Integration of gene expression and GWAS results supports involvement of calcium signaling in Schizophrenia
Introduction
Studies conducted by the International Schizophrenia Consortium attribute one-third of genetic susceptibility for schizophrenia (estimated to be 80% (Sullivan et al., 2003)) to the collective effect of hundreds of common polygenic variants, each contributing a small effect (Purcell et al., 2009, Gejman et al., 2011).
The results of GWAS (Johnson and O'Donnell, 2009) contain many false positives (Burmeister et al., 2008). Current replicable GWAS results account for only a small percentage of the estimated heritability (Ozomaro et al., 2013) and their systematic biological interpretation is lacking. GWAS results have generated several biological hypotheses, such as involvement of ZNF804A through regulation of gene expression (O'Donovan et al., 2008); of infection, through interaction with genes located in the major histocompatibility complex (MHC) region, and involvement of calcium channels, based on two GWAS-derived genes that encode for calcium channel subunits (Ripke et al., 2013). Focusing on a specific GWAS-based variant might be, however, misleading, as each specific variant may be a false positive, and in any case it confers only a small increase of the risk.
Higher-level interpretation of GWAS results in terms of implicated pathways is hindered by the number of reliable GWAS-derived genes (see (Ripke et al., 2013)), too small to yield robust and statistically meaningful pathway enrichment analysis. Ripke et al. (2013) dealt with this limitation by focusing from the outset on the set of SNPs located in genes encoding calcium channel subunits, and found enrichment of SNPS with small p-values in this set. Jia et al. (2010) performed a broader search and identified enrichment of pathways related to metabolism of glutamate, apoptosis, inflammation and immune system. In O'Dushlaine et al. (2011), the pathway of cell adhesion passed multiple testing correction. In Schizophrenia Working Group of the Psychiatric Genomics (2014), a multi-stage GWAS of up to 36,989 cases and 113,075 controls, 108 loci met genome-wide significance, 83 of which have not been previously reported. Protein-coding variants played a limited role, consistently with the hypothesis that most associated variants detected by GWAS exert their effects through altering gene expression rather than protein structure. Genes encoding voltage-gated calcium channel subunits, CACNA1C, CACNB2 and CACNA1I, were found to be significantly associated, extending previous findings implicating members of this family of proteins in schizophrenia.
Expression profiling studies in schizophrenia suggested changes in functional gene groups such as oligodendrocyte and myelin related genes, metabolism, synaptic transmission, GABAergic and glutamatergic pathways (Roussos et al., 2012, Katsel et al., 2005a, Mirnics et al., 2006). Often there is little agreement between different microarray studies regarding which transcripts are differentially expressed in the disease, reflecting the GWAS-based picture of a multifactorial heterogeneous disease, caused by a combination of many genetic and environmental factors. The reported results show only modest changes in genes' expression levels between cases and controls, calling for a more involved and integrated analysis. A model of a complex combination of genetic and environmental factors is concordant with the clinical heterogeneity of the disease (for example, gender difference in the age of onset and outcome) and with delicate structural brain changes that are found in both schizophrenia patients and their relatives (Cooper et al., 2014).
Here we focused on functional changes by analyzing a large-scale genome-wide gene expression dataset of postmortem brain samples, for the identification of biological pathways and processes through which GWAS-derived genes affect schizophrenia. Integration of data from different platforms potentially increases the biological and statistical reliability of the results, reducing significantly the false positives rate resulting from each of the separate analyses. A short list of genes identified by GWAS as relevant to the disease were used as “seed”, and the expression data of these and related genes were analyzed. Correlations between their expression profiles were calculated; high pairwise correlations suggest shared biological pathways. A cluster of genes with high pairwise correlations of expression was identified, and gene ontology (GO) enrichment analysis was performed on the members of the cluster. Since the genes of the cluster have correlated expression, the chance that they belong to the same pathway increases. Indeed we identified significantly enriched pathways—ion channels and, specifically, calcium channel activity.
Since the enrichment analysis was based on a small number of genes, we sought ways to validate this result further. First, we searched for proteins that interact with members of our cluster. The interaction partners were mostly related to calcium signaling; calcium channel subunits and members of the calmodulin pathway emerged. For further validation we extended the list of genes to be searched for enrichment, assembling all genes with high correlation of expression with the average expression profile of the cluster. Enrichment analysis of this extended list identified many pathways that are known to be associated with schizophrenia, and calcium-related pathways re-emerged with high statistical significance.
Section snippets
Subjects
Human brain samples were obtained from the Brain Bank of the Department of Psychiatry of the Mount Sinai Medical Center (New York, NY)/JJ Peters Veterans Administration Medical Center (Bronx, NY). Dissections were performed blind to diagnosis. All cortical dissections and sample preparation were performed as described previously (Hakak et al., 2001, Katsel et al., 2005a, Katsel et al., 2005b). Brain banking activities were approved by the Institutional Review Board of the Mount Sinai School of
36 GWAS-derived genes.
The recent GWAS analysis of (Schizophrenia Working Group of the Psychiatric Genomics, 2014) identified 108 loci that have potential roles in schizophrenia, with genome-wide significance. 58 of these are associated with a single gene symbol. In order to control for background noise, we applied various quality filters (see Supplementary Methods); 36 genes, listed in Table 3, passed these filters and were expressed at a high enough level in our dataset.
Schizophrenia associated genes exhibit robust and statistically significant high correlations of their expression profiles
Expression values (see Methods) of the 36
Discussion
Altered functionality of a group of genes centered on a specific pathway might have a synergistic effect on the risk for the development of a disease. Thus, pathway analysis is important for the biological interpretation of GWAS results. On the one hand, however, the number of genes reliably identified by GWAS as disease associated is small; on the other, lowering the threshold on GWAS-based identification introduces many false positives. Both situations hinder derivation of robust and
Role of funding source
The study was funded by VA-MIRECC, NIH MH066392 and MH064673 and the Leir Charitable Foundation.
Contributors
L. Hertzberg designed and performed the analysis and wrote the manuscript. E. Domany supervised the computational analysis and interpretation of the results and the writing of the manuscript. V. Haroutunian provided the gene expression and clinical data, and supervised the biological interpretation of the results and the writing of manuscript. P. Kastel contributed to the generation of the expression data and commented on the manuscript. P. Roussos contributed to the generation of the
Conflict of interest
All authors declare that they have no conflict of interest.
Acknowledgments
Special thanks to Prof. Avi Weizman, for sharing his knowledge and helpful discussions. Grateful thanks to Prof. Benjamin Lerer and Prof. John Lisman for helpful discussions, and to Assif Yitzhaky for his help with generating the Figures. We would like to thank the Leir Charitable Foundation for bringing together the two groups, at Weizmann and Mount Sinai, and for its support.
References (33)
- et al.
Multimodal voxel-based meta-analysis of structural and functional magnetic resonance imaging studies in those at elevated genetic risk of developing schizophrenia
Psychiatry Res.
(2014) - et al.
Genetic findings in schizophrenia patients related to alterations in the intracellular Ca-homeostasis
Prog. Neuropsychopharmacol. Biol. Psychiatry
(2010) - et al.
Molecular pharmacology of calmodulin pathways in the cell functions
Cell Calcium
(1992) - et al.
Common variants conferring risk of schizophrenia: a pathway analysis of GWAS data
Schizophr. Res.
(2010) - et al.
Variations in differential gene expression patterns across multiple brain regions in schizophrenia
Schizophr. Res.
(2005) - et al.
Variations in myelin and oligodendrocyte-related gene expression across multiple brain regions in schizophrenia: a gene ontology study
Schizophr. Res.
(2005) Calcium signaling dysfunction in schizophrenia: a unifying approach
Brain Res. Brain Res. Rev.
(2003)- et al.
Critical appraisal of DNA microarrays in psychiatric genomics
Biol. Psychiatry
(2006) - et al.
Non-redundant compendium of human ncRNA genes in GeneCards
Bioinformatics
(2013) - et al.
Controlling the false discovery rate: a practical and powerful approach to multiple testing
J. R. Stat. Soc.
(1995)