Introduction

Kabuki syndrome (KS) (OMIM#147920) is a multiple congenital anomaly/mental retardation syndrome.1, 2 Most patients have five cardinal manifestations: a distinctive face, skeletal anomalies, digit abnormalities, mild-to-moderate mental retardation and postnatal growth retardation.3 KS has an estimated incidence of 1 in 32 000, and 500 cases have been reported worldwide.4, 5, 6, 7 The majority of cases of KS are sporadic, and the sex ratio is nearly equal.8 In 2010, KMT2D (previously known as MLL2, NM_003482.3) was identified as the first KS causative gene.8 In 2012, a de novo deletion and a point mutation of KDM6A (NM_021140.2) were described as the second cause of KS.9 KMT2D comprises 54 coding exons and encodes a histone-lysine N-methyltransferase of 5537 amino acids that belongs to the trithorax protein group and regulates the transcription of a diverse set of genes involved in embryogenesis and development.10, 11 KDM6A comprises 29 exons and is one of the X chromosome genes that largely escape X inactivation.12 KDM6A demethylates di- and trimethyl-lysine 27 on histone H3.13 Because both KMT2D and KDM6A are histone modifiers, other pathogenic genes for KS might have functions that are related to those of these two genes.6 It has been suggested that KMT2D and KDM6A act together in the epigenetic control of transcriptionally active chromatin by counteracting Polycomb-group proteins.14

Although several reports have demonstrated the mutational spectrum of KS patients, there are no such reports in Korean patients. Here, we performed exome sequencing and phenotype characterization in 12 patients to identify their underlying disease-causing mutation and to delineate the clinical features of KS in Korean patients.

Materials and methods

Patients and clinical data

A total of 12 patients, 8 females and 4 males, were enrolled in this study from March 2009 to September 2012. The data were collected from three different clinical genetics centers in Korea over a protracted period and were forwarded for review to two of the authors. All patients exhibited at least four of the five cardinal manifestations of KS, as defined by Niikawa et al.3 Clinical manifestations, clinical courses and the results of genetic studies were reviewed retrospectively. Written consent was obtained from all participants who provided identifiable samples. The institutional review boards of two hospitals, Pusan National University Children’s Hospital and Seoul National University Children’s Hospital, approved this study.

Exome sequencing

We performed 12 exome captures using the Illumina TruSeq Exome Enrichment Kit (Illumina, San Diego, CA, USA). Pre-enrichment DNA libraries were constructed according to the Illumina TruSeq DNA Sample Preparation Guide. A 300–400-bp band gel was selected for each library, and exome enrichment was performed using Illumina’s TruSeq DNA Sample Preparation Kit v2, according to the manufacturer’s protocol. A quantity of 500 ng of each sample was pooled before using the TruSeq 62 Mb Exome Enrichment Kit. The final library was then validated on a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) and sequenced via 100 bp paired-end sequencing on a HiSeq2000 instrument (Illumina). Conventional Sanger sequencing was used to validate the variants found in 11 patients and their parents.

Sequence alignment, and detection of single-nucleotide variants and insertions and deletions

Raw reads in FASTQ format from exome sequencing were aligned to the hg19 reference genome from UCSC using Burrows Wheeler Aligner (v. 0.5.9, http://bio-bwa.sourceforge.net) using default parameters, including a seed length of 45 bp. Aligned reads were processed and polymerase chain reaction duplicates were removed using SAM Tools (v. 0.1.16, http://samtools.sourceforge.net), and regional realignment and quality score recalibration were carried out using the Genome Analysis Tool Kit (v. 1.4, BROAD institute, Cambridge, MA, USA). Subsequently, single-nucleotide variants and insertions and deletions were identified using 6 × of read depth coverage. Single-nucleotide variants and insertions and deletions were annotated according to genomic location (intergenic, 5′ untranslated region, intron, coding DNA sequence (CDS), splicing or 3′ untranslated region) using a self-annotation program based on the GRCh37. The splicing region was defined as 2 bp from both ends of the coding exon–exon junction region. Single-nucleotide variants in coding DNA sequence were annotated as synonymous, missense or nonsense. Insertions and deletions in coding DNA sequence were annotated as inframe or frameshift. Functional annotation tools included SIFT (http://sift.jcvi.org/), PolyPhen2 (http://genetics.bwh.harvard.edu/pph2/), MutationTaster (http://www.mutationtaster.org/), Online Mendelian Inheritance in Man (OMIM) and Clinical Signal and Human Gene Mutation Database (HGMD) (HGMD Professional v. 2012.2. 2012, https://portal.biobase-international.com). The suspected dominant and recessive variants were further filtered using 1000 genomes, dbSNP (dbSNP137) and Korean Personal Genome Project (KPGP) information (http://opengenome.net/). Whole-exome sequencing revealed that >90% of the coding regions of KMT2D and KDM6A were covered by an average read depth of 52–76 per sample.

Results

Clinical features of Korean patients with KS

Twelve patients were diagnosed with KS over a period of 3 years and 6 months. The sex ratio had a high proportion of females: four patients were male and eight were female. The mean age at the clinical diagnosis was 50.1 months (range, 8.8–140.7 months), and none of the 12 patients had a family history of KS. Phenotypic characterization of all patients was performed, and its results are summarized in Table 1. All patients showed the characteristic of facial dysmorphism, including long palpebral fissures, everted lower eyelids and broad nasal root. Most patients showed developmental delay/mental retardation (91.7%) and hypotonia (75.0%). Ophthalmologic problems were common, particularly strabismus (41.7%). Congenital heart defects were present in nine patients (75.0%): coarctation of the aorta (33.3%), ventricular septal defect (25.0%), atrial septal defect (16.7%), Scimitar syndrome (8.3%), hypoplastic left heart syndrome (8.3%) and interrupted aortic arch type B (8.3%). Skeletal abnormalities, including 5th finger shortening/clinodactyly (100%), hip dislocation (25.0%) and joint hypermobility (16.7%), were also observed. Sacral dimpling was also observed in five patients (41.7%). Most patients (8 out of 12 patients; 66.7%) had a history of recurrent ear infections. Conventional karyotype and molecular cytogenetic analyses, including array-based comparative genomic hybridization containing BAC clones spaced at 1 Mb intervals across the genome, revealed an absence of pathological abnormalities in all patients.

Table 1 Summary of clinical and genotypic characteristics of 12 Korean patients with KS

Mutational spectrum

Mutational analysis using whole-exome sequencing was performed in 12 patients who were suspected of having KS. We identified KMT2D and KDM6A variants in 10 and 1 patient, respectively (Table 1 and Figure 1). However, these variants were not detected in their parents, and all mutations identified were de novo. Each patient had a unique individual mutation. No recurrent mutations were identified. The mutations identified were relatively evenly distributed across exons 14–46 of the KMT2D gene, without a definite hot spot, although four mutations were located in exon 39. These 10 mutations included four small deletions or insertions causing a frameshift (K3, 6, 8 and 9), four nonsense mutations (K2, 7, 10 and 11) and two missense mutations (K1 and 4). Nine of the 10 (90.0%) KMT2D mutations were novel. Only one (10.0%) mutation, which was identified in patient K7, had been reported previously.7 Eight of the 10 (80.0%) KMT2D mutations were nonsense mutations or frameshift mutations that resulted in protein truncation. The two novel missense mutations were not found in ESP6500 (http://evs.gs.washington.edu/EVS/) as well as 1000 genome, dbSNP (dbSNP137) and KPGP information (http://opengenome.net/). One KDM6A mutation identified in K12 was a novel splice-site mutation.

Figure 1
figure 1

Partial sequences of KMT2D and KDM6A showing the mutations detected in this study. Except the mutation identified in patient K7 (c.12592 C>T, p.R4198*), the other nine KMT2D and one KDM6A mutations were novel.

Discussion

In this study, we investigated the clinical manifestations and genetic characteristics of Korean patients with KS. Causative genes for KS were identified very recently (in 2010 and 2012), and only one patient with a confirmed KMT2D mutation has been reported in Korea.15 To develop a diagnostic approach for a large number of targets, we performed whole-exome sequencing, which was useful for the rapid and simultaneous screening of candidate variants and genes linked to KS. In fact, if the regions of the two genes (KMT2D and KDM6A) are sequenced as 52–76X, the budget for this experiment will not be much higher than that for Sanger sequencing. In addition, whole-exome sequencing not only allows the identification of causative mutations in KMT2D and KDM6A genes simultaneously, but also provides an opportunity to identify another causative gene in patients without a mutation in the KMT2D or KDM6A genes.

Our study showed that 11 out of 12 patients with KS had KMT2D or KDM6A mutations; the mutation-detection rate was higher than the rate expected (55.8–80.0%) based on previous studies.2, 6, 16, 17, 18 The higher yield observed in this study might be due to the strict criteria used to establish the clinical diagnosis, because the mutation-detection rates would depend on the clinical diagnostic level and the inclusion/exclusion of atypical cases. A previous study reported that KMT2D mutations are localized mainly in two exons: 39 (20.3%) and 48 (13.5%).16 In the present study, 4 of the 10 KMT2D variants identified were also located on exon 39. Exon 39 contains several regions that encode long polyglutamine tracts, suggesting the presence of a mutational hot spot.2 Considering the high cost of the genetic testing using conventional sequencing for KS, a stepwise screening approach starting in this region might be an adequate way to achieve a molecular diagnosis of this disease. All but one variant found in the KMT2D gene in this study was novel, and most of them were nonsense or frameshift mutations that were predicted to truncate the polypeptide chain before translation of the SET domain, suggesting a loss of function of the protein. Therefore, haploinsufficiency seems to be the likely mechanism underlying the KS phenotype. The putative functional relevance and pathogenicity of the novel missense mutations identified in the present study were predicted in silico using various software, including SIFT, PolyPhen2 and MutationTaster. The two novel missense mutations (p.C1424F and p.R4420Q) were predicted to affect protein function and to be disease-causing mutations.

Recently, the KDM6A gene was identified as the second causative gene for KS, and de novo deletions and point mutations in the KDM6A coding region have been identified in 9–13% of KMT2D-negative KS patients.6, 9 Reported patients with KDM6A mutations exhibit a broad phenotypic spectrum, ranging from typical KS to a milder clinical presentation.6, 9 Patient K12 in this study, who had a 4-bp deletion in KDM6A, presented with typical KS phenotype, including long palpebral fissures, a short nasal septum, lateral eversion of the lower eyelid, profound short stature, cardiac abnormality, severe mental retardation, recurrent infection and Chiari malformation.

However, the genetic basis of KS remains unknown in 20–45% of patients.16 The emerging spectrum of KMT2D and KDM6A mutations in KS, to which this study added 10 novel mutations, provides further understanding of the etiology of this disease. Because the mutation-detection rate was high in this study, rigorous mutation analysis of KMT2D and KDM6A may be an important first step toward the early diagnosis and genetic counseling of Korean patients with KS.

Regarding the correlation between KS and neoplasia, only five such cases have been reported, one each with Burkitt’s lymphoma,19 acute lymphoblastic leukemia,20 neuroblastoma,21 fibromyxoid sarcoma22 and neuroblastoma and hepatoblastoma.23 In this study, patient K5 was diagnosed as having abdominal Burkitt’s lymphoma (stage IV) at the age of 3 years. He had very typical clinical features of KS, including a congenital heart defect (coarctation of the aorta and ventricular septal defect), mental retardation, seizures and immune dysfunction. However, no KMT2D or KDM6A mutation was detected, and additionally performed oligonucleotide array-based comparative genomic hybridization using NimbleGen CGX-3 135K Whole-Genome Array (Roche Diagnostics, Mannheim, Germany) showed no deletions or duplications in this case. The failure to discover a mutation in one case was one of the limitations of this study, and this brings forth several hypotheses. The first is the presence of locus heterogeneity, including other genomic elements that may also cause clinical KS, similar to KMT2D mutation-positive KS. The second is the possibility of the presence of mutations in regulatory or deep intronic regions. We checked all TruSeq Exome Enrichment Kit (Illumina)-covered regions to address the possibility of the existence of other disease genes. However, whole-genome sequencing, which is more expensive than exome sequencing, would be necessary to read the regions that are not covered by exome sequencing. The third is the possibility of missing a mutation because of poor coverage or insufficient depth in the two candidate genes. No large intragenic deletions or mosaic mutations were identified because exome sequencing has the limitation of structure variation.24, 25 To find mosaic mutations, exome sequencing of the patient’s parent sample would be required. However, we did not perform this analysis because of the associated costs. Therefore, further investigation is required to identify the genetic basis in patients with mutation-negative KS.

In conclusion, this research identified 11 mutations in KS patients, including 10 novel mutations in the KMT2D or KDM6A genes; in particular, it yielded a high mutation-detection rate (91.7%). To the best of our knowledge, this is the first and the largest mutational analysis performed using exome sequencing in Korean patients with KS. The results of this study supported the contention that exome sequencing has become an immensely powerful tool for the analysis of disease-causing variants in extensive genomic data, and might be applied diagnostically to KS.