Linkage Analysis and Fine-Mapping in Autism Pedigrees Confirms and Refines Regions on Chromosomes 7 and 12

Background: Autism is highly heritable with complex genetic architecture. While genetic studies to date have identified more than 100 candidate genes, they have failed to explain all of the underlying genetic risk. Recent studies have been association studies or rare-variant studies looking for de novo variants in small families. These studies do not take advantage of the information that can be gained in linkage analysis of multiplex families. Previous linkage studies, while suggestive, failed to firmly establish linkage regions. Furthermore, recent advances in genotyping technology and growing reference panels from sequencing studies now allow for high-quality imputation of dense genotype data, which should improve our ability to identify linkage and fine-map autism genes using association.

Objectives: We hypothesize that using dense imputation in our sample of large of well-characterized multiplex families with autism, linkage analysis will be a powerful tool to identify genomic risk loci for autism.

Methods: We have ascertained 250 families of European ancestry: 165 families with sibling pairs and 85 with more distant relative pairs. Individuals were genotyped on either the Illumina HumanHap550 or Illumina Human1M genome-wide genotyping platforms. GWAS data were imputed from Haplotype Reference Consortium (HRC version r1.1) panel within the Michigan Imputation Server. Non-parametric multipoint linkage analysis was performed using the MERLIN software, followed by data cleaning (Merlin and Pedwipe), LD pruning (r2=0.16), and manual inspection for potential double recombinants using the Progeny software. Family-based SKAT (famSKAT) was used to test for significantly associated genes within the significant linkage regions.

Results: Non-parametric multipoint analysis identified two regions of significant linkage. The first was chromosome 7q11.23-21.11 (LOD=5.8). This linkage region spans 3Mb with ~40 genes including HSPB1, a gene that responds to environmental stress by translocating proteins, and the YWHAG gene, which is highly conserved and mainly expressed in the brain. The second region of linkage was chromosome 12q14 (LOD=4.5) spanning a 2Mb and 26 genes, including GRIP1, which is important for the function of synapses and previously reported associated with autism. FamSKAT analyses confirmed that several genes within both chromosome 7 and 12 are associated with risk of autism in our families. Both of these regions had been implicated previously as harboring autism risk genes.

Conclusions: We demonstrate the power of linkage analysis with imputed GWAS data in large pedigrees. Our results confirm previously suggestive linkage on chromosomes 12 and 7 with several potential candidate genes confirmed by association studies. Of particular interest is the GRIP1 gene on chromosome 12, previously implicated in autism, and the 7q11.23 region that is deleted in William syndrome and includes several autism candidate genes. The findings of several genes associated with autism, suggests a potential role of rare regulatory and non-regulatory variants accounting for the linkage signal within these regions.