Leveraging blood serotonin as an endophenotype to identify de novo and rare variants involved in autism

Abstract

BackgroundAutism spectrum disorder (ASD) is one of the most highly heritable neuropsychiatric disorders, but underlying molecular mechanisms are still unresolved due to extreme locus heterogeneity. Leveraging meaningful endophenotypes or biomarkers may be an effective strategy to reduce heterogeneity to identify novel ASD genes. Numerous lines of evidence suggest a link between hyperserotonemia, i.e., elevated serotonin (5-hydroxytryptamine or 5-HT) in whole blood, and ASD. However, the genetic determinants of blood 5-HT level and their relationship to ASD are largely unknown.MethodsIn this study, pursuing the hypothesis that de novo variants (DNVs) and rare risk alleles acting in a recessive mode may play an important role in predisposition of hyperserotonemia in people with ASD, we carried out whole exome sequencing (WES) in 116 ASD parent-proband trios with most (107) probands having 5-HT measurements.ResultsCombined with published ASD DNVs, we identified USP15 as having recurrent de novo loss of function mutations and discovered evidence supporting two other known genes with recurrent DNVs (FOXP1 and KDM5B). Genes harboring functional DNVs significantly overlap with functional/disease gene sets known to be involved in ASD etiology, including FMRP targets and synaptic formation and transcriptional regulation genes. We grouped the probands into High-5HT and Normal-5HT groups based on normalized serotonin levels, and used network-based gene set enrichment analysis (NGSEA) to identify novel hyperserotonemia-related ASD genes based on LoF and missense DNVs. We found enrichment in the High-5HT group for a gene network module (DAWN-1) previously implicated in ASD, and this points to the TGF-β pathway and cell junction processes. Through analysis of rare recessively acting variants (RAVs), we also found that rare compound heterozygotes (CHs) in the High-5HT group were enriched for loci in an ASD-associated gene set. Finally, we carried out rare variant group-wise transmission disequilibrium tests (gTDT) and observed significant association of rare variants in genes encoding a subset of the serotonin pathway with ASD.ConclusionsOur study identified USP15 as a novel gene implicated in ASD based on recurrent DNVs. It also demonstrates the potential value of 5-HT as an effective endophenotype for gene discovery in ASD, and the effectiveness of this strategy needs to be further explored in studies of larger sample sizes.Electronic supplementary materialThe online version of this article (doi:10.1186/s13229-017-0130-3) contains supplementary material, which is available to authorized users.

Drosophila Histone Demethylase KDM5 Regulates Social Behavior through Immune Control and Gut Microbiota Maintenance

“…We therefore assessed the serotonin (5-HT) concentration in wt and kdm5 K6801/10424 flies. Similar to the findings in ASD patients (Chen et al, 2017), we observed that the abdomens and gut kdm5 K6801/10424 flies had significantly higher levels of 5-HT compared to wt flies ( Figures 4O and S8L). Interestingly, heads and guts of flies with intestine-specific KDM5 knockdown also showed elevated levels of 5-HT ( Figures 4R and S8M).…”

Differential serotonin transporter (5‐HTT) and 5‐HT2 receptor density in limbic and neocortical areas of adults and children with autism spectrum disorders: implications for selective serotonin reuptake inhibitor efficacy

“…Peripheral hyperserotonemia was the earliest demonstrated neurochemical change in individuals with autism spectrum disorders (ASD) (Schain and Freedman ) and has since become the best replicated biomarker, as recent meta‐analyses conclude that hyperserotonemia is present in 25–45% of the autism population (Gabriele et al ; Chen et al ; Eissa et al ). Furthermore, whole blood maternal serotonin levels during pregnancy are associated with altered cognitive abilities and core neurodevelopmental outcomes in offspring with ASD (Montgomery et al ).…”

Dissecting Autism Genetic Risk Using Single-cell RNA-seq Data

“…We hypothesize that autism risk genes have distinct spatiotemporal expression signatures in developing human brain in neurotypicals. When comparing A-risk to other metrices or methods in prioritizing risk variants, we observed better performance of A-risk in prioritizing candidate risk variants using de novo variant data of 8838 trios from recent publications [6,[20][21][22][23][24]. Furthermore, we showed that A-risk and gene mutation intolerance metrics [25] can be combined to improve prior estimation in an empirical Bayesian model and enables identification of additional novel risk genes.…”

“…To investigate if A-risk can prioritize de novo risk variants detected from exome or genome sequencing studies, we compiled de novo likely gene-disrupting (LGD) variants of 8838 trios from recent published studies [6,[20][21][22][23][24] ( Supplementary Table 5). We calculated enrichment rate of LGD de novo variants in a gene set by the observed number of variants divided by the expected number estimated from background mutation rate models [35,36] ( Table 1).…”

“…To this end, we stratified a total of 18663 protein-coding genes by A-risk score 0.4 and pLI cutoff 0.9, resulting in 4 quadrants of genes (Supplementary Figure 5A): 1195 constrained genes with high A-risk score (quadrant A), 1842 constrained genes with low A-risk score (quadrant B), 1444 non-constrained genes with high A-risk score (quadrant C) and 14182 non-constrained genes with low A-risk score (quadrant D); then we estimated prior parameters by extTADA in each quadrant of genes, using previously reported de novo LGD and D-mis variant data from 8838 trios [6,[20][21][22][23][24]. Consistent with previous simulation [6], in unstratified analysis, π is about 0.04, corresponding to 750 risk genes in total.…”