Meta-analysis of gene coexpression networks in the post-mortem prefrontal cortex of patients with schizophrenia and unaffected controls
1 Canadian Institute of Health Research/Michael Smith Foundation for Health Research (CIHR/MSFHR) Graduate Program in Bioinformatics, University of British Columbia, Vancouver, BC Canada
2 Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory, One Bungtown Road, NY 11724, USA
3 Department of Psychiatry, University of British Columbia, Vancouver BC, Canada
4 Centre for High-throughput Biology, 177 Michael Smith Laboratories, 2185 East Mall, University of British Columbia, Vancouver BC, Canada
BMC Neuroscience 2013, 14:105 doi:10.1186/1471-2202-14-105Published: 26 September 2013
Gene expression profiling of the postmortem human brain is part of the effort to understand the neuropathological underpinnings of schizophrenia. Existing microarray studies have identified a large number of genes as candidates, but efforts to generate an integrated view of molecular and cellular changes underlying the illness are few. Here, we have applied a novel approach to combining coexpression data across seven postmortem human brain studies of schizophrenia.
We generated separate coexpression networks for the control and schizophrenia prefrontal cortex and found that differences in global network properties were small. We analyzed gene coexpression relationships of previously identified differentially expressed ‘schizophrenia genes’. Evaluation of network properties revealed differences for the up- and down-regulated ‘schizophrenia genes’, with clustering coefficient displaying particularly interesting trends. We identified modules of coexpressed genes in each network and characterized them according to disease association and cell type specificity. Functional enrichment analysis of modules in each network revealed that genes with altered expression in schizophrenia associate with modules representing biological processes such as oxidative phosphorylation, myelination, synaptic transmission and immune function. Although a immune-function enriched module was found in both networks, many of the genes in the modules were different. Specifically, a decrease in clustering of immune activation genes in the schizophrenia network was coupled with the loss of various astrocyte marker genes and the schizophrenia candidate genes.
Our novel network-based approach for evaluating gene coexpression provides results that converge with existing evidence from genetic and genomic studies to support an immunological link to the pathophysiology of schizophrenia.