Variant in the X-chromosome spliceosomal gene GPKOW causes male-lethal microcephaly with intrauterine growth restriction

Congenital microcephaly, with or without additional developmental defects, is a heterogeneous disorder resulting from impaired brain development during early fetal life. The majority of causative genetic variants identified thus far are inherited in an autosomal recessive manner and impact key cellular pathways such as mitosis, DNA damage response and repair, apoptosis and splicing. Here, we report a novel donor splice site variant in the G-patch domain and KOW motifs (GPKOW) gene (NG_021310.2:g.6126G>A, NM_015698.4:c.331+5G>A) that segregates with affected and carrier status in a multigenerational family with an X-linked perinatal lethal condition characterized by severe microcephaly and intrauterine growth restriction (IUGR). GPKOW is a core member of the spliceosome that has been shown in numerous model organisms and in human cells to be essential for survival. By investigating GPKOW transcripts in lymphoblastoid cell lines (LCLs) of three carrier females, we show that the GPKOW c.331+5G>A variant disrupts normal splicing of its pre-mRNAs. In a clonal culture expressing only the c.331+5G>A allele isolated from one carrier female LCL, we observed an 80% reduction in wild type GPKOW mRNA, 70% reduction in the full length GPKOW protein and the presence of a truncated GPKOW protein with possible dominant negative effect. Based on our and published data we propose that the GPKOW gene is essential for fetal development and when disrupted, leads to a severe, male-lethal phenotype characterised by microcephaly and IUGR.


INTRODUCTION
Congenital microcephaly, a condition where head circumference at birth is significantly below the mean for age and sex, is the result of impaired brain development during early fetal life. 1 While the genetic causes are diverse, causative variants predominantly show an autosomal recessive mode of inheritance and affect fundamental cellular pathways such as the assembly and function of the mitotic apparatus; DNA replication, damage and repair; apoptosis signaling and mRNA splicing. 1,2 Why the developing fetal brain is particularly sensitive to even subtle alterations in pathways core to all living cells is still being investigated. 2, 3 We studied a previously reported multigenerational family with a male lethal condition primarily characterized by severe microcephaly with IUGR and variable presentation of additional developmental defects, 4 with carrier females possibly presenting with a mild phenotype. We identified a rare variant in a canonical donor splice site of the X-linked splicing factor gene GPKOW as the most likely cause of this condition.

MATERIALS AND METHODS
Details of materials and methods can be found in the Supplementary Information.

RESULTS
X-exome sequencing of obligate carrier III-2 ( Figure 1a) and subsequent variant filtering identified five novel heterozygous variants (Supplementary Table 1), of which the GPKOW (NG_021310.2: g.6126G4A; NM_015698.4:c.331+5G4A) and IRS4 (NG_013239.1: g.7189C4T; NM_003604.2:c.2156C4T) variants were considered candidates. However, only the GPKOW c.331+5G4A splice-site variant segregated in nine additional family members, including one affected male fetus (Figure 1a and Supplementary Table 3). The GPKOW c.331+5G4A did not have a damaging CADD score 5 (Supplementary Table 1), however different in silico prediction tools calculated a decrease in splicing efficiency as compared to wild type sequence (Supplementary Table 2). GPKOW has a probability of loss-of-function intolerance Z score of 0.98 in ExAC, 6 suggesting that disruption of this gene is likely to be damaging (Supplementary Table 1).
Extreme skewing (4 90%) of X chromosome inactivation (XCI) was observed in blood-derived genomic DNA from three females; II-8, III-2 and III-4 (Figure 1a), however skewing of XCI ratios did not co-segregate with obligate carrier status or presence of the GPKOW variant. Namely, I-2 and II-2 are obligate carriers and heterozygous for GPKOW c.331+5G4A variant yet have random XCI, whereas III-4, who has not conceived affected children and is homozygous wild type (WT) at this locus, has skewed XCI to the ratio of 97:3. Furthermore, skewed XCI, as assessed on venous blood DNA, does not appear to be fully protective against the effect of the GPKOW variant in carrier females. For example, II-8 has highly skewed XCI yet her height and occipitofrontal circumference (OFC) measurements are in the o1st and 2nd centiles, respectively. Incidentally, we observed a missense variant in exon 1 of the AR gene (NG_009014.2:g.6285T4A, NM_000044.3:c.170T4A, p.(Leu57Gln), rs78686797) that is in linkage disequilibrium with the GPKOW c.331+5G4A variant (Supplementary Figure 1c). This was supported by determining the AR allele inheritance for affected and unaffected male family members (Supplementary Table 3). Subsequently, we used the methylation status of the AR sequence tagged site (NCBI Probe database Pr012386469) that flanks rs78686797 as a proxy to indicate WT or variant GPKOW allele expression. LCLs from carrier females I-2, II-2 and III-2 were all highly skewed toward expression of the WT GPKOW allele (Supplementary Figure 3). XCI measured on DNA from LCLs differed to XCI measured from venous blood DNA for obligate carriers I-2 and II-2 (see Supplementary Figure 3A).
Given the severity of the disorder we did not have access to a viable cell source from an affected male. To investigate whether the GPKOW c.331+5G4A donor splice site variant alters GPKOW pre-mRNA splicing and creates alternative mRNA species harboring premature termination codons (PTC) that are subject to Nonsense Mediated mRNA Decay (NMD), we inhibited NMD in low passage LCLs from three carrier (I-2, II-2 and III-2) and three control females with cycloheximide (CHX). Increased levels of bona fide NMD target GADD45B mRNA confirmed NMD inhibition (Supplementary Figure 2).  Table 3). Foetus II-6 was a missed miscarriage at 25/40, III-3 died in the neonatal period at 36/40, III-5 was stillborn at 34/40, and pregnancies III-7 and IV-1 were terminated at 21/40 and 18.5/40, respectively, following diagnosis by ultrasound. IUGR, intrauterine growth restriction. RT-PCR amplification of GPKOW mRNA using primers located within exons 1 and 5 revealed four aberrant GPKOW transcripts in addition to the full-length transcript (Figures 1b-d). Transcripts A and B were present in all CHX-treated LCLs suggesting they result from normal splicing events unrelated to the c.331+5G4A variant, and as they were largely undetectable in the untreated LCLs, are naturally degraded by NMD. Sanger sequencing revealed transcript A lacks exon To further investigate the c.331+5G4A variant in isolation from sequence-confirmed WT allele expression, we isolated clonal cultures from the LCL of carrier female II-2. From 13 successfully isolated clones, 12 expressed the WT GPKOW allele and one the c.331+5G4A variant allele (Supplementary Figure 3b). Real time quantitative PCR (RT-qPCR) revealed that full length GPKOW levels were reduced by 80% in the c.331+5G4A expressing clone as compared to three independent WT expressing clones and three unrelated controls (Figure 2a). Western blot analysis showed a corresponding 70% reduction in full-length GPKOW protein, with the presence of an additional band at approximately 40 kDa, corresponding to the size predicted for transcript D translation product (Figures 2b and c). Indirect immunofluorescent detection of GPKOW protein in clonal LCL cultures demonstrated significant reduction in GPKOW levels but no discernible effect on its localization in the c.331+5G4A variant cells compared to WT cells and unrelated controls ( Figure 3).

DISCUSSION
That RNA metabolism is fundamental to normal brain development and neuronal function is demonstrated by the complexity of alternative precursor messenger RNA (pre-mRNA) splicing in the human brain. 7 In excess of 350 disorders, including many neurodevelopmental disorders, are consequences of genetic variants affecting pre-mRNA splicing. 8 Furthermore, variants in splicing factors themselves are reported to cause a wide variety of human diseases. [8][9][10] In this work we implicate GPKOW in a male-lethal developmental disorder. GPKOW is a ubiquitously expressed nuclear RNA-binding protein that has been identified as a core member of the spliceosome. 11 GPKOW associates with catalytically active complexes B act and C 11,12 and acts as an essential cofactor for the two DEAH-box helicases, DHX16 and DHX38, which sequentially catalyze the key transesterification reactions. GPKOW has been classified as essential for human cell survival, 13 and there are no known paralogs in the human genome. Moreover, there are no homozygous or hemizygous loss of function variants thus far listed in ExAC. 6 However, DECI-PHER lists a patient (#286121) with a 1.4 Mbp deletion including the GPKOW gene and a 33 Mbp pathogenic deletion of chromosome 21, presenting microcephaly as one of the clinical symptoms. GPKOW is highly conserved and has been shown to be essential for life in numerous model organisms. For example, in Caenorhabditis elegans, RNAi manipulation of its GPKOW ortholog R11A8.2 results in embryonic lethality in more than 80% of cases and the minority that survived had morphological growth defects. 14 RNAi silencing of the GPKOW ortholog CG10324 in Drosophila melanogaster neuroblasts is early lethal and causes under proliferation. 15 To our knowledge, no vertebrate model of a GPKOW knockout has been generated.
The GPKOW c.331+5G4A variant we identified was predicted in silico to adversely impact spliceosomal processing of exon 2 donor splice site (Supplementary Table 2). Although adenine at the +5 position occurs in approximately 9% of human donor splice sites genome-wide, 16 there are numerous reported cases of +5G4A disease-associated variants causing aberrant pre-mRNA processing. 17 Blocking NMD in LCLs from three carrier females revealed the expression of the c.331+5G4A allele is correlated with the production of aberrant transcripts C and D. Current evidence shows that GPKOW is subject to XCI 18 and as such in a female the gene is expressed only from the active X-chromosome. Using clonal WT and c.331+5G4A In the c.331+5G4A allele-expressing LCL clone, we observed fulllength transcript levels were reduced by 80% as compared to expression levels in WT-expressing cultures (Figure 2a). Importantly, we also observed a corresponding 70% reduction in full-length GPKOW protein, and the presence of a truncated GPKOW protein, corresponding to the transcript D translation product (Figures 2b and c). The truncated protein is predicted to lack amino acids 59-189 (Figures 1b and c), spanning the first two-thirds of the G-patch domain, which has been shown to be important for GPKOW binding to DHX16. 12 Consequently, we predict that the truncated protein, even if folded and localized correctly, would be unlikely to function properly as a cofactor in spliceosomal reactions. We also speculate that the truncated GPKOW protein may have a dominant negative effect by competing with residual WT GPKOW for binding sites and/or binding to and sequestering other spliceosomal factors.
Testing of X-chromosome inactivation provided conflicting data. Three females (II-8, III-2 and III-4) of whom two are obligate carriers (II-8 and III-2), had highly skewed XCI, while two other obligate carriers (I-2 and II-2) showed random XCI. However, when we tested XCI on DNA isolated from LCLs of three obligate carriers, two of whom have variable XCI in blood (I-2 and II-2), we found all three to have highly skewed XCI favoring expression of the wild type GPKOW allele (Supplementary Figure 3a). Skewing of XCI also doesn't appear to be protective of the possible mild phenotype observed in the obligate carrier females (Supplementary Table 3). On aggregate, the combined data on XCI in blood and LCL DNA and on multiple obligate carrier and wild type females in this family show a complex, yet to be resolved mechanism of XCI skewing in different tissues. It also appears that a factor unrelated to the GPKOW variant and involved in high XCI skewing, might also be segregating in this family (e.g. wild type female III-4 with high skewing of XCI; Figure 1a).
How significantly reduced levels of WT GPKOW along with any function of the truncated GPKOW lead to the observed male-lethal phenotype remains to be determined. However, there are a number of severe congenital disorders with the shared features of microcephaly and growth retardation that are attributed to genetic variants in splicing factors. For example, mandibulofacial dysostosis, Guion-Almeida type (MFDGA, OMIM #610536) is caused by heterozygous variants in EFTUD2; Verheij syndrome (OMIM #615583) has recently been attributed to heterozygous variants in PUF60; 19 and microcephalic osteodysplastic primordial dwarfism type I (MOPD1, OMIM #210710) caused by recessive variants in RNU4ATAC. Furthermore, the particular sensitivity of males to mutations in X-linked splicing factors is demonstrated with the example TARP syndrome (OMIM #311900), a male-lethal developmental disorder caused by mutations in the X-linked gene RNA Binding Motif Protein 10 (RBM10). 20 Our findings show that a variant in the X-chromosome essential splicing factor GPKOW is the likely cause of a male-lethal disorder featuring microcephaly and IUGR, and add to the growing number spliceosomal gene variants implicated in severe developmental disorders, highlighting the crucial role of pre-mRNA splicing in human development. 8 Further investigations to elucidate the role of GPKOW in normal development are warranted. grants 1041920 and 1091593 to J.G., MS McLeod Research Fellowship to M.C, and by the EU FP7 project GENCODYS, grant number 241995 to VMK. This study makes use of data generated by the DECIPHER Consortium. A full list of centres who contributed to the generation of the data is available from https:// decipher.sanger.ac.uk/ and via email from decipher@sanger.ac.uk. Funding for the project was provided by the Wellcome Trust.