Main

Joubert syndrome (JS) and related disorders (JSRD) are autosomal recessive and X-linked disorders characterized by hypoplasia of the cerebellar vermis with the characteristic neuroradiological ‘molar tooth sign’ and accompanying neurological symptoms including dysregulation of breathing pattern, ataxia and developmental delay. JSRD are classified into six subtypes: pure JS, JS with ocular defect, JS with renal defect, JS with oculorenal defects, JS with hepatic defect and JS with orofaciodigital defects.1 To date, 17 causative genes have been identified in JSRD: INPP5E,2 TMEM216,3 AHI1,4 NPHP1,5 CEP290,6 TMEM67,7 RPGRIP1L,8 ARL13B,9 CC2D2A,10 OFD1,11 TTC21B,12 KIF7,13 TCTN1,14 TMEM237,15 CEP41,16 TMEM138,17 and C5ORF42.18 Because of the clinical and genetic heterogeneity in JSRD, it can be very difficult to identify the causative mutations in individual cases.

We encountered five non-consanguineous Japanese families with JSRD (Figure 1a) and molar tooth sign was observed in all patients (Figures 1b–e, Supplementary Table 1). Peripheral blood samples were obtained from patients and their family members after written informed consent was given. To identify causative mutations, we performed whole-exome sequencing (WES) in five probands of the five families (one proband from each family). DNA was processed using the SureSelectXT Human All Exon 50 Mb library or V4 (51 Mb) library (Agilent Technologies, Santa Clara, CA, USA), and sequenced on a Genome Analyzer IIx sequencer (Illumina, San Diego, CA, USA) with 108 bp paired-end reads, or on a HiSeq2000 sequencer (Illumina) with 101 bp paired-end reads and 7 bp index reads. Image analysis and base calling were performed by Illumina pipeline. Approximately 3.8–6.0 Gb of sequence data were mapped to the human reference genome (GRCh37.1/hg19) with Novoalign or Burrows-Wheeler Aligner. The mean depth of coverage was 55–125 reads, with 88–96% of all coding exons being covered by 5 × or more reads.

Figure 1
figure 1

Familial pedigree and brain MRI of the patients. (a) JSRD families and mutations. (b) T2-weighted axial images of III-1, family 1. (c) T2-weighted axial images of III-2, family 1. (d) T2-weighted axial images of III-1, family 2. (e) T2-weighted axial images of III-2, family 2. The molar tooth sign is visible in all patients (arrowheads).

Full size image

Out of all variants within exons and ±20-bp intronic regions from the exon–intron boundaries, those registered in dbSNP135, 1000 Genomes and ESP5400 and located within the segmental duplications were removed. Homozygous or compound heterozygous variants of 17 JSRD causative genes were then picked up. In patients 1, 2, 3 and 4 whose DNA was captured by the SureSelectXT Human All Exon 50 Mb library, ∼90% of the entire coding regions in 13 of 17 causative genes were covered by 5 × reads or more. In patient 5 captured by the V4 (51 Mb) library, >90% of the coding region was covered by 5 × reads or more (Supplementary Table 2), indicating that the V4 library offered superior coverage to the SureSelectXT library around the regions of the JSRD genes.

All patients from the five families possessed novel compound heterozygous mutations or a homozygous mutation in known genes later confirmed by Sanger sequencing (Figure 1a): c.1862G>A (p.R621Q)/c.700dupC (p.L234Pfs*56) in INPP5E (9q34.3) for family 1; c.5788A>T (p.K1930*)/c.6012-12A>T in CEP290 (12q21.32) for family 2; c.329A>G (p.D110G)/c.2322+5delG in TMEM67 (8q22.1) for family 3; homozygous c.6012-12A>T in CEP290 for family 4; and c.214G>T (p.E72*)/c.6012-12A>T in CEP290 for family 5. No other variants within 17 known genes have been identified after excluding the variants of dbSNP135, 1000 Genomes and ESP5400. Clinical phenotypes caused by respective mutated genes are discussed in Supplementary text. In families 1, 2 and 4 in which parental samples were available, all parents were heterozygous carriers of one of the mutations. As parental samples were unavailable from families 3 and 5, we determined whether two mutations resided on different alleles by cloning an reverse transcriptase-PCR (RT-PCR) product amplified from total RNA of lymphoblastoid cells into a pCR4-TOPO vector (Life Technologies, Carlsbad, CA, USA) and sequencing. Each mutation was found in a different allele for both families (data not shown). Another variant, c.1894A>G (p.K632E) in CEP290, of family 2 was not found to be pathogenic based on web-based analyses such as SIFT, PolyPhen-2 and Mutation Taster (Supplementary Table 3). In families 2, 4 and 5 with a CEP290 abnormality, c.6012-12A>T was shared. On the basis of our in-house 135 exome data, the allele frequency of the mutation was 1/270 allele (0.74%), indicating that it may be a rare variant in Japanese. The other mutations were not found in our in-house 135 exome data.

Splicing effects were examined in families 3 and 4. RT-PCR was performed on RNA from lymphoblastoid cells of family members using primers spanning exons 42/43 and 45/46 in family 4 and exons 20/21 and 24/25 in family 3 (sequence information available on request). In family 4, only an aberrant cDNA was detected in II-2, whereas the parents (I-1 and I-2) showed two different products including one wild-type, which was detected in a control (Supplementary Figures 1a, b). Sequencing of the mutant product revealed a 57-bp insertion corresponding to the 3′-side of intron 43. As a result, a premature stop codon was introduced at intron 43. In family 3, RT-PCR detected a mutant cDNA in II-1 together with a wild-type product, which was detected in a control. Sequencing of the mutant product confirmed the skipping of exon 22, resulting in an in-frame 27 amino-acid deletion (Supplementary Figures 1c, d).

WES has proved a powerful tool for the identification of novel genes in genetic diseases. It also has tremendous potential for clinical diagnosis and is now being applied in the molecular diagnosis of single-gene disorders such as neurofibromatosis type 1, Marfan syndrome and multi-gene disorders such as retinitis pigmentosa.19 As shown here, WES would also be suitable for the diagnosis of JSRD, another multi-gene disorder. Though the read-coverage of the old version of SureSelect did not sufficiently collect genomic DNAs for four genes (INPP5E, TMEM216, KIF7 and TCTN1), the performance of the V4 (51 Mb) library was satisfactory for all genes. Further, as exome capture technology is based on hybridization it can be refractory to homologous regions, so other methods such as multiplex PCR amplification and multiple microdroplet PCR technology could be useful in addition.

In conclusion, we were able to identify causative mutations in five non-consanguineous families with JSRD using WES. The diagnostic utility of WES is obvious, implying that WES or other next-generation sequencing technologies will be a main factor of molecular diagnosis.