DNA methylation is a metastable epigenetic mark that reflects genetic and environmental influences on health and disease. As the epigenome is established during the perinatal period, environmental exposures experienced during this time have the greatest impact on the methylome, as well as resulting changes in gene expression and phenotype. This unique ability of DNA methylation to reflect both etiology and phenotypes of disease makes it an attractive clinical biomarker, particularly if assayed from perinatal tissues such as placenta and newborn blood. In this dissertation, we assay DNA methylation of perinatal tissues to gain etiological insights and identify potential biomarkers of congenital heart defects (CHDs) in individuals with Down syndrome (DS) and of autism spectrum disorders (ASD).
DS is a genetic disorder caused by trisomy 21 that results in developmental and intellectual delays. CHDs affect approximately 50% of individuals with DS, though the molecular reasons behind this incomplete penetrance is unknown. We investigated an epigenomic signature of DS-CHD using whole genome bisulfite sequencing (WGBS) in newborn blood from DS patients with and without CHDs. We identified sex-specific differentially methylated regions (DMRs), which were enriched for terms related to cardiac and immune functions. Machine learning algorithms selected 19 DMRs identified in males that could distinguish CHD from non-CHD. This study showed a sex-specific signature of DS-CHD and showed that the newborn blood methylome can reflect the variability in phenotypes in DS.
ASD is a group of neurodevelopmental disorders that arise from a poorly understood combination of genetic factors, environmental factors, and gene-environment interactions. To improve our understanding of the combinatorial effects of potential risk and/or protective factors for ASD, we identified co-methylation networks from WGBS of placental and newborn blood samples and identified modules that were significantly associated with ASD diagnosis in the child. In both tissues, these ASD co-methylation modules mapped to genes previously implicated in neurodevelopment, including CSMD1, AUTS2, and GALC from placenta, and RANBP17, TLX3, PTPRJ and TRIM49B from newborn blood. These ASD modules also associated with potential risk factors for ASD, including polychlorinated biphenyl levels in maternal serum and maternal hypertension (placental modules) and grandparental cigarette smoking, alcohol use, and advanced age when their child was born (newborn blood modules). These network analyses showed that dysregulated DNA methylation at particular loci may provide a biological connection between ASD diagnosis and multivariate factors in the parents and grandparents. We also identified a pan-tissue epigenomic signature of ASD through DMR analysis of newborn blood, placenta, cord blood, and post-mortem cortex from ASD and TD individuals. We found that the epigenomic signature of ASD was stronger in females than males, with DMRs from all four tissues showing greater overlap in females and enrichment for gene ontology terms related to neurodevelopment. In females, one or more DMRs from all four tissues mapped to BCOR, GALNT9, and OPCML, making these interesting targets for future ASD studies.
The findings in this dissertation increase our understanding of the etiological basis of DS-CHD and ASD and provide evidence for the utility of perinatal methylation signatures in the development of clinical biomarkers. This work benefitted from access to perinatal tissue samples and robust participant data from multiple cohorts; methylation sequencing over the entire genome; and bioinformatic tools for DMR and co-methylation network analysis. Overall, these studies help us move towards earlier interventions for patients with DS-CHD or ASD by improving our understanding of risk factors and biomarkers for these conditions.
