Investigation of the genetic basis of schizophrenia and cognitive function

Abstract

Schizophrenia (SZ) is a chronic psychiatric disorder affecting approximately 1% of the population. Symptoms range from delusional thoughts and hallucinations to anhedonia and social withdrawal. Affected individuals also display cognitive deficits in IQ, memory and attention. My first study focused on epigenetic mechanisms, which are a heritable means of regulating various genomic functions to orchestrate processes such as brain development. These processes, when perturbed, are thought to contribute to SZ pathophysiology. I investigated these mechanisms using data from genome wide association studies (GWAS) which have identified 108 chromosomal regions associated with risk of SZ.

In my first study, I identified risk genes for SZ that function in epigenetic processes. Risk variants in 8 candidate genes (BCL11B, CHD7, EP300, EPC2, GATAD2A, KDM3B, RERE and SATB2) were analysed using a dataset of Irish psychosis cases and controls (n=1,235), to test for effects on IQ, memory and attention. Five of the eight genes were found to be associated with at least one cognitive task. Variants with strong associations were tested in 2 independent samples from Wales (n= 772) and Germany (n=920) but failed to replicate.

Following on from this, I focused on three genes identified in the first study (SATB2, BCL11B and GATAD2A) that regulate stages of neurodevelopment via epigenetic mechanisms. SATB2 mediates the projection of neurons across the cerebral hemispheres by regulating the activity of BCL11B via the NuRD nucleosome remodelling complex. GATAD2A encodes a component of the complex. Given the risk signals at these genes, genes within the NuRD complex or regulated by SATB2 may contribute to SZ etiology. I used ChIP-seq data from Satb2 mouse models via a collaborative project with a lab in the University of Innsbruck, to generate three gene-sets that contain genes either functionally related to or targeted by SATB2 at different developmental stages. Each was tested for enrichment using the largest available GWAS dataset for SZ (36,989 cases, 113,075 controls) and enrichment analysis was performed for SZ and other neurodevelopmental disorders using rare variant data. Based on the genetic overlap between SZ and cognition, gene-sets were also investigated for a genetic contribution to educational attainment (EA) using the largest available GWAS dataset (N=293,723). SATB2 gene-sets were enriched for genes containing common variants associated with both phenotypes, and were enriched for genes containing rare variants for SZ, autism and intellectual disability. For neuron-expressed genes, synapse-related genes, or genes expressed in specific brain regions, gene-sets targeted by SATB2 were found to be more strongly enriched for genes associated with SZ or EA than gene-sets not targeted by SATB2.

Satb2 regulates of synaptic plasticity and long-term memory in the adult cortex. To address whether the differences in SATB2 function at different developmental stages are reflected by differences in the SATB2 protein interactome, the third study, carried out by collaborators in Innsbruck, compared SATB2 interacting proteins in neonatal vs. adult mouse cortex. A proteomic approach was used to characterize SATB2 protein-protein interactions at the two stages. I used this data to test the contribution of the SATB2 interactome to cognition using GWAS datasets for EA and cognitive ability (CA). Proteomic analysis identified 95 SATB2 binding partners that form distinct multiprotein complexes. The SATB2 interactome overlapped with the molecular machinery involved in chromatin loop formation. The adult but not neonatal SATB2 interactome was found to be enriched with common variants associated with EA and CA.

Disruptive and damaging ultra-rare variants (dURVs) in highly constrained (HC) genes influence EA in the general population and are also associated with an increased risk of SZ. In SZ, the enrichment of dURVs is most concentrated in neuronally-expressed genes with synaptic functions. In my final study, I analysed GWAS-identified genes for SZ and EA to determine if, like genes harbouring dURVs for these phenotypes, they show similarities in level of constraint, cell expression, and synaptic function. I performed gene-set enrichment analysis on the largest available GWAS datasets for SZ and EA. Sets of HC genes were found to be enriched for genes associated with SZ and EA in comparison to genes under weak selective pressure. Neuronally-expressed, synaptic genes were most enriched for association signals for SZ and EA.