Genetic vulnerability, environmental exposures and neurodevelopmental disorders : clinical insights and in-vitro consequences

Abstract

An understanding of how different genetic backgrounds and environmental exposures interact and contribute to diverse medical conditions like neurodevelopmental disorders (NDDs), is key to better health outcomes. A considerable overlap exists in the underlying genetics and physiology, often leading to their co-diagnoses. There is a lack of robust biomarkers for ASD and ADHD. Furthermore, no efficient resource exists to evaluate potential gene-environment interactions during early human neurodevelopment. This thesis addressed these knowledge gaps through five clinical and in-vitro studies.

Study I explored the utility of genetic information from exome sequencing in predicting intervention outcomes of a social skills group training (SSGT) clinical trial for ASD. A genetic score was developed for common and rare variants in relevant genetic pathways, followed by generating a predictive machine learning (ML) model for individual responses. Variant carriers demonstrated significantly less improvement after standard care at postintervention. A higher rare variant genetic score for synaptic transmission was linked to less efficacy after SSGT at follow-up, while an opposite effect was observed for regulation of transcription from RNA polymerase II. The ML model emphasised the importance of rare variants in predicting intervention outcomes. Study II deployed urine-based untargeted metabolomics to investigate ASD-related biomarkers in a twin cohort with ultra-high performance liquid chromatography and mass spectrometry (UHPLC-MS). For the first time, any associations with autistic traits were also evaluated. No metabolite was found to be significantly associated with ASD. Based on nominal significance, an elevation in phenylpyruvate and taurine, and a decline in carnitine were detected, amongst others. These were found to be enriched in the arginine and proline metabolism pathway. More nominally significant metabolites were associated with autistic traits, and indole-3-acetate was positively associated with autistic traits within twin pairs. Study III also utilised a twin cohort to detect urinary and faecal metabolites associated with ADHD using nuclear magnetic resonance (NMR) and UHPLC-MS, respectively. Males with ADHD had increased levels of urinary hippurate, a metabolite produced by microbial-host co-metabolism. Hippurate was also negatively associated with intelligence quotient (IQ) levels in males and differentially associated with faecal metabolites from the gut microbiome. ADHD faecal profiles were characterised by higher levels of 1-stearoyl-2-linoleoyl-snglycerol (SLG), flavine adenine dinucleotide (FAD) and 3,7-dimethylurate. Reduced levels of aspartate, xanthine, orotate and other metabolites were also detected. Study IV dissected the impact of six environmental factors (lead, valproic acid, bisphenol A, ethanol, fluoxetine and zinc deficiency) in human induced pluripotent stem cell (iPSC) derived neuronal progenitors after differentiation for 5 days using the fractional factorial experimental design (FFED) coupled with RNA-sequencing. This was followed by a stratified analytical approach. Several gene and pathway level changes, that were both convergent and divergent for the environmental factor exposures, were identified. Pathways related to synaptic function and lipid metabolism were significantly elevated by lead and fluoxetine, respectively. Furthermore, fluoxetine increased the levels of several fatty acids when validated with direct infusion electrospray ionisation mass spectrometry (ESI-MS). Study V evaluated the differential in-vitro effects of four commonly prescribed selectively serotonin reuptake inhibitors (SSRIs: fluoxetine, citalopram, sertraline and paroxetine) in iPSC-derived neuronal progenitors. Total reactive oxygen species (ROS) and adenosine triphosphate (ATP) levels were determined at day 5 and 28 of differentiation. Concurrently, untargeted metabolomics was performed using ESI-MS. Sertraline and paroxetine significantly decreased ROS and ATP levels. Sertraline mediated early metabolite changes at day 5, while both sertraline and paroxetine drove such effects at day 28. Combined effects were driven by LPC 18:0 and LPC 16:0. Overall, metabolites were enriched in phospholipid biosynthesis and amino acid metabolism pathways.

In conclusion, this thesis highlighted that genetic information can be used as an indicator for ASD interventions, encouraging further exploration. Urine and faecal metabolites are potential biomarkers for ASD and ADHD, pending validation. A multiplexable resource for studying gene-environment interactions was developed, along with a rich dataset outlining molecular changes in ASD. Lastly, the thesis demonstrated that different SSRIs elicit both shared and unique in-vitro responses, with a need to evaluate probable in-utero effects. The findings can guide future clinical studies to generate greater insights into ASD, ADHD and other conditions with aberrant neurodevelopmental trajectories.