Identification of a DLG3 stop mutation in the MRX20 family

Abstract

Here, we identified the causal mutation in the MRX20 family, one of the larger X-linked pedigrees that have been described in which no gene had been identified up till now. In 1995, the putative disease gene had been mapped to the pericentromeric region on the X chromosome, but no follow-up studies were performed. Here, whole exome sequencing (WES) on two affected and one unaffected family member revealed the c.195del/p.(Thr66ProfsTer55) mutation in the DLG3 gene (NM_021120.4) that segregated with the affected individuals in the family. DLG3 mutations have been consequently associated with intellectual disability and are a plausible explanation for the clinical abnormalities observed in this family. In addition, we identified two other variants co-segregating with the phenotype: a stop gain mutation in SSX1 (c.358G>T/p.(Glu120Ter)) (NM_001278691.2) and a nonsynonymous SNV in USP27X (c.56 A>G/p.(Gln19Arg)) (NM_001145073.3). RNA sequencing revealed 14 differentially expressed genes (p value < 0.1) in 7 affected males compared to 4 unaffected males of the family, including four genes known to be associated with neurological disorders. Thus, in this paper we identified the c.195del/p.(Thr66ProfsTer55) mutation in the DLG3 gene (NM_021120.4) as likely responsible for the phenotype observed in the MRX20 family.

Introduction

According to the fifth edition of the “Diagnostic and statistical manual of mental disorders (DSM-5)”, intellectual disability (ID) is a neurodevelopmental disorder characterized by deficits in cognition and adaptive function with an onset during the developmental period [1]. It is estimated that 1–3% of the population is affected, with a male to female ratio of 1.6:1 [2]. The sex difference in frequency is commonly attributed to the excess of ID genes on the X chromosome [3]. Although the X chromosome covers only 5% of the human genome, it contains 15% of the genes currently known to be associated with ID [3]. Before whole exome sequencing (WES) facilitated the analysis of all genes in the genome, families with X-linked inheritance were prioritized for disease-gene identification studies because an X-linked pattern of inheritance facilitated the identification of affected families. Subsequent linkage analysis narrowed down the region of interest to a specific region of the X chromosome, further reducing the number of potential candidate genes. A classification system was put in place for families with a LOD-score above 2, indicating significant linkage to the X chromosome, and such families were numbered in sequential order of discovery as MRX (for non-syndromic) or MRXS (for syndromic) families [3–5]. This distinction between syndromic and non-syndromic, in retrospect, has often been arbitrary since even despite a careful clinical evaluation, the syndromic features common to all members of any family can be difficult to recognize and may be age-dependent [4]. In total, 105 families with MRX received MRX numbers, after which this tradition was no longer continued. For 67 of these families, a causative gene has been reported, significantly aided by large-scale initiatives such as Euro-MRX and GenCodys [3]. X-linked families thus contributed significantly to the discovery of MRX genes, of which the total number is estimated to be 141 according to the latest update [3].

Here, we studied the MRX20 family, a large pedigree in which initial linkage studies mapped the putative disease gene to a 55.6 cM interval in the pericentromeric region of the X chromosome, between the short tandem repeat polymorphism markers DXS1068 (Xp11.4-p21, hg19: chrX:38908118-38908368) and DXS454 (Xq21.1-q23, hg19: chrX:97986121-97986265) [6]. This family presents with ID but no obvious other clinical manifestations were found. In this study, we were able to identify a causal mutation in the Discs Large MAGUK Scaffold Protein 3 (DLG3) gene. Further, we found two other variants in the linked region, which may or may not play an additional role in the disease manifestation in this family.

Methods

Collection of patient data

V.3 was referred to AL and MKM for genetic counseling regarding a family history of ID. Her brother, V.2, was examined by clinical geneticist Cheryl S. Reid, M.D. At 21 years of age he presented as an affable young man with obesity, a mildly gynecoid habitus, mild micrognathia, and dysarthric speech but had no physical findings suggestive of a specific syndrome. Available medical records showed a birth weight of 2.3 kg and an initial referral for neurologic evaluation only at an age of 5 years and 8 months for learning difficulties and hyperactivity.

Family history revealed three maternal uncles of the proband reported to be similarly affected, as were three cousins of these uncles, thus likely three obligate carrier females (II.3, III.3, and III.6). The limited medical records still extant for IV.2 and IV.4, were accessed and reported below and in the original publication [6]. Both wards of the state and 50-year residents of group homes, they appeared more severely debilitated than V.2 who lived at home nurtured by a supportive family where he had managed to hold supervised employment.

Mindful of the power presented by the extensive pedigree to identify a causative gene, relevant blood samples for DNA analysis were deposited at the Human Genetic Mutant Cell Repository at the Coriell Institute (Camden, New Jersey, USA) with informed consent.

Post-mortem cerebellar tissue of an unaffected 9-year-old child was obtained from the institute Born-Bunge vzw IBB NeuroBioBank with approval of the Ethics Committee of the Antwerp University Hospital.

Cell culture

Epstein-Barr virus transformed lymphoblastoid cell lines of 19 family members were obtained from Coriell Institute: III.1 (Coriell 400718), III.4 (Coriell 400719), III.5 (Coriell 400717), IV.1 (Coriell 400706), IV.2 (Coriell 400705), IV.3 (Coriell 400703), IV.4 (Coriell 400708), IV.5 (Coriell 400707), IV.6 (Coriell 400709), IV.7 (Coriell 400715), IV.8 (Coriell 400716), IV.9 (Coriell 400710), IV.10 (Coriell 400711), IV.11 (Coriell 400704), IV.12 (Coriell 400712), IV.13 (Coriell 400701), V.1 (Coriell 400714), V.2 (Coriell 400700), V.3 (Coriell 400713). All cell lines were cultured in RPMI (Life Technologies, Carlsbad, California, USA), supplemented with 15% fetal bovine serum (Life Technologies), 1% penicillin/streptomycin (Life Technologies), 1% sodium pyruvate (Life Technologies), and 1% GlutaMAX (Life Technologies).

DNA extraction and WES

Genomic DNA was extracted using the DNeasy® Blood & Tissue Kit (Qiagen, Hilden, Germany) following manufacturer’s instructions. WES was executed on DNA of two affected members (IV.3 and V.2) and one unaffected member (III.3) by BGI (Copenhagen, Denmark). Sequencing was performed using TruSeq DNA sample preparation (Illumina, San Diego, California, USA) and SureSelect Human All Exon V5 kit (Agilent, Santa Clara, California, USA) according to the standard protocols. Sequencing was performed on an Illumina HiSeq 4000 using a 2 × 150 bp sequencing run. Data analysis was done using an in-house pipeline as described before [7]. Data filtering and annotation of variants in the linkage interval were performed with VariantDB [7, 8].

Sanger sequencing validation

Sanger sequencing was performed with primers listed in Supplementary Table 1.

X-inactivation experiments

An X-inactivation assay was performed on the genomic DNA of the following female family members: III.1, III.5, IV.1, IV.6, IV.12, IV.13, and V.3 based on the protocol described by Jones et al., 2014 [9]. X-skewing was determined by fragment analysis of the AR gene and the RP2 gene. Fragments were analyzed on an ABI3130XL (Applied Biosystems, Waltham, Massachusetts, USA) in the presence of an internal sizing standard (ROX). Amplicon sizes were determined using GeneMarker v2.6.4 (SoftGenetics, State College, Pennsylvania, USA). Calculation of the X-inactivation ratio was performed using the areas under the allele peaks with or without HpaII cleavage. The ratio of X-inactivation is interpreted as follows: <80:20 is random; 80:20 to 90:10 is moderately skewed; >90:10 is highly skewed.

DLG3 expression analysis

DLG3 mRNA levels were quantified in lymphoblastoid cell lines using Real-Time PCR (RT-PCR) as described below. Protein expression was evaluated according to the methods as previously published [10]. Briefly, 20 µg of protein was separated using SDS-PAGE and transblotted to a nitrocellulose membrane which was incubated with N-terminal SAP102 (Invitrogen, Waltham, Massachusetts, USA; PA5-51626, 1/1000 dilution) and C-terminal SAP102 (Abcam, Cambridge, UK; ab288436, 1/1000 dilution) primary antibodies.

RNA extraction and sequencing

RNA was extracted using the Quick-RNA™ Miniprep Kit (Zymo Research, Irvine, California, USA) following manufacturer’s instructions. RNA sequencing was performed on the following affected males; IV.2, IV.3, IV.4, IV.9, IV.10, IV.11, V.2, and unaffected males; IV.5, IV.7, IV.8, V.1, of the family. Fragment analysis was performed using the RNA kit (DNF-471), standard sense RNA analysis kit (15nt) of Agilent and RNA sequencing was performed using the QuantSeq 3′ mRNA-Seq Library Prep Kit FWD for Illumina (Lexogen, Vienna, Austria) following manufacturer’s instructions. The RNA-seq data was analyzed by the trimming and cleaning with bbduk [11], the alignment with STAR [12], and the feature extraction with Subread featureCounts [13]. Differential expression analysis for the protein coding genes was conducted using DESeq2 [14] in R and Benjamini-Hochberg adjusted p-values controlling for false discovery rate at 10%.

Pathway enrichment analysis

Pathway analysis was performed using the Web-based Gene set analysis toolkit WebGestalt with the GSEA method, and KEGG database [15].

Identification of enriched transcription factors among differentially expressed genes

Enriched transcription factors among the fourteen differentially expressed genes (p value < 0.1) observed through RNA sequencing were searched for with the plugin IRegulon v1.3 in Cytoscape [16]. Following settings were used: motif collection of 10 K [9,713 position weight matrices (PWMs)], track collection of 1,120 ChiP-seq tracks, the putative regulatory region of 20 kb centered around transcription start site (TSS), motif ranking database 20 kb centered around TSS (seven species), and track ranking database of 20 kb centered around TSS (ChiP-seq-derived). In addition, we used a normalized enrichment score (NES) threshold of 5.0, a ROC threshold for AUC calculation of 0.03, and a rank threshold of 5000. For transcription factor (TF) prediction, the maximum False Discovery Rate on the motif similarity threshold was 0.001. To strengthen the link between the targets of the top ranked enriched transcription factor, we used the TFlink database, that provides comprehensive information on transcription factors and their targets [17].

RT-PCR validation

Real-time PCR (RT-PCR) was used to examine the differential expression. 1 µg of total RNA was converted to cDNA using the Superscript III First-Strand Synthesis System (Invitrogen). The primer design was performed using an in-house automated pipeline [18] with primers listed in Supplementary Table 1. Quantitative PCR was performed using the qPCR Mastermix Plus for SYBR Green I – no ROX (Eurogentec, Seraing, Belgium) following manufacturer’s instructions on a CFX384 Real-time system (Bio-Rad, Hercules, California, USA). Statistical analysis was performed using the qBASE+ software (CellCarta, Montreal, Canada). The data was normalized to ACTB, UBC, and YWHAZ, and the stability of these reference genes was checked with the qBASE+ software. Statistical analysis was performed in GraphPad Prism 9.0 using a two-tailed Mann-Whitney U test.

Results

Patient data

ID was observed in seven males across two generations in a family originating from the USA (Fig. 1) [6]. More extensive clinical information was retrieved from individuals IV.2, IV.3, and V.2 of which relevant abnormalities are briefly summarized here.

Fig. 1. Pedigree of the MRX20 family.

Pedigree is shown as in original publication [6]. Segregation of variants in DLG3 (c.195del/p.(Thr66ProfsTer55)) (NM_021120.4), SSX1 (c.358G>T/p.(Glu120Ter)) (NM_001278691.2) and USP27X (c.56A>G/p.(Gln19Arg)) (NM_001145073.3) are shown for all family members of which DNA was available as well as X-inactivation patterns of seven women of the family. The ratio of X chromosome inactivation for the AR and RP2 alleles were interpreted as follows: <80:20 = random (R); 80:20 to 90:10 = moderately skewed (MS); >90:10 = highly skewed (HS). NI Non-informative.

V.2 demonstrated mild ID, learning difficulties and hyperactivity. There were no complications during pregnancy. His psychomotor development was delayed as he did not walk until 19 months of age. At 9 years of age, impairment in both gross and fine motor function was observed as well as microcephaly (10th percentile). At 18 months of age, alternating exotropia was observed, which was confirmed at a check at 9 years of age where he exhibited with 10° of exotropia. During consultation at 21 years of age, no exotropia could be observed. No facial dysmorphia was observed apart from mild micrognathia.

IV.2 presented with severe ID and a mental age of 4 years and 11 months (IQ ratio 27 on Stanford-Binet intelligence Scale) at age 55.

IV.3 presented with moderate ID. At 46 years of age he scored 51 for Verbal IQ, 52 for Performance IQ and 48 for full scale IQ at the Wechsler Adult Intelligent Scale (WAIS). At 54 years of age, he scored 36 on Stanford-Binet intelligence Scale and his mental age was 5 years and 9 months.

Individuals IV.4, IV.9, IV.10, and IV.11 were diagnosed with ID without further specifications.

Gene identification/mutation identification

A single base pair mutation in the DLG3 gene was identified using WES; c.195del/p.(Thr66ProfsTer55) (NM_021120.4). Segregation of this stop mutation was confirmed in all affected individuals of this family and obligate carrier females using Sanger sequencing (Figs. 1, 2). The mutation was not identified in the unaffected relatives. Loss-of-function (LoF) mutations in DLG3 have consistently been associated with ID and several other families had been described [19–26]. Beyond the published mutations, ClinVar reports 17 additional pathogenic or likely pathogenic LoF or splice site variants. We conclude that this DLG3 mutation on its own may explain the clinical presentation of this family.

Fig. 2. DNA sequencing chromatogram of the variants found in DLG3, SSX1, and USP27X in the MRX20 family.

The reverse complement sequence is shown. An example of the wild-type allele, an alternative allele, which refers to a mutant allele for DLG3, but to a non-reference allele for SSX1 and USP27X, and a heterozygous combination of both alleles in a carrier female are represented. The chromosomal positions (GRCh38) of these variants are: NC_000023.11:g.70445396del, NC_000023.11:g.48263809G>T and NC_000023.11:g.49880363A>G, respectively.

In addition, we found a LoF variant (stop gain), c.358G>T/p.(Glu120Ter) (NM_001278691.2), in the SSX1 gene with a CADD-Phred 1.4 score of 33.0 and a variant of unknown significance, c.56A>G (nonsynonymous SNV)/p.Gln19Arg (NM_001145073.3), in the USP27X gene with a CADD-Phred 1.4 score of 20.2, both segregating with the disease in the same manner as the DLG3 mutation (Figs. 1, 2).

X-inactivation assay

Skewed X chromosome inactivation occurs frequently in families with X-linked ID and has been previously observed in a family with a deleterious DLG3 mutation [26, 27]. Here, moderately skewed X-inactivation was observed in the only available obligate carrier based upon two independent markers. Apart from the obligate carrier, we determined the skewing pattern for six additional females in the family, of which four presented with a nonrandom X-inactivation pattern (Fig. 1).

DLG3 expression

DLG3 encodes the SAP102 protein, a post-synaptic density protein. There are approximately ten different transcripts reported originating from the DLG3 gene (Ensemble and GTEx), of which four are protein coding [28, 29]. Two of these transcripts, ENST00000374360.8 and ENST00000194900.8, are translated to a protein of 90–93 kDa and are predominantly brain-specific. These isoforms contain 19 and 21 exons, respectively. The shorter transcripts, ENST00000374355.8 and ENST00000542398.1, contain 14 and 12 exons and are translated to proteins of 58 kDa an 42 kDa, respectively. These isoforms are more widely expressed throughout different human tissues. However, these transcripts lack exon 1 and hence the described mutation. We investigated DLG3 expression at the mRNA and protein level to compare the expression of several isoforms in lymphoblastoid cell lines of our family and an unaffected human brain, as a positive control. At the RNA level, we could detect expression of both the small and large transcripts in the control brain. However, in the lymphoblastoid cell lines only low quantities of the shorter transcript and potentially even lower quantities of the larger transcript could be detected (Supplementary Fig. 1A). This is in line with the data available on GTEx [29]. At the protein level, we observed the canonical SAP102 isoform in the human brain, with both an N-terminal (detecting the large isoforms, Supplementary Fig. 1B) and C-terminal antibody (detecting both the large and small isoforms, Supplementary Fig. 1C). However, we were not able to detect this larger isoform in lymphoblastoid cell lines derived from both affected and unaffected family members of the MRX20 family. In addition, a smaller SAP102 isoform (~42 kDa), detected by the C-terminal antibody, showed modest expression in the control human brain, but was too low to detect in lymphoblastoid cell lines (Supplementary Fig. 1C).

Differential expression analysis

To unravel the impact of the WES-identified variants, we conducted 3′ mRNA sequencing and performed differential expression analysis on lymphoblaistoid cell lines derived from the males of this family. Here, we identified 14 genes with an adjusted p value lower than 0.1, of which 10 showed an adjusted p value lower than 0.05 (Fig. 3A). In the affected family members WWTR1, HLA-DRA, HLA-DPA1, LDHA, CDCA4 and PPP1R16B were upregulated, in contrast to TMEM51, BCL11A, TFE3, NMT2, FRY, DNAJC5, SELENOW and PEX26 which were downregulated. From these differentially expressed genes, four are known transcription factors (WWTR1 [30],CDCA4 [31], BCL11A [32], TFE3 [33]) and four are associated with neurological disorders (BCL11A [34], TFE3 [35], FRY [36], and DNAJC5 [37]). Through pathway analysis using WebGestalt [15], we found a significant enrichment of the “hematopoietic cell lineage” (Fig. 3B, Supplementary Table 2).

Fig. 3. Differential expression and enriched pathway analysis.

A Differential expression analysis of the MRX20 family. Hierarchical clustering heatmap showing the differential expressed genes; all genes with an adjusted p value < 0.1 are represented. * Represents the differentially expressed genes with an adjusted p value < 0.05. B Enriched pathways based on the differential expression data. Bar chart shows enrichment ratio or NES of results with direction. Enriched pathways with a False Discovery Rate (FDR) ≤ 0.05 are shown in a darker shade. Analysis was performed using WebGestalt [14]. C Validations of differential expression of genes CDCA4, WWTR1, PEX26, LDHA, and NMT2 using RT-PCR normalized to ACTB, UBC, and YWHAZ in the MRX20 family. Statistical significance was obtained by a two-tailed Mann-Whitney U test. * represents a p value < 0.05 and ** represents a p value < 0.01.

Subsequently, we confirmed a differential expression pattern using RT-PCR for the following five genes using the strict criteria of a two-tailed Mann-Whitney U test; CDCA4, WWTR1, PEX26, LDHA, NMT2 (Fig. 3C). Differential expression of genes BCL11A, PPP1R16B were found to be borderline significant with a p value < 0.1 and >0.05 (Supplementary Fig. 2).

Identification of enriched transcription factors among differentially expressed genes

The iRegulon prediction tool was used to determine enriched transcription factors of the fourteen differentially expressed genes (p value < 0.1) [16]. We reported all transcription factors with multiple target genes amongst our differentially expressed genes and with a statistically significant normalized enrichment score (NES) above five (Supplementary Table 3). TFLink was used to provide extra evidence for the link between the most enriched transcription factor and its target genes [17]. Here, the four targets, namely WWTR1, PEX26, BCL11A, and TFE3, which were enriched for the transcription factor SRF, were found in additional databases compared to the original (Supplementary Table 4).

Discussion

The gene found to be causal for ID in this family is the DLG3 gene on the X chromosome (For an overview of all known published causal variations/mutations in DLG3; see Supplementary Table 5 and Fig. 4). The canonical form of this gene contains 19 exons and its encoded protein, synapse-associated protein 102 (SAP102), is the major member of the membrane-associated guanylate kinase (MAGUK) family expressed in neurons during the early brain development [28, 38]. MAGUKs are known to be central building blocks for the postsynaptic density, linking surface-expressed receptors to an intracellular signaling molecule [39]. The SAP102 protein was first described by Müller et al., who stated that this SAP102 protein contains three tandem PDZ domains, an scr homology (SH3) domain and a guanylate kinase (GK) domain [28]. This widely expressed protein is found in dendrites as well as axons in the cytoplasm and postsynaptic density [38]. A knockout mouse model of DLG3 revealed the importance of the SAP102 protein for NMDA receptor-driven plasticity, behavior, and signal transduction [40]. The mutant mice revealed cognitive deficits with a specific spatial learning deficit, which could be overcome by additional training. Typically, MAGUKs are thought of as stabilizing synaptic proteins, but in contrast to others, the SAP102 is also known to play a role in clearing NMDARs from the synaptic site [41].

Fig. 4. Representation of the synapse-associate protein 102 (SAP102) encoded by the Discs Large MAGUK Scaffold Protein 3 (DLG3) gene.

The protein contains three tandem PDZ domains, an scr homology (SH3) domain and a guanylate kinase (GK) domain. Known protein coding mutations are indicated by arrows (NM_021120.4/ NP_066943.2) (Supplementary Table 5 gives a representation of all published mutations in the DLG3 gene).

In the last decade, mutations in the DLG3 gene were identified in several patients diagnosed with mild to severe ID [19–26]. Including the mutation described here, a total of ten stop mutations and two splice donor site mutations have been reported. A representation of all mutations published to date can be found in Fig. 4 and Supplementary Table 5.

As stated before, the DLG3 gene gives rise to approximately ten different transcripts (Ensemble and GTEx), of which two (ENST00000374360.8 and ENST00000194900.8) are translated to the larger isoforms of the SAP102 protein and two (ENST00000374355.8 and ENST00000542398.1) are translated to shorter isoforms. The two splicing regions are called I1 and I2 and are situated respectively on the N-terminus and between the SH3 and GK domains of SAP102 [28, 42]. The larger isoforms, including the canonical form, are mainly expressed in brain tissue, whereas the shorter isoforms are more widely expressed throughout the body (GTEX) [28, 29]. The isoforms share their C-terminal sequence but the N-terminal sequence is unique to the larger isoforms. The variant found in this study in DLG3 is positioned at exon 1 and thus is solely harbored by the larger brain-specific isoforms. We determined the expression of DLG3 both at RNA and at protein level in the Epstein-Barr virus-transformed lymphoblastoid cell lines originating from the MRX20 family members, as well as a control brain sample (Supplementary Fig. 1). Of these experiments we can conclude that Epstein-Barr virus-transformed lymphoblastoid cell lines express the shorter isoform and may express some mRNA of the larger isoform of DLG3, but lack detectable levels of the SAP102 protein. Our results are in contrast with the data presented by Kumar et al. that showed expression of the SAP102 protein in lymphoblastoid cell lines on western blot [19]. In an attempt to reproduce the protein detection, we repeated the western blot experiments using the extraction protocol as described in their paper (data not shown), but nevertheless, no protein band at the right size could be observed. We have no explanation for this apparent discrepancy in protein detection between the two studies, except that we used a different antibody as compared to the abovementioned study. Unfortunately, no brain tissue of this family is available to replicate this study on disease-relevant tissue.

Nevertheless, using differential transcriptome analysis on lymphoblastoid cell lines, we found 14 genes which were differentially expressed in the affected family members (p-value < 0.1), thereby identifying the “hematopoietic cell lineage”-pathway as significantly enriched. Enriched transcription factors for the differently expressed genes include the Serum Response Factor (SRF), an important known transcription factor in the brain [43]. The link between SRF and its involved targets was strengthened by the findings in an additional database (TFLink), which relies on chromatin immunoprecipitation assay data [17]. Furthermore, we should take into account that the transcriptomic analysis was performed on RNA extracted from Epstein-Barr virus-transformed lymphoblastoid cell lines, which can affect cellular gene expression profiles and activities of cellular pathways [44]. Of note, while the clinical presentation of the affected individuals is most likely a result of the DLG3 mutation, the observed differential expression observed might be influenced by two other variants co-segregating in this family.

The stop gain variant in the SSX1 gene; c.358G>T/p.(Glu120Ter), is unlikely causative on its own, as multiple hemizygous LoF variants in non-neurological controls are present in gnomAD. Exceptionally, gnomAD states that more LoF and missense variants then expected are present in control populations (pLI = 0, Z_missense = −4.37). SSX1 is a primate-specific gene that is mainly expressed in the brain and testis [45]. It competes with SMARCB1 for nucleosome acidic patch binding. SMARCB1 is a subunit of mSWI/SNF complex just as BCL11A which we found to be downregulated in this family [46]. This complex is a chromatin remodeling complex which plays an important role in neurological disorders [47]. Remarkably, haploinsufficiency of the BCL11A gene causes Logan-Dias syndrome [34]. Logan-Dias syndrome was discovered by Dias et al. and is known to be an intellectual developmental disorder with the persistence of fetal hemoglobin (HbF). Unfortunately, family members were not available for the testing of HbF persistence. A second variant co-segregating with the disease is a nonsynonymous variant in the ubiquitin-specific protease 27X (USP27X) gene; c.56A>G/p.(Gln19Arg). USP27X is reported as a candidate gene for ID by Hu et al. [48]. However, independent confirmation has not been reported. In addition, USP27X is important in the maintenance of neural stem/progenitor cells by regulating HES1 [49] and aberrant expression of USP27X resulted in reduced neuronal differentiation. However, the effect of the missense mutation with CADD 1.4 score of 20,2 in our family is difficult to predict.

In this study, we have identified the causal mutation c.195del/p.(Thr66ProfsTer55) for the ID phenotype in the MRX20-family in the DLG3 gene. Further, differential expression analysis revealed 14 significantly differentially expressed genes between affected and unaffected males in this family (p value < 0.1). The differential expression pattern might be influenced by two other variants which are co-segregating with the phenotype in this family in the genes, SSX1 and USP27X.

Author contributions

Clinical examination and counseling of the family were performed by AL and MKM. JH was responsible for the conceptualization and overview of the experiments under the supervision of RFK and GV. Primers were developed by JH and EE. Experiments were executed by JH, EE, KEVR, and BC. GV analyzed the WES data. Differential expression was analyzed by LM. Pathway enrichment analysis and identification of enriched transcription factors were performed by JH and LM. Analysis of RT-PCR data and preparation of the corresponding figures were performed by JH and CPD. Western Blotting was performed by CPD. JH drafted the manuscript, which was reviewed and approved by all authors.

Funding

The authors acknowledge the support of the Research Fund of the University of Antwerp OEC-Methusalem grant “GENOMED”.

Data availability

The datasets generated during the current study are not publicly available due consent restrictions, but are available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.

Ethical approval

The cell lines used in this study were donated with informed consent to the Human Genetic Mutant Cell Repository at the Coriell Institute (Camden, New Jersey, USA) for research purposes.

Supplementary information

The online version contains supplementary material available at 10.1038/s41431-024-01537-7.