Article Text

Short report
Whole exome sequencing identifies a splicing mutation in NSUN2 as a cause of a Dubowitz-like syndrome
  1. Fernando Jose Martinez1,
  2. Jeong Ho Lee1,
  3. Ji Eun Lee1,
  4. Sandra Blanco2,
  5. Elizabeth Nickerson3,
  6. Stacey Gabriel3,
  7. Michaela Frye2,
  8. Lihadh Al-Gazali4,
  9. Joseph G Gleeson1
  1. 1Department of Neurosciences and Pediatrics, Neurogenetics Laboratory, Institute for Genomic Medicine, Howard Hughes Medical Institute, University of California, San Diego, California, USA
  2. 2Wellcome Trust Centre for Stem Cell Research, University of Cambridge, Cambridge, UK
  3. 3The Broad Institute of MIT and Harvard, Cambridge, Massachusetts, USA
  4. 4Department of Paediatrics Faculty of Medicine & Health Sciences United Arab Emirates University Al Ain, United Arab Emirates
  1. Correspondence to Dr Joseph G Gleeson, Department of Neurosciences and Pediatrics, Neurogenetics Laboratory, Institute for Genomic Medicine, Howard Hughes Medical Institute, University of California, San Diego, CA 92093, USA; jogleeson{at} Dr Lihadh Al-Gazali, Department of Pediatrics, United Arab Emirates University, School of Medicine and Health Sciences, 17666 Al Ain, United Arab Emirates.


Background Dubowitz syndrome (DS) is an autosomal recessive disorder characterized by the constellation of mild microcephaly, growth and mental retardation, eczema and peculiar facies. Over 140 cases have been reported, but the genetic basis is not understood.

Methods We enrolled a multiplex consanguineous family from the United Arab Emirates with many of the key clinical features of DS as reported in previous series. The family was analyzed by whole exome sequencing. RNA splicing was evaluated with reverse-transcriptase PCR, immunostaining and western blotting was performed with specific antibodies, and site-specific cytosine-5-methylation was studied with bisulfite sequencing.

Results We identified a homozygous splice mutation in the NSUN2 gene, encoding a conserved RNA methyltransferase. The mutation abolished the canonical splice acceptor site of exon 6, leading to use of a cryptic splice donor within an AluY and subsequent mRNA instability. Patient cells lacked NSUN2 protein and there was resultant loss of site-specific 5-cytosine methylation of the tRNA(Asp GTC) at C47 and C48, known NSUN2 targets.

Conclusion Our findings establish NSUN2 as the first causal gene with relationship to the DS spectrum phenotype. NSUN2 has been implicated in Myc-induced cell proliferation and mitotic spindle stability, which might help explain the varied clinical presentation in DS that can include chromosomal instability and immunological defects.

  • Dubowitz
  • NSUN2
  • MISU
  • RNA methylation
  • microcephaly
  • genetics
  • neurosciences
  • developmental
  • cancer, head and neck
  • cell biology
  • MicroRNA
  • oncology

Statistics from

Dubowitz syndrome (DS, Omim %223370) is a rare, developmental disorder characterised by small stature, intellectual disability, mild microcephaly and a distinct facial appearance consisting of blepharophimosis, broad nasal bridge, sloping forehead and micrognathia.1 2 Although first reported in 1965 by Dubowitz as a condition characterised by familial low birthweight, dwarfism with unusual facies and skin eruption,3 the phenotypic spectrum has been expanded to include a wide variety of less frequent features such as congenital heart defects, frequent infections and chromosomal instability.4 DS is distinguished from other blepharophimosis-mental retardation syndromes such as Say-Barber-Biesecker variant of Ohdo syndrome,2 recently linked to mutations in the KAT6B gene encoding a histone acetyltransferase.5 To date, there are approximately 140 DS reported cases with no causative mutations identified. Although there is a presumed recessive mode of inheritance in most cases, there are reports suggesting that more complex modes of inheritance can occur.6

We identified a consanguineous multiplex family 1455, of Lebanese origin, in the United Arab Emirates (figure 1A). The eldest sibling was healthy with no pathological features. The three affected members displayed features consistent with DS, including low birth weight (5% or less), and oligohydramnios in the two older affected individuals (table 1), which are compared with typical features of DS.2 By 1 year of age, growth retardation and developmental delay were noted. Facial features included blepharophimosis, telecanthus, hypertelorism and high or broad nasal bridge (figure 1B). Oral features included smooth philtrum, downturning upper lip and dental anomalies. Eczema, at the time of evaluation, was clearly apparent over the chest and back in the youngest affected and required treatment, although the two older affected individuals did not have a history of notable eczema. The neurological features included microcephaly (−2% ile occipital-frontal head circumference at birth, progressing to −4 SD at 2 years of age), axial hypotonia and autistic features. The eldest affected displayed epileptic seizures, controlled on valproate, a feature not observed in the younger two affected individuals at the time of evaluation. The brain MRI was unremarkable. The remainder of the history and physical was non-contributory and serum chemistry, immune profiles and routine karyotype were normal. However, a notable hoarse voice, triangular face, round nose tip and abnormal ears, occasionally observed in DS, were not reported. We therefore cautiously refer to this condition as a Dubowitz-like condition, although it may, in fact, represent DS and the pedigree is consistent with a recessive mode of inheritance.

Figure 1

Pedigree, phenotype and predicted effect on splicing of the NSUN2 mutation. (A) Family-1455 presents three affected children from a first-cousin marriage. (B) Photograph of affected individuals, demonstrating blepharophimosis, telecanthus, hypertelorism, high/broad nasal bridge, smooth philtrum and downturning upper lip in 2 and 3. Eczema on the nape and back are visible in 4 (arrow). (C) Invariant conservation across species of the G residue at the predicted exon 6 splice acceptor site (arrow), with mutant G>C base indicated.

Table 1

Phenotype of published compared with current patients

In order to identify the gene responsible for this condition, we performed whole exome sequencing on two affected siblings. The study was approved by the ethics committee, and the family provided informed consent for the study. Blood DNA was extracted using Qiagen reagents, then subjected to exome capture with the Agilent SureSelect Human All Exome 50 Mb kit, sequenced on an Illumina HiSeq2000 instrument, resulting in ∼94% target coverage at >10× depth. GATK7 was used for variant identification and intersected with identity-by-descent blocks identified by HomozygosityMapper8 and then filtered for homozygous variants shared between the two affected individuals using custom Python scripts (available upon request).

We identified a homozygous variant in an intronic region of the gene NSUN2, located one base pair upstream of the start of exon 6 (chr. 5 base position 6,622,214 C>G; hg19) at the canonical splice acceptor site (figure 1C). The variant was found to be homozygous in all three affected cases, occurred with a 19.5 cM block of homozygosity in both affected members between chr. 5 base position 2,752,824-22,078,584 (online supplementary figure 1), was absent in the single unaffected sibling, and heterozygous in each parent, and thus, segregated according to a recessive mode of inheritance.

There were six other shared novel homozygous genetic variants, each leading to amino acid transversions annotated to be potentially damaging, occurring in the CUL4A, RIMS2, TARP, PKHD1L1, CACHD1 and TRHR genes. However, the reference amino acid for each was not fully conserved across evolution, and further, upon direct testing was found to fail the test of segregation in the family. There were no other likely deleterious variants that segregated, according to the presumed recessive mode of inheritance, supporting pathogenicity of the NSUN2 variant. The variant was not encountered in our in-house exome database of ∼1000 Middle Eastern individuals, and was not reported in any public databases. We also found the variant absent in a panel of 96 ethnically matched Arab controls. The mutated position was highly conserved across species, and annotation with SeattleSeq predicted it to be critical for splicing. Direct Sanger sequence analysis of four additional patients with DS-like phenotypes for all coding and splice sites yielded no obvious mutations, suggesting genetic heterogeneity.

The NSUN2 gene is alternatively spliced with at least four splice variants and three proposed alternative start sites of transcription. To determine whether the exon 6 splice acceptor mutation leads to altered mRNA levels, we performed quantitative reverse transcriptase PCR. Dermal fibroblasts were obtained from two affected individuals, the mother and one unrelated control, cultured under similar conditions, and total RNA was reverse-transcribed using SuperScript III (Invitrogen, Carlsbad, CA). Amplifications from exon 5 to 6 and exon 6 to 7 were compared with a control reaction against TATA-binding protein using the LightCycler 480 SYBR Green I (Roche, Indianapolis, IN) assay. All assays were normalised to glyceraldehydes 3-phosphate dehydrogenase (GAPDH), and then to the expression level of the unrelated control. We found a severe 95% reduction in normalised expression for both products from both affected patients, whereas, the obligate carrier mother showed about a 50% reduction in amplification, suggesting a functional consequence of the mutation on the NSUN2 mRNA (figure 2A). Because exon 6 is not present in all deposited NSUN2 mRNAs, and its absence maintains the open reading frame, we additionally assessed amplifications from exon 5 to 7, 7 to 8 and 8 to 9. In all reactions, we found comparable 70–80% reduction in NSUN2 mRNA, which excluded the possibility of trivial exon skipping. The results suggest a global reduction in mRNA levels from altered splicing, resulting in either partial or complete loss of mRNA in the two affected individuals.

Figure 2

Altered NSUN2 mRNA levels and aberrant splicing in affected individuals from Family-1455. (A) Each experiment compares quantitative reverse transcriptase PCR (qRT-PCR) levels from healthy unrelated control with mother and two affected members. Severely reduced incorporation of exon 6 was detected from exon 5 to 6 and 6 to 7 reactions, whereas, the mother displayed about 50% of the control. TATA binding protein qRT-PCR, used as control. Reactions from exon 5 to 7, 7 to 8 and 8 to 9 showed overall reduction in NSUN2 mRNA levels. (B) Agilent Bioanalyzer visualisation of aberrant splicing of NSUN2 mRNA in Family-1455. Amplification from exon 5 to 7 produced a 210 bp product in control and mother, severely reduced in the affected individuals, who instead, show a 250 bp product, also evident in the mother at reduced level, containing an AluY transposon.

To determine the effect on splicing, we analysed RT-PCR products from the fibroblasts of the affected patient and control by Bioanalyzer (Agilent, Inc., Santa Clara, CA) electrophoresis. Amplification from exon 5 to 7 demonstrated an aberrant-sized product of about 250 bp from the two affected individuals, also evident in the mother's sample, whereas, the predicted size product of 210 bp was notably reduced in the two affected individuals. Products from exon 5 to 6 were reduced without an obvious difference in molecular weight (figure 2B). Analysis of the reference genome for possible cryptic splice acceptor and donor sites identified an AluY transposon located between exon 6 and 7. Replacement of exon 6 with a single arm of the transposon is predicted to yield a 250 bp cDNA product from exon 5 to 7. In order to determine if this AluY exon was incorporated in the mRNA, we sequenced the aberrant 250 bp band. As predicted, we identified a sequence corresponding precisely to the 3′ arm of the AluY transposon (online supplementary figure 2A). This sequence trace demonstrated splicing from exon 5 to the 3′ AluY arm to exon 7, which is predicted to maintain the reading frame (online supplementary figure 2B, 3). Alu repeats have a two-part structure consisting of 5′ and 3′ arms separated by an ‘A’-rich linker region. The 3′ arm is terminated by a polyA tail.9 Novel insertions of Alu elements within the genome have been linked to disease, and their incorporation into mRNAs is frequently associated with microRNA-mediated degradation. Indeed, dozens of human microRNAs are predicted to target conserved regions of Alu transposons.10

In order to test for an effect on NSUN2 protein expression, we generated whole cell lysates from cultured fibroblasts and performed Western analysis by loading each sample in duplicate. We probed the blot for NSUN2 protein using a rabbit affinity-purified polyclonal antibody (Proteintech Catalog 20854-1), generated against full-length recombinant human protein. In the control and mother cell lines, we detected robust protein expression at the expected size of about 100 kD, similar to the published molecular weight.11 Protein loading was controlled by probing the blot with anti-alpha-tubulin. In both patient cell lines, however, there was undetectable protein (figure 3A). To confirm these results, we fixed and stained these fibroblasts with NSUN2 antibody. Because NSUN2 is a nuclear and nucleolar protein, we counterstained nuclei with Hoechst. In the control and mother lines, NSUN2 staining overlapped precisely with nuclei and nucleoli. In both patient lines, in images acquired with identical settings, there was striking reduction in staining for NSUN2, without detectable signal in either nucleoplasm or nucleoli (figure 3B). The data suggests severe reduction in NSUN2 protein levels in patient cells.

Figure 3

Reduced NSUN2 protein levels in affected cell lines from Family-1455. (A) Western blot from whole cell extracts of the control, mother and two affected individuals (Aff #1 and Aff #2), loaded in duplicate, show undetectable protein, compared with alpha-tubulin loading control. (B) Immunofluorescence of same cells showing undetectable NSUN2 reactivity (green) but intact Hoechst (blue) staining in affected cells compared with cells from control and mother. Staining is specific to nuclei and is enriched at nucleoli as previously described. Scale bar 10 uM.

NSUN2 is an RNA methyltransferase known to operate downstream of c-MYC during cell proliferation. It contains a NOL1/NOP2/Sun domain, and was demonstrated to have in vitro activity to methylate cytosine in tRNA and hemimethylated DNA (polydI:dC).11 12 In yeast, the closest orthologue is the gene SUN (Trm4/Ncl1), which is capable of methylating tRNAs at all four known sites (positions 34, 40, 48, 49), and where gene inactivation completely blocks tRNA methylation.13 A recent study using metazoan cells demonstrated a requirement for NSUN2 in cytosine methylation (m5C) of residue 34 in tRNALeu(CAA) and residues 47 and 48 in tRNAAsp(GTC),14 whereas, residue 38 in tRNAAsp(GTC) is methylated by the alternative methyltransferase DNMT2.15 NSUN2-deficient mice lack m5C of tRNALEU(CAA) at position C34. These mice also display a low body weight phenotype and partial alopecia, as well as altered cell cycle timing and self-renewal capacity in epidermal stem cells.16

To test for similar defects in human NSUN2-mutant cells, RNA was isolated from dermal fibroblasts, and processed for bisulfite sequencing as previously described.16 Primers were designed to amplify mature tRNAASP(GTC) (figure 4A, Fw: 5′-GTTAGTATAGTGGTGAGTAT-3′, Rev: 5′-CTCCCCATCAAAAAATCAAAC-3′). Visualisation of bisulfite sequencing data demonstrated the loss of cytosine-5 methylation at these NSUN2 target sites (C47 and C48) in affected individuals but not the unaffected mother (figure 4B). Position C38, the target of DNMT2, was methylated in all individuals. These results suggest that loss of the NSUN2 gene results in a loss of site-specific 5-cytosine methylation in patient cells. DNMT2 and NSUN2 are the only known tRNA:m5C methyltransferases known, and although recently demonstrated to protect tRNA from cleavage and degradation,15 the significance of this post-translational modification is not well understood.

Figure 4

Site-specific loss of cytosine-5 methylation (m5C) in two affected members (Aff #1 and Aff #2). (A) Secondary structure of tRNAAsp. Prediction is that C47 and C48 require NSUN2 for m5C, whereas C38, methylated by DNMT2, will be unaffected. (B) RNA bisulfite sequencing of tRNAAsp(GTC) in the mother as a control and two affected members. In affected members, m5C is specifically lost at NSUN2-dependent sites C47 and C48 (highlighted in bold), whereas, m5C of C38 (highlighted with a star) remains unchanged. Blue indicates cytosine-5 not methylated; yellow indicated cytosine-5 methylated.

NSUN2 has been shown to interact with the mitotic spindle and to play an important role in cell division, and interestingly, its effect on mitotic spindle stability is independent of this methyltransferase activity.17 Thus, it has been hypothesised to play a dual role in mammalian cells. Phosphorylation of NSUN2 by the cell-cycle-related kinase, Aurora, reduces methyltransferase activity,11 further supporting this dual-role hypothesis. Cultured cells depleted of NSUN2 exhibit increased apoptosis, reduced proliferation and a variety of spindle defects leading to abnormal mitoses,17 which might relate to the frequent reports of potential defects in immunological function and chromosomal instability in DS. Our discovery of a damaging mutation in NSUN2 leading to a DS-like condition further highlights the role of cell proliferation genes in normal brain development and emphasises the connection between abnormal cell cycle and microcephaly phenotypes.

Web resources

The URLs for data presented herein are as follows:


The authors thank J Meerloo at the UCSD Neurosciences Core for microscopy services (US National Institute of Neurological Disorders and Stroke P30NS047101), Bruno Reversade for sequencing NSUN2 in additional Dubowitz probands (from the cohort), Broad Institute (U54HG003067 to Eric Lander) for sequencing support and analysis. This work was supported by the US National Institutes of Health R01NS048453, R01NS052455 and P01HD070494 (JGG), the Daland Fellowship from the American Philosophical Society (JHL), the American Heart Association 09POST2250641 (JEL), the Simons Foundation Autism Research Initiative, and the Howard Hughes Medical Institute (JGG).


Supplementary materials


  • FJM and JHL contributed equally to this work.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval Ethics approval was provided by the UCSD Institutional Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.