De novo non-synonymous TBL1XR1 mutation alters Wnt signaling activity

Here we report de novo non-synonymous single-nucleotide variants (SNVs) by conducting whole exome sequencing of 18 trios consisting of Japanese patients with sporadic schizophrenia and their parents. Among nine SNVs, we explored the functional impact of the de novo mutation in TBL1XR1 [c.30 C > G (p.Phe10Leu)], a gene previously found to be associated with autism spectrum disorder and epilepsy. Protein structural analysis revealed that Phe10Leu mutation may decrease the structural stability of the TBL1XR1 protein. We demonstrate that Phe10Leu mutation alters the interaction of TBL1XR1 with N-CoR and β-catenin, which play critical roles in regulation of Wnt-mediated transcriptional activity. Consistently, TBL1XR1-mediated activation of Wnt signaling was up-regulated by Phe10Leu mutation. These results suggest that a de novo TBL1XR1 point mutation could alter Wnt/β-catenin signaling activity. Further studies are required to clarify the involvement of TBL1XR1 mutations in neuropsychiatric conditions.


Results
De novo SNVs identified in schizophrenia trio samples by exome sequencing. We conducted exome sequencing of 18 trios. On average, we obtained 15.4 GB of raw sequence data per sample, and 96.9% of these data were mapped to the reference genome (hg19). On detection of de novo mutations, we found 82 de novo SNVs in 18 trios. Of these 82 de novo SNVs, 17 were predicted to be non-synonymous mutations. Of these 17 de novo non-synonymous SNVs, we validated nine mutations in eight trios by Sanger sequencing (Table 1). Among these nine de novo non-synonymous SNVs, two mutations, one in the ABCD4 [ATP-binding cassette, sub-family D (ALD), member 4] gene and one in the TBL1XR1 [transducin (beta)-like 1 X-linked receptor 1] gene, were predicted as damaged by all of three software tools ( Table 1). Given that de novo TBL1XR1 point mutations have been found in other neuropsychiatric conditions, including autism spectrum disorder (ASD) and West syndrome 14,15 , we conducted subsequent structural and functional analyses of the observed de novo non-synonymous TBL1XR1 mutation [c.30 C > G (p.Phe10Leu)]. The TBL1XR1 point mutation (p.Phe10Leu) was not observed either in the independent 1,191 patients with schizophrenia nor in the 1,986 non-psychiatric control subjects.
Phe10Leu mutation impairs structural stability of TBL1XR1 protein. We next examined the effect of the TBL1XR1 point mutation (p.Phe10Leu) on the protein structure of TBL1XR1 (Fig. 1a). Surface areas of Phe and Leu amino acids were 175 and 137 Å 2 , respectively (Table 2a). Van der Waals (VdW) volumes of Phe and Leu were 135 and 124 Å 2 , respectively. Thus, we predict that substitution of Leu for Phe may decrease the volume and surface area of the 10th residue of TBL1XR1. TBL1XR1 is composed of two structural domains, the N-terminal domain (NTD) and tryptophan-aspartic acid 40 (WD40) repeat. The Phe10Leu mutation is located within the NTD. To assess the substitution in the context of protein structure, we built structural models of the NTD of the control and Phe10Leu TBL1XR1, and we obtained tetramer models of the NTD (Fig. 1b). In a monomer model of the control NTD, the side chain of Phe10 interacts with those of Tyr13, Arg14, Ile34, Ile39, and Val44 within the monomer (Fig. 1c). In the Phe10Leu model, the region around the 10th residue appears sparser than the control (Fig. 1d). We therefore calculated contact areas of the surrounding residues with the rest of the structures and found that those of the surrounding residues, except for Val44, were decreased in the Phe10Leu mutant (Table 2b). Calculated stability of protein structures was lower in the Phe10Leu NTD model than in the control (Table 2c). It is remarkable that statistics associated with VdW potentials were higher in the Phe10Leu NTD. Collectively, the results suggest that the Phe10Leu substitution of NTD decreases structural stability due to the decreases in the contact area of the 10th residue with the surrounding residues.
Phe10Leu mutation of TBL1XR1 alters Wnt signaling activity. TBL1XR1 is a component of the nuclear receptor corepressor (N-CoR) and silencing mediator of retinoic acid and thyroid hormone receptor (SMRT) protein complex, which regulates transcriptional repression machinery 16 . TBL1XR1 also plays a critical role in recruiting β-catenin to Wnt target gene promoters for transcription activation 17 . To examine whether the Phe10Leu mutation affects TBL1XR1-mediated transcription mechanisms, either wild-type TBL1XR1 or the Phe10Leu mutant (TBL1XR1 Phe10Leu ) was exogenously expressed in 293T cells and co-immunoprecipitated with N-CoR and β-catenin. While the Phe10Leu mutation disturbed the interaction between TBL1XR1 and N-CoR, we observed increased binding of TBL1XR1 to β-catenin (Fig. 2a). Consistently, reduced binding of TBL1XR1 to N-CoR and increased interaction between TBL1XR1 and β-catenin were observed in HT22 cells, mouse hippocampal neuronal cells overexpressing TBL1XR1 Phe10Leu (Fig. 2b). Next, we examined the effect of the Phe10Leu mutation on Wnt/β-catenin-mediated transcriptional activity by using the TOPFlash Wnt reporter assay 18 . Consistent with previous studies 17,19 , overexpression of wild-type TBL1XR1 enhanced Wnt transcriptional activity, which was further up-regulated by Phe10Leu mutation (Fig. 2c). Interestingly, although Phe10Leu mutation decreases the interaction between TBL1XR1 and N-CoR, overexpression of N-CoR suppresses enhancement of Wnt transcriptional activity induced by either wild-type TBL1XR1 or Phe10Leu mutant (Fig. 2c).

Discussion
To the best of our knowledge, this is the first study conducting trio-based exome sequencing using Japanese subjects with sporadic schizophrenia. The observed exome point mutation rate in schizophrenia in the present study was similar to those of previous exome sequencing studies 11,13 . While no genes reported in previous trio-based genetic studies of schizophrenia 7-13 was detected in our cohort, we identified the de novo mutation in TBL1XR1 (p.Phe10Leu), a gene previously found to be associated with neuropsychiatric conditions. O'Roak et al. reported two de novo point mutations (p.Leu282Pro and p.Ile397SerfX19) in sporadic cases of ASD 14,20 . Saitsu and colleagues found a de novo point mutation (p.Gly70Asp) in a patient with West syndrome and three missense variants in this gene (p.Ala116Ser, p.Gly405Glu, and p.Asn407Ser) in patients with epilepsy 15 . Deletions on 3q26.32, encompassing the TBL1XR1 gene, have been reproducibly associated with intellectual disability 21,22 . These results suggest that mutation of the TBL1XR1 gene may contribute to a genetic vulnerability to multiple neurodevelopmental psychiatric conditions. The results we obtained in protein structural and functional analysis suggest potential pathogenic impact of Phe10Leu mutation in TBL1XR1 function. TBL1XR1 is a component of the quaternary corepressor complex composed of N-CoR, SMRT and histone deacetylase 3 (HDAC3), which plays a key role in regulating transcription repression 16,[23][24][25] . Recent studies demonstrated that TBL1XR1 also plays a critical role in recruiting β-catenin to Wnt target gene promoters for transcriptional activation 17 . These results suggest that TBL1XR1 may act as a molecular switch for transcriptional activation and repression, which may be regulated by its posttranslational modification, such as SUMOylation 26 . The results of our protein structural analysis indicate that Phe10Leu substitution in the NTD of TBL1XR1 decreases structural stability of the NTD, which may influence binding to other components of the corepressor complex. In fact, we observed that Phe10Leu substitution decreases interaction between TBL1XR1 and N-CoR, whereas binding of TBL1XR1 with β-catenin is increased, which may explain observed up-regulation of Wnt/β-catenin-mediated transcriptional activity induced by Phe10Leu substitution. Accumulating evidences suggest that altered Wnt signaling may be implicated in etiopathophysiologies of neurodevelopmental psychiatric conditions, such as schizophrenia and ASD 14,27,28 . Up-regulation of Wnt/β-catenin signaling and dysregulated function of NCoR have been documented in multiple types of cancers 29,30 . High expression of TBL1XR1 has been reported to be associated with a poor prognosis of colorectal cancer 31 . Consistently, Wnt/β-catenin signaling and the N-CoR/TBL1XR1 complex play critical roles in cell proliferation and differentiation 17,26,29 . Thus, it would be of interest to investigate how altered Wnt signaling pathway induced by TBL1XR1 mutations may affect cellular processes in brain development.
The ABCD4 point mutation (p.lle344Thr) found in the present study was also predicted as damaged by in silico functional analysis. ABCD4 is an ABC transporter that has been classified as a member of the D subfamily of peroxisomal ABC transporters, and mutations in ABCD4 cause a new inborn error of vitamin B12 (cobalamin) metabolism 32 . Given that decreased levels of cobalamin have been reported in the frontal cortex in schizophrenia and autism compared with controls 33 , the effect of ABCD4 point mutations in the metabolic processes of cobalamin in brains should be investigated in future studies.
In conclusion, we report nine novel de novo non-synonymous SNVs as a result of the whole-exome sequencing of 18 trios. In particular, a de novo TBL1XR1 point mutation could alter Wnt/β-catenin signaling activity, supporting the potential involvement of altered Wnt signaling pathway in neurodevelopmental psychiatric disorders. Further studies are required to clarify the involvement of TBL1XR1 mutations in neuropsychiatric conditions.

Materials and Methods
Sample collection. We recruited 18 trios of patients with schizophrenia and their unaffected parents from the Tokushima University Hospital, Ehime University Hospital, Kochi University Hospital, and Nagasaki University Hospital in Japan. Patients with schizophrenia (1191 in total, 696 men and 495 women, mean age: 59.5 ± 14.5 years) were independently recruited from Tokushima University Hospital in Japan. Schizophrenia was diagnosed according to Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV criteria by at least two expert psychiatrists based on extensive clinical interviews and a review of medical records (Supplementary Table 1). Non-psychiatric control subjects (1986 in total, 833 men and 1153 women, mean age: 38.6 ± 13.3 years) were selected from volunteers recruited from hospital staff, students, and company employees with no     Figure. (c) The TOPFlash Wnt reporter assay showed that overexpression of wild-type TBL1XR1 increased Wnt transcription activity, which was further enhanced by an F10L mutation in TBL1XR1 (*P < 0.05 and **P < 0.01). Overexpression of N-CoR suppresses an increase in Wnt transcriptional activity induced by either wild-type TBL1XR1 or F10L mutant (**P < 0.01). Luciferase activities were determined 48 hours post-transfection and normalized against Renilla values. Bars represent averages of each group in three independent experiments. AU, arbitrary unit. All data are presented as the mean ± s.e.m.
Data processing for identification of de novo SNVs. Raw sequencing data for each individual was mapped to the human reference genome (build hg19) by using the Burrows-Wheeler Aligner (BWA v0.5.9) 34 . BWA-generated SAM files were converted into BAM format, sorted and indexed using SAMtools v.0.1.19 35 , then processed by Picard (v1.90) to mark duplicated reads. The BAM-formatted files were further processed using the Genome Analysis Toolkit (GATK, v2.6-4 or v2.  according to the GATK's best-practice recommendations. In brief, the BAM files were processed with GATK tools (RealignerTargetCreator, IndelRealigner, BaseRecalibrator and PrintReads) to perform local realignment around indels and recalibration of the base quality scores, followed by the data compression with the downstream GATK tool (ReduceReads). When using all of the processed BAM files from 18 trios, multi-sample variant calling with the UnifiedGenotyper tool in GATK (v2.  was done to identify SNV and indel candidates. The resulting Variant Call Format file (VCF, version 4.1) was applied to variant quality score recalibration with the GATK VariantRecalibrator tool. Genomic annotations associated with the variants in the VCF file were added using snpEff (v2.0.5d) with the GRCh37.64 database. According to the GATK's recommendations for variant filtering, the annotated VCF file was used to extract potential true positive variants on the following values: QD < 2.0, MQ < 40.0, FS > 60.0, HaplotypeScore > 13.0, MQRankSum < −12.5, and ReadPosRankSum < −8.0. After the multi-sample VCF file was split into each of the 18 trios using GATK, the trio-based VCF files generated were used to obtain variants that violated Mendel's law of segregation in the respective families as candidate de novo variants. Of these variants, we extracted de novo SNV candidates on the following settings: 1) read depth of coverage at SNV sites were 30 or more in both the proband and parents from a trio of interest, 2) SNVs were not present in the dbSNP v137 database and 3) SNVs were seen in only the proband of interest. Plasmids and antibodies. The FLAG-tagged TBL1XR1 expression construct was a gift from Dr. Cun-Yu Wang (University of California, Los Angeles) 17 . The FLAG-tagged mutant TBL1XR1 (TBL1XR1 Phe10Leu ) expression construct was made by a PCR-based mutagenesis protocol 43 . For biochemical experiments, the following antibodies were used: rabbit polyclonal anti-β-catenin antibody (Sigma, Beverly, MA, USA), rabbit polyclonal anti-N-CoR antibody (EMD Millipore, Billerica, MA, USA), as well as mouse monoclonal and rabbit polyclonal anti-FLAG antibodies (Sigma, Beverly, MA, USA).