Leveraging Hyperserotonemia and Whole Exome Sequencing in Autism Spectrum Disorder Families to Tackle Genetic Heterogeneity

Background:  Autism spectrum disorder (ASD) is characterized by a complex and extremely heterogeneous genetic etiology. Numerous lines of investigation implicate serotonergic dysfunction in ASD. Examples include hyperserotonemia, or significantly elevated levels of platelet serotonin (5-HT) in ~35% of cases, and (2) efficacy of selective serotonin reuptake inhibitors in ameliorating anxiety and irritability related to insistence on sameness (IS) and other rigid-compulsive behaviors (RCBs). Notably, platelet serotonin is itself highly heritable, and correlation of hyperserotonemia with OCD in first-degree relatives has been documented. Common variants individually show small effects on risk, yet in aggregate, common genetic variation contributes significantly to heritability. Both CNV and whole exome sequencing (WES) point to de novo and inherited rare variants (RVs) as a major class of autism risk. 

Objectives:  We hypothesized that examination of WES data in autism families identified as having hyperserotonemic probands will reduce genetic heterogeneity and offer insight into ASD risk, specifically relevant to serotonergic dysfunction.

Methods:  To tease out genetic factors related to RCBs and also hyperserotonemia as a heritable biomarker in ASD, we conducted WES of families selected on the basis of these traits. While most families were parent-child trios, one was a large 4-generation, multiplex family with several affected members across generations. In addition to filtering WES data for de novo mutations (DNMs), partially reported in Neale et al 2012, we also filtered phased family exome data to identify autosomal or X-linked genes harboring inherited compound heterozygous (CH; ‘2-hit’), or X-lined hemizygous (in males) ‘functional’ variants (missense, nonsense, consensus splice site and read-through). A gene-collapsing transmission disequilibrium test (cTDT) was employed to explore potential over-transmission of RVs in conjunction with a Consensus Genotyper for Exome Sequence (CGES) to improve call accuracy and reduce the artifactural bias towards an excess of observed non-transmissions. Finally, various algorithms and multiple databases were used to explore gene/protein network enrichment.

Results:  Numerous loci were identified to harbor ‘functional’ CH genotypes and DNMs. Various analyses indicated an enrichment of genes so identified in pathways related to cell adhesion, extracellular matrix proteins and genes previously implicated in ASD, intellectual disability or other neurological phenotypes. Analysis of the multiplex family revealed twelve functional variants shared across all affected members and obligate (‘unaffected’) carriers. As expected, there is a bias towards larger genes harboring such variants, but pathway enrichment nevertheless suggested biological relevance. TDT analysis initially showed an expected over-representation of genes harboring non-transmitted functional variants, reflecting false negative calls in probands or false positive calls in parents. The application of multiple genotype calling algorithms in the cloud-based CGES dramatically improves genotype call metrics and reduces bias towards overall under-transmission. We are working across the board to integrate functional rare variants and CNVs in all families.

Conclusions:  We have used a hyperserotonemic and RCB-selected subset of ASD to identify potential risk factors in ASD and have found numerous genes harboring ‘functional’ de novo mutations, inherited two-hit, compound heterozygous variants and CNVs that provide clues into specific gene sets and functional pathways of relevance.