJ11 Mutations and Risk of Neonatal Diabetes for Subsequent Siblings

Abstract

Context

Activating mutations in the KCNJ11 gene, which encodes the Kir6.2 subunit of the pancreatic β-cell K_ATP channel, result in permanent and transient neonatal diabetes. The majority of KCNJ11 mutations are spontaneous, but the parental origin of these mutations is not known.

Objective

Our objective was to determine the parental origin of de novo KCNJ11 mutations and investigate the possibility of mosaicism in transmitting parents.

Design

We identified 68 index cases with a KCNJ11 mutation where neither parent was known to be affected. DNA was available from both parents of 41 probands. The parental origin of the mutation was determined in 18 families by examination of pedigrees, microsatellite analysis, or allele-specific PCR.

Results

A nonsignificant excess of paternally derived mutations was found with 13 of 18 (72%) shown to have arisen on the paternal allele. There was no evidence to suggest an association with increased age at conception. In two families, there were half-siblings with permanent neonatal diabetes born to an unaffected father, suggesting germline mosaicism that was confirmed by the presence of the R201C mutation in one father’s semen. Somatic mosaicism was detected in one unaffected mother, and this mutation will also be present in her germ cells.

Conclusion

De novo KCNJ11 mutations can arise either during gametogenesis or embryogenesis. The possibility of germline mosaicism means that future siblings are at increased risk of neonatal diabetes, and we recommend that molecular genetic testing is routinely offered at birth for subsequent siblings of children with de novo KCNJ11 mutations.

NEONATAL DIABETES MELLITUS (NDM) is a rare disorder that can be permanent (PNDM) or transient (TNDM) (1). Heterozygous activating mutations in the KCNJ11 and ABCC8 genes, which encode the K_ATP channel subunits Kir6.2 and SUR1, have recently been identified (2–4). KCNJ11 mutations are most common, accounting for 47% of PNDM (4, 5) and a few cases of TNDM (6, 7). Most of these patients achieve improved glycemic control after transfer from insulin injections to sulfonylurea tablets (8).

Approximately 84% of patients with KCNJ11 mutations have no family history of neonatal diabetes, and their diabetes results from a de novo (spontaneous) mutation (9). De novo mutations arise in the parent either during gametogenesis (germline de novo mutations) or as an early postzygotic event. We previously reported a family with paternal germline mosaicism where two paternal half-siblings were affected with PNDM (10). If a postzygotic mutational event happens before the separation between somatic and germinal lineages, the individual will show both somatic and germline mosaicism, although the level of mosaicism will vary between tissues and may not be detectable in leukocyte DNA. The percentage of mosaicism in the germ cells will depend upon the timing of the mutation, but if all primordial germ cells are affected, then the recurrence risk for future siblings can be up to 50% (11).

The aim of this study was to determine the origin of de novo KCNJ11 mutations causing neonatal diabetes. The possibility of somatic mosaicism (and hence germline mosaicism) was investigated in the transmitting parents using a quantitative real-time PCR assay.

Subjects and Methods

Subjects

We identified 68 probands with a KCNJ11 mutation and no family history of neonatal diabetes (12). DNA from both parents was available for 41 cases. The majority were recruited after a request for referrals to the International Society of Pediatric and Adolescent Diabetes (ISPAD). Informed consent was obtained from all subjects.

Genetic analysis of KCNJ11

Leukocyte DNA was extracted using standard methods. DNA was extracted from semen as described previously (13).

Haplotype analysis and family relationship confirmations were performed by analyzing microsatellites D11S902, D11S419, D11S1397, D11S1901, D11S921, and D11S1888. Gender identity of parental samples was confirmed by analysis of the amelogenin locus.

The KCNJ11 gene was sequenced as described previously (13) in probands and parental samples to confirm that the mutations occurred spontaneously in the proband and to genotype the coding single-nucleotide polymorphisms (SNPs) E23K (rs5219), A190A (rs5218), and I337V (rs5215) for the molecular haplotyping assay.

Molecular haplotyping

PCR primers were designed to amplify a single copy of the KCNJ11 gene using a primer complementary to one allele of a heterozygous SNP or the heterozygous mutation (Table 1). Sequence analysis of the PCR amplicon revealed the coinherited allele and hence the parental origin of the mutation.

Table 1. Allele-specific primers for molecular haplotyping.

Primer	Primer sequence (5’–3’)	Genotype of single allele
1	CGCCTGGCAGAGGACCCTGCT G	E23K (GAG)-F
	CTGGATGAGCAGCAGGGGGAG	E23K (GAG)-R
2	CGCCTGGCAGAGGACCCTGCT A	E23K (AAG)-F
	CTGGATGAGCAGCAGGGGGAG	E23K (AAG)-R
3	CAAGCATGCGGTGATCGCC	A190A (GCC)-F
	ACGCCTTCCAGGATGACGAT	A190A (GCC)-R
4	CAAGCATGCGGTGATCGCT	A190A (GCT)-F
	ACGCCTTCCAGGATGACGAT	A190A (GCT)-R
8	CCGCTTCCTGCAGGACGG	V59 (GTG)-F
	CTGTGGTCCTCATCAAGC TG	V59 (GTG)-R
11	AAGCAGAGGCGGCCGTGGCGCAGG	A190A (GCC)-F
	CCGAGAGGACTCTGCAGTGA	A190A (GCC)-R
12	GAGTGGTGTGGGCACTTTGC C	I337V (GTC)-F
	TCAGCAAGCATGCGGTGATC	I337V (GTC)-R
13	GAGTGGTGTGGGCACTTTGCT	I337V (ATC)-F
	TCAGCAAGCATGCGGTGATC	I337V (ATC)-R
14	CTTGCGGAGGTCACCCACG C	R201H (CGT)-F
	TCACCGGAGCCATGCTGTC	R201H (CGT)-R
15	CTTGCGGAGGTCACCCACGT	R201H(CAT)-F
	TCACCGGAGCCATGCTGTC	R201H (CAT)-R

The italicized 3’-terminal nucleotide is allele specific. The underlined nucleotides are mismatched from the normal sequence to enhance allelic discrimination. F, Forward; R, reverse.

Real-time PCR quantification

Genomic DNA from leukocytes was investigated for low-level mosaicism using an allelic discrimination assay as described previously (10) with the addition of a preamplification stage of 10 cycles using fragment 1 or 2 primers (13) for analysis of the V59M, E227K, and G53R mutations.

Restriction enzyme digest

The restriction endonuclease AflIII (New England Biolabs Ltd., Hitchin, Hertfordshire, UK) was used to digest the normal allele from 1 μg semen or leukocyte DNA for 24 h at 37 C using 20 U enzyme. The digested DNA was then subject to 35 cycles of amplification using fragment 2 primers (12). This amplicon was then sequenced.

Results

De novo KCNJ11 mutations were proven in 41 families where neither parent was shown to carry the mutation, and microsatellite analysis of markers flanking the KCNJ11 locus confirmed family relationships. There were four families where after an initial spontaneous mutation, there were then two generations affected with neonatal diabetes (Fig. 1). Genotyping and microsatellite analysis established that these R201H and E227K mutations originated as de novo events in the grandparental generation and arose in two families on the paternal allele (one R201H and one E227K) and in two families on the maternal haplotype (Fig. 1 and Table 2). We previously reported a family (ISPAD 58) with two affected half-siblings born to an unaffected father (10). In this study, a second family was identified (ISPAD 115) where the proband and her affected paternal half-sibling were heterozygous for the R201C mutation, but the father was not affected.

Fig. 1.

Microsatellite analysis of markers flanking the KCNJ11 gene in four families with two generations affected with neonatal diabetes. Black bars indicate the haplotype cosegregating with neonatal diabetes on which the de novo mutation has arisen. Squares denote males, circles females, and black symbols show individuals affected with neonatal diabetes. N/M, Heterozygous mutation; N/N, mutation not present. The KCNJ11 gene is located between markers D11S902 and D11S921. Arrows point to the proband in each family.

Table 2. Origin of de novo KCNJ11 mutations.

ISPAD family ID	KCNJ11 mutation	PNDM/TNDM	Technique for determining chromosome of origin	Parental origin	Germline/somatic mosaicism	Age of parent at conception (yr)	Country of origin	Ref.
137	H46Y	PNDM	Allele-specific PCR	Maternal	Not tested	19	Slovakia	12
79	G53R	TNDM	Allele-specific PCR	Paternal	Not detected	31	UK	6
138	V59M	PNDM	Allele-specific PCR	Maternal	Not detected	25	UK	12
25	V59G	PNDM	Allele-specific PCR	Paternal	Not tested	30	Czech Republic	4
50	I182V	TNDM	Allele-specific PCR	Maternal	Somatic and germline	39	UK	6
22	R201H	PNDM	Allele-specific PCR	Paternal	Not detected	20	Australia	4
110	R201H	PNDM	Allele-specific PCR	Paternal	Not detected	Not known	Holland	12
19	R201H	PNDM	Microsatellite analysis	Paternal	Not detected	24	Holland	4
NECKER4	R201H	PNDM	Allele-specific PCR	Paternal	Not tested	Not known	France	20
113	R201H	PNDM	Microsatellite analysis	Maternal	Not detected	31	Canada	12
139	R201H	PNDM	Allele-specific PCR	Paternal	Not detected	45	Canada	Not published
140	R201H	PNDM	Allele-specific PCR	Paternal	Not detected	19	Slovakia	12
41	R201C	PNDM	Allele-specific PCR	Paternal	Not detected	31	Holland	4
115	R210C	PNDM	Pedigree analysis	Paternal	Germline mosaic	49	Bulgaria	Not published
58	R201C	PNDM	Pedigree analysis	Paternal	Inferred germline mosaic	34	Canada	10
57	E227K	TNDM	Microsatellite analysis	Maternal	Not detected	20	German	Not published
114	E227K	TNDM	Microsatellite analysis	Paternal	Not detected	28	Canada	Not published
10	E229K	TNDM	Allele-specific PCR	Paternal	Not tested	Not known	UK	Not published

An allele-specific PCR assay was used to determine the chromosome of origin in 12 of the remaining 35 families in whom informative SNPs were located within 500 base pairs of the mutation. The molecular haplotyping method was validated by confirming the paternal origin of the R201H mutation in family ISPAD 19 (Fig. 2). We found that nine of these mutations had arisen on the paternal allele, and three were of maternal origin (Table 2). The parental origin of KCNJ11 mutations was therefore determined in a total of 18 families and demonstrated a moderate but not significant bias toward a paternal origin (13 of 18 mutations; χ ² P = 0.17).

Fig. 2. Allele-specific PCR to determine chromosome of origin in family ISPAD 19.

A, The heterozygous R201H (c.602G→A) mutation arose spontaneously in the proband’s mother, who is also heterozygous for SNP rs5218 (C/T; A190A). Her mother is homozygous and her father heterozygous for this SNP. B, Standard KCNJ11 PCR detected the heterozygous mutation and polymorphism in the mother (product 1). Allele-specific PCR using a specific primer for the T allele of the SNP amplifies a single product (product 2). Sequencing of this allele-specific PCR product revealed an A at nucleotide 602. This demonstrates that the mutant allele is in cis with the T allele of the rs5218 SNP, and therefore the R201H mutation arose on the paternal chromosome. Squares denote males, circles females, and black symbols show individuals affected with neonatal diabetes. An arrow is used to identify the proband.

A quantitative real-time PCR assay was used to investigate the possibility of low-level somatic mosaicism in the transmitting parent. The limit of sensitivity for the TaqMan probes was calculated as more than 0.4% for the R201C allele-specific probe (10), whereas all others had a probe sensitivity of more than 0.8% (data not shown). There was no evidence of somatic mosaicism in leukocyte DNA from 13 parents (Table 2), but the I182V mutation was present at 2.9% in the mother of a child with TNDM (ISPAD 50).

Real-time PCR analysis of leukocyte and semen DNA samples from the unaffected father in family ISPAD 115 did not show somatic or germline mosaicism for the R201C mutation present in his two affected children. We investigated this further by using an AflIII restriction digest of genomic DNA to specifically cleave the normal sequence and thus enrich for the mutant R201C allele. These digested DNA samples were then subject to PCR amplification, and sequencing of the resulting PCR products showed the presence of the mutant allele in the paternal semen DNA sample but not in his leukocyte DNA (Fig. 3). It was not possible to perform this analysis in the father of family ISPAD 58 because he had undergone a vasectomy.

Fig. 3.

Analysis of the R201C mutation in somatic and germ cells. Sequence analysis of PCR product derived from leukocyte DNA from the ISPAD 115 proband in whom the R201C (c.601C→T) mutation is heterozygous (A), leukocyte DNA from the unaffected father after enrichment for the mutation (B), and semen DNA from his unaffected father after enrichment for the mutation by restriction endonuclease digestion of the normal allele (C). The position of nucleotide 601 is indicated by an arrow.

The median parental age at conception was 31 yr (range, 19–49 yr) for KCNJ11 mutations of paternal origin and 25 yr (range, 19–39 yr) for maternal origin (P = 0.5, Mann-Whitney U test). Comparison with the population median age at conception was not possible due to the small sample size as a result of the diversity of populations represented within this study (Table 2).

In addition to the 41 cases where a de novo mutation was proven by molecular genetic testing, we identified KCNJ11 mutations in 27 probands in whom there was no parental history of neonatal diabetes but DNA samples from one or both parents were not available for testing. In only the two families described above was there a second sibling affected with neonatal diabetes (two of 68, 3% of families).

Discussion

The majority (29 of 41, 71%) of de novo mutations identified in this study were substitutions at CpG nucleotides that are likely to be the result of the spontaneous deamination of methylcytosine to thymine. The most common KCNJ11 mutations (V59M, R201H, R201C, E227K, and E229K) all occur within the context of a CpG dinucleotide and encode C→T or G→A transitions.

It was possible to determine the parent of origin in 18 families, and a moderate excess of paternal mutations (13 of 18) was observed. In two families, the inheritance pattern suggested paternal germline mosaicism because two half-siblings had PNDM, but neither parent was affected. Paternal germline mosaicism was confirmed in one family by analysis of semen DNA. The R201C mutation was present at a very low level (less than 0.4%), which was unexpected because two of his three children are affected with neonatal diabetes. The germline mosaicism is likely to be the result of a postzygotic mutation during embryonic development, and therefore the mosaicism may not be restricted to the germline. Although there was no evidence of somatic mosaicism in either father’s leukocyte DNA, samples from other tissues (e.g. buccal cells, hair roots, or fibroblasts) were not available for testing, and hence somatic mosaicism cannot be excluded.

Somatic mosaicism was detected in a single transmitting parent of 14 tested. This mother (ISAPD 50) was mosaic for the I182V KCNJ11 mutation. The presence of the mutation in leukocyte DNA is consistent with a postzygotic mutation occurring early in embryonic development. The I182V mutation will also be present in the mother’s germ cells, although it is not possible to measure the mutation load and hence the risk to future siblings. However, if the mutation arose early enough to affect all the primordial germ cells, then the recurrence risk could be up to 50% (11). This is the first report to suggest that neonatal diabetes may be caused by maternal germline mosaicism for a KCNJ11 mutation.

The occurrence of both paternal and maternal mutations is consistent with KCNJ11 mutations resulting from postzygotic mutations during embryogenesis in an unaffected parent and from mitotic errors during gametogenesis. Other genes showing a similar pattern of de novo mutations include APC, VHL, NF1, NF2, and EFNB1 mutations (causing familial adenomatous polyposis, Von Hippel-Lindau, neurofibromatosis, or craniofrontonasal syndrome) (14–16). Postzygotic mutations occurring during embryogenesis will show no sex bias because they are random events, and no correlation with increased parental age at conception is predicted.

The situation for KCNJ11 mutations is distinct from the exclusive paternal origin of de novo FGFR2 and FGFR3 mutations in Apert syndrome and achondroplasia where there is a significant association with increased paternal age at conception (17, 18). Recent evidence suggests that these gain-of-function mutations confer a selective advantage to spermatogonial cells leading to their accumulation over time (13, 19).

Germline mosaicism was indicated in two families by pedigree analysis and in one of 12 additional parents tested for somatic mosaicism. Within our cohort of 68 families with presumed or proven de novo KCNJ11 mutations, a second affected child was born to two families. From this series, the empiric recurrence risk for affected siblings due to parental germline mosaicism is 3%. Although this may be an overestimate due to the increased likelihood of referral for genetic testing for families with two affected children, it is important that the possibility of germline mosaicism is discussed during counseling. In view of the increased risk of neonatal diabetes compared with the population, we recommend that the parents of all children with KCNJ11 mutations are offered molecular genetic testing for future siblings at birth. A cord blood sample can be tested quickly (within 3 d), and a normal result will provide reassurance. In the unlikely event of a mutation being identified, monitoring of blood glucose levels can be initiated promptly to ensure early diagnosis and appropriate treatment to achieve good glycemic control from the outset.

Abbreviations

ISPAD

International Society of Pediatric and Adolescent Diabetes

NDM

neonatal diabetes mellitus

PNDM

permanent NDM

SNP

single-nucleotide polymorphism

TNDM

transient NDM