Homozygous Mutations in NEUROD1 Are Responsible for a Novel Syndrome of Permanent Neonatal Diabetes and Neurological Abnormalities

This study was conducted in accordance with the Declaration of Helsinki. The study protocol was approved by the local ethics committee, and written informed consent was obtained from the parents or guardians of each patient.

We studied 44 probands with PNDM diagnosed before 6 months of age, who had been referred to the Molecular Genetics Laboratory at the Peninsula Medical School in Exeter, U.K.. Mutations in KCNJ11, ABCC8, INS, and GCK had been excluded. The relevant clinical information was obtained from the medical records.

Genomic DNA was extracted from peripheral leukocytes using standard procedures. The single coding exon of NEUROD1 was PCR amplified in three overlapping fragments using specific primers for each amplicon tagged with 5′ M13 tails to allow sequencing to be performed with a universal M13 primer (primers and conditions available upon request). Single-strand sequencing was carried out using standard methods on an ABI 3730 sequencer (Applied Biosystems, Warrington, U.K.). Sequences were compared with the published template (accession no. NM_002500) using Mutation Surveyor (version 3.20; SoftGenetics). Sequence variants were tested for their presence in family members whenever DNA was available.

High-density single nucleotide polymorphism (SNP) genotyping was carried out on the Affymetrix human 10K Xba chip by Medical Solutions Nottingham (formerly GeneService) (Nottingham, U.K.). Processing of genomic DNA was performed in accordance with the Affymetrix protocol. In-house Perl scripts were developed to automatically identify genomic homozygous segments, defined by at least 20 consecutive homozygous SNPs marking a region that exceeded 3 cM (13).

Two novel homozygous mutations in NEUROD1, a single base pair duplication (c.364dupG) and a two-base pair CT deletion (c.427_428del), were identified in two unrelated probands. Both mutations result in a frameshift and a premature truncation of the C terminus of the expressed protein (p.Asp122Glyfs*12 and p.Leu143Alafs*55, respectively), leading to mutated proteins completely lacking the transactivation domain (Fig. 1). These mutations had not been previously documented and were not present in 200 alleles from healthy unrelated individuals. No mutations were identified in the remaining 42 patients.

The two homozygous probands inherited the mutation from their heterozygous parents (Fig. 2). In family A with the c.364dupG mutation, parents were known to be first cousins and, consistent with parental consanguinity, SNP genotyping analysis of the proband revealed a total genomic homozygosity value of 6.0% (13). The mutation-containing homozygous segment was the largest homozygous segment (46.6-Mb long) and spanned 2q31.1–2q36.1 delimited by the SNPs rs726032 to rs724149. In contrast, in family B, the parents of the patient with the homozygous c.427_428del mutation were not known to be related and, in keeping with this, total genomic homozygosity value was very low (0.3%). However, the mutation in both parents was inherited on an extended haplotype of 10.4 Mb between positions Chr2q31.1–32.1 (SNPs rs2884471–rs722385), suggesting that the mutation arose from a single common ancestor.

The two probands were diagnosed with permanent diabetes within the first 2 months of life and had presented with intrauterine growth retardation (birth weights 1,490 and 2,230 g at 34 and 38 weeks of gestation, respectively), reflecting reduced insulin secretion in utero. They had no evidence of pancreatic exocrine dysfunction and normal pancreatic size on abdominal scanning (see supplementary information, available in an online appendix [http://diabetes.diabetesjournals.org/cgi/content/full/db10-0011/DC1]). In addition to diabetes, they presented with a similar pattern of neurological abnormalities including moderate-to-severe developmental delay, profound sensorineural deafness, and visual impairment due to myopia and diffuse retinal dystrophy. Brain magnetic resonance imaging scans showed severe cerebellar hypoplasia with no other major intracranial abnormalities (Fig. 3 and supplementary information). A more detailed clinical description is given in Table 1.

There was limited availability of other family members for genetic and clinical testing. The diabetes status, age of diagnosis, treatment, and genetic testing result of family members are shown in Fig. 2. We assessed glucose tolerance in the four parents of the two probands who were proven heterozygous carriers of the mutations. In family A (c.364dupG mutation), the mother had been diagnosed with type 2 diabetes at 33 years, despite having a normal BMI, and was treated with glicazide. In contrast, the father (also aged 33 years) had normal fasting (4–6 mmol/l) and postprandial (5–7 mmol/l) blood glucose levels on several occasions. In family B (c.427_428del mutation), the mother and father underwent standard oral glucose tolerance tests (aged 33 and 37 years, respectively) that confirmed normal glucose tolerance (6.2 and 4.8 mmol/l, respectively, at 2 h). No heterozygous family members in either family had any developmental delay or neurological features on clinical examination.

We report the first two cases of PNDM caused by homozygous mutations in NEUROD1. The patients with this novel autosomal recessive syndrome not only had early-onset permanent diabetes but also presented with developmental delay, cerebellar hypoplasia, and hearing and visual impairment. This is the 13th gene in which mutations have been described in patients with permanent neonatal diabetes.

NEUROD1, a tissue-specific member of the bHLH family of transcription factors, is expressed in developing pancreatic islets and in mature β-cells. It forms a heterodimer with the ubiquitous bHLH transcription factor E47 that binds to specific E-box motifs on specific target genes, including INS, GCK, and ABCC8, to regulate their expression (14,–16). The two homozygous NEUROD1 mutations both introduce a frameshift that results in truncated proteins lacking the transactivation domain, which has been shown to be important for the interaction of NEUROD1 with its main coactivator, p300 (17). These are likely to have no biological activity, as shown previously for a different frameshift mutation (c.616dupC, p.His206Profs*38) identified in a patient with NEUROD1-MODY (9). The two patients have a remarkably consistent phenotype (Table 1), with clinical features in keeping with the known expression and biology of this transcription factor, and this provides further evidence that the homozygous mutations in NEUROD1 are causative.

Both patients have neonatal diabetes but a normal pancreas on scanning and no evidence of exocrine dysfunction. This is consistent with the central role of NEUROD1 in islet development. Mice lacking Neurod1 die shortly after birth from severe diabetic ketoacidosis (8). Histological examination of the Neurod1-deficient pancreas shows an impaired islet morphogenesis with a reduction in the number of endocrine cells—especially β-cells (8).

In addition to diabetes, our two patients presented with a similar pattern of neurological features, including developmental delay, cerebellar hypoplasia, and visual and hearing impairment. This is in keeping with the abundant expression of NEUROD1 in the developing and mature nervous system. Interestingly, the initial Neurod1-null mice that rapidly died from diabetes had no obvious anatomic and histologic abnormalities of the brain (8). However, it is possible to explore the role of Neurod1 in the nervous system by rescuing Neurod1-null mice either by expressing a transgene encoding the mouse Neurod1 gene under the insulin promoter (18) or by crossing the null mutation into a different genetic background to reduce the severity of the diabetes (19). The rescued Neurod1-null mice show a similar neuronal phenotype consisting of impaired balance, ataxic gait, circling, and swaying head movement as a result of impaired cerebellum development (18,–20). Furthermore, rescued Neurod1-deficient mice have abnormal hearing and vision as a result of severe sensory neuronal defects in the inner ear and neural retina, respectively (20,–22). The main feature seen in the mouse that was not present in our patients was epilepsy (19). The remarkable similarity between the NEUROD1-deficient patients and the Neurod1-deficient mice (Table 2) strongly supports a similar biological role of this transcription factor across species.

Homozygous mutations in PTF1A, which encodes another bHLH transcription factor, also cause a syndrome of neonatal diabetes and cerebellar hypoplasia/agenesis (23). However, in this condition the pancreatic phenotype is not limited to the islets; affected patients have pancreatic hypoplasia/aplasia. In keeping with the islets representing <1% of the endocrine pancreas, the size of the pancreas was found to be normal in our two patients with homozygous NEUROD1 mutations. This suggests that shared developmental pathways are important during development in the pancreas and the cerebellum.

Although heterozygous loss-of-function mutations in NEUROD1 have previously been identified as a very rare cause of diabetes in humans (9,,–12), diabetes was present in only one of four heterozygous mutation–carrying parents. Their age at the time of the study ranged from 33 to 39 years and does not exclude the possibility of developing diabetes later in life. In addition, incomplete penetrance has been described in some of the families with NEUROD1 diabetes (9). Homozygous mutations in other known MODY genes, namely GCK and IPF1, have previously been associated with isolated PNDM and isolated pancreatic agenesis, respectively (24,25). We have shown that homozygous mutations in another MODY gene are also associated with a more severe phenotype of neonatal diabetes.

In conclusion, homozygous mutations in NEUROD1 constitute a rare novel autosomal recessive cause of neonatal diabetes with severe neurological abnormalities. This confirms the important role that NEUROD1 plays in the development of both the pancreas and the nervous system in humans.

This work was funded by European Union FP6 Integrated Program 2006–2010 EuroDia (LSHM-CT-2006-518153) and the Welcome Trust (grants 067463/Z/2/Z and 083270/Z/07/Z). O.R.-C. is supported by an “Ayuda para contratos post-Formación Sanitaria Especializada” from the “Instituto de Salud Carlos III” (FIS CM06/00013). S.E. is employed as a core member of staff within the National Institute for Health Research–funded Peninsula Clinical Research Facility. A.T.H. was a Wellcome Trust Research Leave Fellow.

No potential conflicts of interest relevant to this article were reported.

O.R.-C. researched clinical data and wrote the manuscript. J.A.L.M. researched molecular genetic data and reviewed the manuscript. I.K. and D.W. researched clinical data and reviewed the manuscript. S.E. and A.T.H. reviewed and edited the manuscript.

We thank Ann-Marie Patch from the Peninsula Medical School for her excellent support with homozygosity mapping analysis.