Genomic sequencing of a dyslexia susceptibility haplotype encompassing ROBO1

Abstract

Background: The DYX5 locus for developmental dyslexia was mapped to chromosome 3 by linkage study of a large Finnish family, and later, roundabout guidance receptor 1 (ROBO1) was implicated as a candidate gene at DYX5 with suppressed expression from the segregating rare haplotype. A functional magnetoencephalographic study of several family members revealed abnormal auditory processing of interaural interaction, supporting a defect in midline crossing of auditory pathways. In the current study, we have characterized genetic variation in the broad ROBO1 gene region in the DYX5-linked family, aiming to identify variants that would increase our understanding of the altered expression of ROBO1.

Methods: We have used a whole genome sequencing strategy on a pooled sample of 19 individuals in combination with two individually sequenced genomes. The discovered genetic variants were annotated and filtered. Subsequently, the most interesting variants were functionally tested using relevant methods, including electrophoretic mobility shift assay (EMSA), luciferase assay, and gene knockdown by lentiviral small hairpin RNA (shRNA) in lymphoblasts.

Results: We found one novel intronic single nucleotide variant (SNV) and three novel intergenic SNVs in the broad region of ROBO1 that were specific to the dyslexia susceptibility haplotype. Functional testing by EMSA did not support the binding of transcription factors to three of the SNVs, but one of the SNVs was bound by the LIM homeobox 2 (LHX2) protein, with increased binding affinity for the non-reference allele. Knockdown of LHX2 in lymphoblast cell lines extracted from subjects from the DYX5-linked family showed decreasing expression of ROBO1, supporting the idea that LHX2 regulates ROBO1 also in human.

Conclusions: The discovered variants may explain the segregation of dyslexia in this family, but the effect appears subtle in the experimental settings. Their impact on the developing human brain remains suggestive based on the association and subtle experimental support.

Keywords: Dyslexia; ROBO1; Whole genome sequencing.

Fig. 1

Pedigree of the DYX5-linked family. Squares denote males and circles females. The 19 affected individuals marked with dots share the dyslexia susceptibility haplotype [11], and their DNA samples were pooled for sequencing on the Illumina platform. The DNA from the two affected individuals denoted by arrows was used in the CGI WGS. Circled dots indicate the individuals whose DNA samples were used in the Sanger sequencing of the exonic SNPs

Fig. 2

ROBO1 expression and phenotype. The results from phonological coding tests were scored according to z-point comparisons to control group mean values, such that the larger the score, the more problems the subject had in phonological coding. The test scoring was 0 (z ≥ −1.0), 1 (−2.0 ≤ z < −1.0), 2 (−3.0 ≤ z < −2.0), or 3 (z < −3.0) [8]. The ROBO1 gene expression was measured by real-time PCR from lymphocytes from the same subjects with higher values indicating higher expression [19]. Blue diamonds denote males and red diamonds females. The plotted values show a tendency for negative correlation, supporting the idea that the less ROBO1 is expressed, the more deficit there is in phonological awareness

Fig. 3

Predicted 8-mer binding sites for LHX2 in the genomic area of SNV4. The reference sequence (shown on the top row) was scanned for 8-mer binding profiles for transcription factors from the UniPROBE database [33]. The rows below the reference sequence show predicted 8-mer binding sites for LHX2, so that the bases that are not included in the 8-mer are shown in grey. The enrichment scores from the protein binding microarray data are shown rightmost on each row. The TAATTA consensus site for homeobox is shown in green, and the reference allele (cytosine) at SNV4 is shown in red. The alternative allele (thymidine) is shown in blue

Fig. 4

EMSA for SNV4. Nuclear extracts from HEK293 cells overexpressing LHX2-V5 or GFP-V5 were used. The arrow shows the location of the bands that showed differences in the amount of probe bound in a protein-DNA complex. A mixture of two antibodies against LHX2 was used to confirm that the protein-DNA complex at the level indicated by the arrow contains LHX2. An unlabeled probe was used to compete the binding of LHX2 to the SNV probes. The T allele seemed to be more resistant than the C allele to the competing probe as seen when comparing the bands above the asterisks

Fig. 5

Luciferase assay for SNV4. The SNV4-containing pGL3-promoter vectors were tested for transcriptional activity in HEK293 cells by using luciferase assays. When combined as a group, the SNV4-containing vectors showed increased luciferase promoter activity when compared to the control empty vector (P < 0.05 on Student’s t test). We did not detect a significant difference between the C and G alleles. The error bars indicate standard deviation

Fig. 6

Correlation between ROBO1 and LHX2 expression. a LHX2 was knocked down by lentiviral shRNA constructs in lymphoblast cell lines extracted from the DYX5-linked family and from control individuals. Expression levels of LHX2 and ROBO1 were measured by quantitative real-time PCR. The fold-change values indicate the difference between LHX2-shRNA and the control scramble shRNA-treated cells. Blue indicates the DYX5-linked family and red indicates controls. The regression lines show that low expression of LHX2 correlates with low expression of ROBO1. b ROBO1 and LHX2 expression in 22 different brain tissues from the FANTOM5 database. The red dots and lines show the co-expression between LHX2 and ROBO1 shorter splice variant b. The blue dots and lines show the co-expression between LHX2 and ROBO1 longer splice variant a