How genes and environments modify intergenerational risk for depression - using polygenic scores to translate between rodents and humans

Rodent and human studies suggest the hippocampal dentate gyrus (DG) region is crucial for resilience to stress and depression. We showed that maternal stress is associated with offspring DG structural changes in a mouse model of maternal stress and in human offspring. DG microstructure is also associated with susceptibility versus resilience to depression across species. However not all human offspring exposed to maternal stress have decreased DG microstructure and increased depressive symptoms. Using cross-species (functional) genomics, we now investigate whether gene x environment interactions might explain which children exposed to maternal distress will be susceptible and who will be resilient. Specifically, we translate brain region specific gene expression data from our mouse models to novel, biologically informed and region specific expression based polygenic scores (PGS) in the human cohorts to assess whether differences in predicted gene expression can explain DG structural difference and depression development in children at risk for depression. We use two human cohorts, three generation family study (N=306; TGS) and the population based Adolescent Brain and Cognitive Development Study (N=6285; ABCD) to ensure replication of findings. Gene networks associated with maternal stress exposure were extracted from mouse DG RNA sequencing data using weighted gene correlation network analysis. Using GTEx we translated the mouse gene network to a human expression based PGS that predicts over/underexpression of the gene network in the human hippocampus (DG ePGS). In addition, broad depression PGS was calculated from the Howard et al. 2018 summary statistics at different GWAS p-value thresholds (p < 10-5-p < 1). We performed enrichment analysis using Functional Mapping and Annotation of Genome-Wide Association Studies (FUMA_GWAS) on all polygenic scores. For MRI, we extracted DG mean diffusivity with FreeSurfer and MRtrix pipeline. Clinical data on the mother and offspring was assessed. Regressions in a generalized estimating equation framework, accounted for family structure with adjustments for sex, age, population stratification and scanner head motion. For MRI, we extracted DG mean diffusivity with FreeSurfer and MRtrix pipeline. Clinical data on the mother and offspring was assessed. Regressions in a generalized estimating equation framework, accounted for family structure with adjustments for sex, age, population stratification and scanner head motion. Higher DG ePGS was significantly associated with higher rates of depression across cohorts. DG ePGS interacted with maternal stress in the ABCD study, such that only children with high DG ePGS who were exposed to maternal stress have disrupted DG microstructure and higher rates of depression development. We also show that GWAS-based depression PGS predicts DG microstructure and depressive symptoms. Enrichment analysis showed that the genes associated with the polygenic score that best predicted DG structure were most differentially expressed in the hippocampus and are involved in pathways associated with neurogenesis and neurodevelopment. Converging results across mice and two independent human samples suggest that the dentate gyrus is an important predictor of resilience to stress and depression, and it is moderated by both genetic and environmental risk factors. Gene networks associated with the effects of maternal stress are conserved across species and associated with depression. Expression based polygenic scores provide an avenue of using gene expression data from mouse models or deceased patients to predict outcomes in live populations.