The use of spoken and written language is a capacity that is unique to humans. Individual differences in reading- and language-related skills are influenced by genetic variation, with twin-based heritability estimates of 30-80%, depending on the trait. The relevant genetic architecture is complex, heterogeneous, and multifactorial, and yet to be investigated with well-powered studies. Here, we present a multicohort genome-wide association study (GWAS) of five traits assessed individually using psychometric measures: word reading, nonword reading, spelling, phoneme awareness, and nonword repetition, with total sample sizes ranging from 13,633 to 33,959 participants aged 5-26 years (12,411 to 27,180 for those with European ancestry, defined by principal component analyses). We identified a genome-wide significant association with word reading (rs11208009, p=1.098 × 10−8) independent of known loci associated with intelligence or educational attainment. All five reading-/language-related traits had robust SNP-heritability estimates (0.13–0.26), and genetic correlations between them were modest to high. Using genomic structural equation modelling, we found evidence for a shared genetic factor explaining the majority of variation in word and nonword reading, spelling, and phoneme awareness, which only partially overlapped with genetic variation contributing to nonword repetition, intelligence and educational attainment. A multivariate GWAS was performed to jointly analyse word and nonword reading, spelling, and phoneme awareness, maximizing power for follow-up investigation. Genetic correlation analysis of multivariate GWAS results with neuroimaging traits identified association with cortical surface area of the banks of the left superior temporal sulcus, a brain region with known links to processing of spoken and written language. Analysis of evolutionary annotations on the lineage that led to modern humans showed enriched heritability in regions depleted of Neanderthal variants. Together, these results provide new avenues for deciphering the biological underpinnings of these uniquely human traits.

Despite high prevalence and high heritability, few candidate genes have been identified for reading disability. To identify novel genetic variants we performed a genome-wide association study (GWAS) using high-depth whole genome sequencing and predicated on reading performance in 407 subjects enrolled in a longitudinal study of response-to-intervention, called the New Haven Lexinome Project. The primary GWAS identified a single peak of 31 SNPs on chromosome 19 that achieved the threshold for genome-wide significance (rs2599553 P=3.13×10−8) located over an expression quantitative trait locus (eQTL) for GARRE1 (Granule Associated Rac And RHOG Effector 1). Little is known about the function of GARRE1, except that it is highly and developmentally expressed in human cerebellum relative to cortex. Local ancestry regression showed the strongest association for the lead variant in African or Admixed American populations, who have been under-represented in previous genetic studies of reading. We replicated our chromosome 19 results in the Genes, Reading, and Dyslexia (GRaD) cohort and found a moderating effect of age with implications for the consideration of developmental effects in the design of future analyses. Growth curve modeling demonstrated that minor alleles of the lead SNP are related to reading longitudinally from Grade 1 to Grade 5, and that children with at least 1 minor allele of rs2599553 persistently underperformed relative to their peers by 0.33 to 0.5 standard deviations on standardized assessments of non-word decoding and reading fluency.Significance StatementTo the best of our knowledge, this work represents the only GWAS predicated on longitudinal reading performance data. Starting with initial discovery, we replicate our association in a second cohort, address common causes of type I error, localize the signal to a single gene, implicate a region of the brain most likely to be affected by variation in our candidate, show a gene-by-age effect with implications for study design in this field, and demonstrate that minor alleles of our lead SNP are associated with significant and persistent clinical effects on reading development in children.

Recent studies of co-occurring reading disability (RD) and attention deficit/hyperactivity disorder (ADHD), and co-occurring RD and language learning disability (LLD), support a core disability plus co-occurrence model focused on language and attention. Genetic factors have been associated with poor reading performance. However, little is known about whether different genetic variants independently contribute to RD co-occurrence subtypes. We aimed to identify subgroups of struggling readers using a latent profile analysis (LPA) in a sample of 1,432Hispanic American and African American youth. RD classes were then tested for association with variants of READ1, a regulatory element within the candidate RD risk gene, DCDC2. Six groups were identified in the LPA using RD designation as a known-class variable. The three RD classes identified groups of subjects with neurocognitive profiles representing RD+ADHD, specific phonological deficit RD, and RD+LLD. Genetic associations across RD subtypes were investigated against functional groupings of READ1. The RU1-1 group of READ1 alleles was associated with RD cases that were marked by deficits in both processing speed and attention (RD + ADHD). The DCDC2 microdeletion that encompasses READ1 was associated with RD cases showing a phonological deficit RD profile. These findings provide evidence for differential genetic contribution to RD subtypes, and that previously implicated genetic variants for RD may share an underlying genetic architecture across population groups for reading disability. Effect of READ1 on latent profiles of reading disability and comorbid attention and language learning disability subtypes