SETD1B encodes suppressor of variegation (Su[var], enhancer of zeste, Trithorax) (SET) domain-containing protein 1B (SETD1B), a component of a histone 3 lysine 4 methyltransferase [1, 2]. In mammals, six histone 3 lysine 4 methyltransferases, including SETD1B, are derived from the yeast Set1 orthologue, and mediate universal epigenetic marks [3]. SETD1B binds to a unique set of target genes, with a non-redundant role in gene expression and epigenetic control of chromatin structure [4]. High transcriptional level SETD1B expression was previously reported in the human fetal and adult brain. The SET domain consists of 130 amino acids that are conserved among vertebrates [5], and is seen in a variety of chromosomal proteins.

De novo missense variants in SETD1B were reported in association with neurodevelopmental disorders; epilepsy, developmental delay, intellectual disabilities, autistic behavior, and craniofacial dysmorphic features in two patients [1]. In the present report, we identified a de novo truncating frameshift SETD1B variant in a patient with neurodevelopmental delay and epilepsy by whole-exome sequencing. The patient was a Japanese girl born to non-consanguineous healthy parents. At the age of 1 year 11 months, she experienced jerks with fever. Since then, she exhibited daily myoclonus seizures when falling asleep as well as awake state associated with epileptic discharges on electroencephalogram (EEG) (Supplemental Fig. 1). Interictal EEG did not show epileptic discharges. She showed developmental delay and autistic behavior. We performed whole-exome sequencing for this patient, and detected a de novo variant in SETD1B (NM_015048.1: c.5644_5647del:p.(Ile1882Serfs*118)), but no other candidate variants that may cause the disease. The allele frequency of this SETD1B variant was 0 in ExAC (Exome Aggregation Consortium; http://exac.broadinstitute.org/), gnomAD (http://gnomad.broadinstitute.org/), and the 1000 Genomes Project (http://www.internationalgenome.org/). Sanger sequencing of the parents’ genomic DNA and the patient’s DNA confirmed that this variant occurred de novo (Fig. 1a, b). Microsatellite haplotype analysis confirmed the biological parentage (data not shown). This variant was located in the last exon of SETD1B (Fig. 1d). A web-based prediction program predicted that this variant may be disease-causing (Mutation Taster, http://www.mutationtaster.org/). SETD1B is very likely to be a dominant gene by DOMINO (https://wwwfbm.unil.ch/domino/index.html). The probability of being loss-of-function intolerant was 0.02 in ExAC, suggesting that this gene may be tolerant to haploinsufficiency [6]. This variant is predicted to escape from nonsense-mediated decay (NMD). If premature terminal codons by pathogenic variants are located more than 50−55 base pairs upstream of the last exon−exon junction, variant alleles suffer from NMD [7]. However, as this frameshift variant is located in the last exon, NMD may not be involved. This frameshift variant is predicted to disrupt the posterior SET domain and add an aberrant 118 amino acids (Fig. 1e). Using the ACMG/AMP (American College of Medical Genetics and Genomics/ the Association for Molecular Pathology) 2015 clinical guidelines [8], we classified the mutation as ‘likely pathogenic’ (PS2: de novo, PM2: absent from controls, PP3: multiple lines of computational evidence support a deleterious effect on the gene or gene product). However, using the eXome Hidden Markov Model [9], we found no pathogenic copy number variations. Thus, the SETD1B frameshift variant was likely causative in this patient.

Fig. 1
figure 1

Familial pedigree and SETD1B variants. a A familial pedigree with the SETD1B genotype. b Sanger sequencing shows a de novo 4-base pair deletion in SETD1B. c Sanger sequencing of SETD1B cDNA containing wild-type and mutant alleles. Inhibition of nonsense-mediated mRNA decay (NMD) using cycloheximide did not alter expression of the mutant transcript, suggesting that the mutant allele is not subjected to NMD. d Schematic presentation of reported 12q24.3 deletions involving SETD1B, as well as pathogenic SETD1B variants. Reported variants (p.(Arg1842Trp) and p.(Arg1859Cys)) [1] are at the SET domain, which is highly conserved among vertebrates. The variant in this study (NM_015048.1:c.5644_5647del: p.(Ile1882Serfs*118)) is also located in the last exon (17 out of 17) of the SET domain. The SETD1B protein consists of 1923 amino acids, and the SET domain located from 1784 to 1901 amino acids. The amino acid sequence from 1837 to 1901 is shown. e Schematic presentation of amino acid sequences of wild-type and mutant alleles, and the SET domain. The mutant allele terminates the latter part of the SET domain, and added a completely different amino acid sequence (p.(Ile1882Serfs*118))

Full size image

We checked the SETD1B transcripts in the patient’s lymphoblastoid cell lines (LCLs), and confirmed that the mutant allele was not degraded by NMD [7] (Fig. 1c). Therefore, the aberrant transcript may be translated to a truncated protein.

To date, two de novo missense variants in SETD1B (NM_015048.1) have been reported in patients with developmental delay, autistic behavior, and myoclonic seizures. These mutations were located at the evolutionally well-conserved SET domain, which is predicted to modulate gene activity and/or chromatin structure. The SET domain has catalytic function for histone methyltransferase [4]. SET domains of the mammalian histone methyltransferase SUV39H1 catalyze tri-methylation of the H3K9 N-terminus, which suppresses transcription [10]. By contrast, a subset of the SET domain with adjacent cysteine-rich regions (pre-SET and post-SET) likely restricts histone methyltransferase activity, causing enhanced transcription [11]. Some chromatin-associated proteins including SETD1B have this conserved SET domain [12].

Labonne et al. [13]. reported that patients with a 360-kb microdeletion at 12q24.31 involving SETD1B showed intellectual disability, autism, epilepsy, and craniofacial anomalies. Baple et al[14]., Qiao et al. [15]., and Palumbo et al. [16]. also reported patients with a 12q24.3 deletion involving SETD1B who share some clinical features with our case, including developmental delay, intellectual disability, and epilepsy. The clinical features of individuals with SETD1B deletions are summarized in Table 1. All of the 12q24.3 deletions contain six genes (KDM2B, ORA1, MORN3, TMEM120B, RHOF, SETD1B). However, the functions of these deleted genes remain unclear, making it difficult to perform a genotype−phenotype correlation. Some of the clinical features of these patients may be caused by SETD1B deletion.

Table 1 Clinical features of individuals with SETD1B variants and 12q24.3 deletions (Hiraide et al. [1] ; Labonne et al. [13]; Baple et al. [14]; Qiao et al. [15]; Palumbo et al. [16])
Full size table

We found three variants in gnomAD; c.87G>A:p.(Trp29*), c.5598+1G>C, and c.544+2T>C (Table 2). These three variants are suspected to be degraded by NMD. Thus, we speculate that the loss of one SETD1B allele may not be damaging. Denovo-db reported a possible NMD-inducible variant (c.3013dupA) in a female with the developmental disorder [17]. There remains a possibility that this variant (c.3013dupA) may be related to the phenotype in this individual, however, neither detailed phenotype nor pathogenesis of the variant is not clearly described. Thus, we do not know this variant truly caused the disease. Instead, two of the previously reported de novo variants (c.5524C>T:p.(Arg1842Trp) and c.5575C>T:p.(Arg1859Cys)) [1] were missense, in which NMD was not involved. In addition, MutPred-LOF software, which predicts the pathogenicity of frameshift and stop-gain variants (score 0−1, benign to pathogenic of loss-of-function pathogenicity) [18], showed a probability score of our variant (p.(Ile1882Serfs*118)) of 0.367, suggesting that loss-of-function is less likely. However, the accuracy of these in-silico prediction tools is uncertain, and gain-of-function may possible in a patient with an SETD1B mutation. The C-terminal SET domain of SETD1B interacts with a subunit of histone methyltransferase complexes [19], which methylate promoters and regulate gene expression [20]. Thus, our frameshift variant may alter the C-terminal conformation of SETD1B to cause epigenetic changes, affecting normal gene expression in patients’ cells. However, further studies in patients and functional analyses are required.

Table 2 Comparison of SETD1B damaging missense variants and non-damaging LoF variants in our case, reported cases, and in databases
Full size table

In summary, we report a novel de novo frameshift variant in SETD1B in association with autistic behavior, developmental delay, intellectual disability, and myoclonic epilepsy.