Multi-omics analysis of the molecular response to glucocorticoids - insights into shared genetic risk from psychiatric to medical disorders

Janine Knauer-Arloth,Anastasiia Hryhorzhevska,Elisabeth B Binder
DOI: https://doi.org/10.1016/j.biopsych.2024.10.004
2024-10-09
Abstract:Background: Alterations in the effects of glucocorticoids have been implicated in mediating some of the negative health effects associated with chronic stress, including increased risk for psychiatric disorders as well as cardiovascular and metabolic diseases. This study investigates how genetic variants influence gene expression and DNA methylation (DNAm) in response to glucocorticoid receptor (GR)-activation, and their association with disease risk. Methods: We measured DNAm (n=199) and gene expression (n=297) in peripheral blood before and after GR-activation with dexamethasone, with matched genotype data available for all samples. A comprehensive molecular quantitative trait locus (QTL) analysis was conducted, mapping GR-response methylation (me)QTLs, GR-response expression (e)QTLs, and GR-response expression quantitative trait methylation (eQTM). A multi-level network analysis was employed to map the complex relationships between the transcriptome, epigenome, and genetic variation. Results: We identified 3,772 GR-response meCpGs corresponding to 104,828 local GR-response meQTLs that did not strongly overlap with baseline meQTLs. eQTM and eQTL analyses revealed distinct genetic influences on gene expression and DNAm. Multi-level network analysis uncovered GR-response network trio QTLs, characterized by SNP-CpG-transcript combinations where meQTLs act as both eQTLs and eQTMs. GR-response trio variants were enriched in GWAS for psychiatric, respiratory, autoimmune and cardiovascular diseases and conferred a higher relative heritability per SNP than GR-response meQTL and baseline QTL SNP. Conclusions: Genetic variants modulating the molecular effects of glucocorticoids are associated with psychiatric as well as medical diseases and not uncovered in baseline QTL analyses.
What problem does this paper attempt to address?