Next-generation DNA sequencing identifies novel gene variants and pathways involved in specific language impairment

Abstract

A significant proportion of children have unexplained problems acquiring proficient linguistic skills despite adequate intelligence and opportunity. Developmental language disorders are highly heritable with substantial societal impact. Molecular studies have begun to identify candidate loci, but much of the underlying genetic architecture remains undetermined. We performed whole-exome sequencing of 43 unrelated probands affected by severe specific language impairment, followed by independent validations with Sanger sequencing, and analyses of segregation patterns in parents and siblings, to shed new light on aetiology. By first focusing on a pre-defined set of known candidates from the literature, we identified potentially pathogenic variants in genes already implicated in diverse language-related syndromes, including ERC1, GRIN2A, and SRPX2. Complementary analyses suggested novel putative candidates carrying validated variants which were predicted to have functional effects, such as OXR1, SCN9A and KMT2D. We also searched for potential "multiple-hit" cases; one proband carried a rare AUTS2 variant in combination with a rare inherited haplotype affecting STARD9, while another carried a novel nonsynonymous variant in SEMA6D together with a rare stop-gain in SYNPR. On broadening scope to all rare and novel variants throughout the exomes, we identified biological themes that were enriched for such variants, including microtubule transport and cytoskeletal regulation.

Figure 1. Variants of putative significance in candidate genes.

(a) ERC1, Proband 23. Chr12:1137072, NM_178039:exon2:c.G3A:p.M1I (start-loss). Both parents report history of speech and language problems. All children have special educational needs. (b) GRIN2A, Proband 4. Chr16:9916226, rs77705198, NM_001134407:exon10:c.G2063C:p.G688A (de novo). Mother reports history of speech and language problems (although both parents have low NWR scores). Proband has special educational needs. (c) SRPX2, Proband 41. ChrX:99922289, rs121918363. NM_014467:exon9:c.A980G:p.N327S. Parents do not report history of speech and language problems. All children have special educational needs. Proband is denoted by arrow. Individuals carrying variant allele are denoted by a plus symbol. Affected individuals are shaded black, unaffected are white, unknown are grey. Parents are always shaded as unknown as the language tests employed were for children only. Self-reported family history is given in text. Additional genotypic and phenotypic information is presented in inset table. Variant alleles are shown in bold. Affection status for all children was defined as CELF-R receptive (RLS) or expressive (ELS) language score >1.5 SD below mean (see Methods for details). We also present information regarding nonword repetition ability (NWR) and verbal and non-verbal IQ (VIQ and PIQ respectively). Although these additional scores were not used to ascertain affection status, they can provide useful information regarding specific deficits in individuals. NWR is thought to provide an index of phonological short term memory, while the IQ measures indicate a general level of verbal and non-verbal ability. All measures are standardized with a mean of 100 and a SD of 15. Scores >1.5 SD below the mean are shown in bold.

Figure 2. Co-segregating stop-gain variants.

(a) OXR1, Proband 29. Chr8:107738486, rs145739822, NM_001198534:exon1:c.G15A:p.W5X. Father reports history of speech and language problems. No DNA sample was available for father. Proband and sibling 2 have special educational needs. Sibling 1 does not have language or IQ scores available, but has been diagnosed with dyslexia. (b) MUC6, Proband 8. Chr11:1027390, rs200217410, NM_005961:exon17:c.C2109A:p.C703X. Mother reports history of speech and language problems. Proband has special educational needs. For key for symbols used in this figure, please refer to Fig. 1.

Figure 3. Probands with multiple hits of putative interest.

(a) Proband 19. Rare AUTS2 variant, stop and rare variant in OR52B2, rare variants in KIAA0586 and STARD9. Parents do not report history of speech and language problems. No sample available for father. All children have special educational needs. (b) Proband 12. Multiple rare variants in KMT2D and STARD9. No family history available but maternal NWR score in normal range. No sample available for father. (c) Proband 30. SYNPR rare stop variant and SEMA6D novel nonsynonymous variant. Parents do not report history of speech and language problems (although mother has low NWR score). Proband has special educational needs. For key for symbols used in this figure, please refer to Fig. 1.

Figure 4. Clusters of significant GO terms enriched with variants of different frequency.

Enriched GO terms were identified using three gene lists marked by variant frequency (novel, less than 1%, and between 1–5%). The resulting GO terms associated with the three gene lists are colour-coded (Cyan: between 1–5%; Gold: less than 1%; Red: novel) and with size representing the number of genes within each GO term. The GO terms were clustered based on their functional similarity. Five major functional categories could be identified, namely “Extracellular Matrix Disassembly”, “Cell Proliferation in Forebrain”, “Microtubule-based Movement”, “Release of Sequestered Calcium ion into Cytosol”, and “Cellular response to interleukin-4”. Lines connecting the GO terms indicate levels of similarity between each connected pair.