A boy with homozygous microdeletion of NEUROG1 presents with a congenital cranial dysinnervation disorder [Moebius syndrome variant]

The 6-year-old proband was the first child of healthy consanguineous Turkish parents. His parents were second cousins. The parents reported that their common grandfather developed hearing impairment before age 50 years and that a great-granddaughter of their grandfather´s sister had received cochlea implants due to congenital deafness. Further data were not available. The proband was born by vacuum extraction after an uneventful pregnancy of 40 weeks gestation. The birth weight was 3,750 g (50th to 75th percentile), length 52 cm (50th percentile), head circumference 35 cm (50th percentile) and Apgar scores were 8 and 9 at 1 and 5 minutes. After birth, he was ventilated for one day and treated with antibiotics due to suspected neonatal sepsis. At age 4 months, he was hospitalized because of episodes of screaming followed by sudden tonus loss. Global developmental delay, muscular hypotonia and torticollis were noticed. A metabolic disease was suspected but no diagnosis was made. A brainstem evoked response audiometry (BERA) test at age 8 months indicated profound bilateral deafness. Cranial computed tomography imaging showed bilateral stenosis of the internal auditory canal with a maximum diameter of 1.1 mm and a hypoplastic cochlea with only a single widened cochlear turn (Figure 1). Magnetic resonance imaging confirmed the abnormalities and Mondini malformation was diagnosed. The vestibulocochlear nerve was assumed to be aplastic because no nerval structure reaching the cochlea could be identified. At age 8 months the boy was fitted with hearing aids and at age 12 months with a cochlear implant (CI) on the left, but without any improvement in hearing or speech acquisition. At age 2 ½ years he began to walk freely. At age 4 years, he was fitted with a cochlear implant on the right, but again without improving hearing or speech.

When last seen at age 5 11/12 years, his length was 110 cm (5th percentile), weight was 21 kg (50th percentile) and head circumference 53 cm (75th percentile). He had plagioscaphocephaly, low-set posteriorly rotated ears, high arched eyebrows, elongated palpebral fissures, high and very narrow palate, funnel chest, sacral dimple, and hypoplastic genitalia. He wore ortheses because of faulty foot posture. His mother reported that he could ride a tricycle, but not a bicycle. He could stand on one leg only for a few seconds and an equilibrium disorder was assumed (Figure 2). Neurological findings included muscular hypotonia, increased salivation and a severe gulp and mouth motor disorder. He could not chew and needed mashed food. Intelligence testing showed large variations in results: his IQ was 70 using the SON-R (Snijders-Oomen non-verbal intelligence test) but 92 using the CPM-Raven (Raven´s Coloured Progressive Matrices), indicating normal abilities to think clearly and make sense of complexity (eductive ability), plus normal abilities to store and reproduce information (reproductive ability).

At age 1 year, karyotyping and mutation analysis of the GJB2 and GJB6 genes were performed with normal results. At age 3 years, a subtelomere screen using multiplex ligation-dependent probe amplification (MLPA) yielded normal results. Genome-wide microarray and NGS analyses were performed on samples taken at age 5 11/12 years.

The proband was among >200 patients studied by us using genome-wide microarray analysis. The Affymetrix Genome-Wide Human SNP Array 6.0 (~1,850,000 probes, 1.7 kb average distance between neighbouring probes) was used with the GeneChip Genome-Wide SNP Assay Kit 6.0 (Affymetrix, Santa Clara, CA). Data were analysed with the Genotyping Console 4.0 software (Affymetrix). Detection of CNVs required at least 5 consecutive abnormal probes spanning at least 100 kb in length. Results were verified using quantitative real-time PCR (qPCR).

Using one flow cell lane of an Illumina HiSeq 2000, we generated 115,420,519 50 nt single end reads, with 101,701,184 reads passing standard quality filters. Reads were aligned to the human reference genome (GRCh37/hg19) using the Bowtie algorithm (version 0.12.7) [12]. 7,417,427 (7%) of the reads failed to align and 17,125,015 (17%) aligned to multiple locations and were removed, leaving 77,455,596 (76%) unambiguously aligned reads. After sorting reads based on genomic position using the samtools software package [13], breakpoints were determined using a statistical approach implemented in the R programming language. Based on the existing genetic and microarray data, we expected a homozygous deletion and thus expected no sequence reads to align to the deleted region. Away from the suspected deletion and outside of centromeres, we determined the distribution of a genomic position to the closest sequence read. Using this distribution, we determined the likelihood that a given position was in a homozygously deleted locus: a genomic position far from the closest sequencing read is more likely to be in a deleted locus. For a given position, the assigned p-value represents the fraction of all genomic positions that are an equal or greater distance from the closest read. The region identified by the Affymetrix SNP 6.0 GeneChip and the FREEC algorithm was examined in detail, including the aligned reads and the resultant p-values (Figure 3A, B) and identified a region where no reads aligned.

The precise deletion sequence locus was determined using PCR with primers designed based on the NGS results, GCTTTTTCTCCTAAATTCTCTGG (forward) and TCAGCCATCTTTGTTGTTTCC (reverse) (Figure 3C). PCR was performed according to standard protocols. The reaction mixture consisted of 2.5 μl 10x PCR buffer, 2.5 μl 50 mM MgCl2, 2.5 μl 10 mM dNTP mix, 1.0 μl (100 ng) of each forward and reverse primer, 0.5 μl (2.5 U) Taq polymerase, 14 μl PCR-grade water und 100 ng template DNA. PCR amplification was carried out with an initial denaturation step at 94°C for 3 min, 35 cycles of 94°C for 30 sec, primer-specific annealing temperature of 55°C for 30 sec, 72°C for 60 sec and a final extension step at 72°C for 10 min. Following Exo/SAP digestion, PCR products were sequenced using the CEQ DTCS Quick Start Kit (Beckman Coulter, Krefeld, Germany) and separated on a Beckman Coulter CEQ 8000 Genetic Analysis System. Sequencing data were analysed using the BLAST program (http://​www.​ensembl.​org/​Homo_​sapiens/​blastview and http://​www.​ncbi.​nlm.​nih.​gov/​BLAST).

Microarray analysis showed a homozygous deletion arr 5q31.1(134,764,990–134,875,777)x0 (GRCh37/hg19, ISCN2009) in the patient and identical but heterozygous deletions in the parents (Figure 4). The deletions were confirmed using qPCR. The proximal breakpoint on chromosome 5q31.1 was identified between 134,764,553 and 134,764,990 bp and the distal breakpoint between 134,875,777 and 134,881,349 bp (GRCh37/hg19), indicating a deletion size between 110,787 bp (minimum size) and 116,796 bp (maximum size) by microarray analysis (6,009 bp uncertainty).

Based on the NGS data for chromosome 5 (Figure 3), the proximal and distal breakpoint on 5q31 mapped between 134,764,588 and 134,764,722 bp (GRCh37/hg19) and 134,879,972 and 134,880,106 bp (p ≤ 0.05), respectively, indicating a deletion sized between 114,922 bp and 115,518 bp (269 bp uncertainty). By use of a breakpoint spanning PCR, we then mapped the precise breakpoints to 134,764,601 bp (last preserved base, proximal breakpoint) and 134,879,936 bp (first preserved base, distal breakpoint), indicating a deletion size of (exactly) 115,334 kb. The finding matched well with the approximate deletion interval established by microarray analysis and confirmed that the deletion interval comprised three genes, NEUROG1, TIFAB and DCNP 1, encompassing all exons and likely the corresponding promotor regions (Figure 4).