Joint methylome- and genome-wide association studies in blood and brain identifies new disease mechanisms for schizophrenia. E. J. C. G. Van den Oord1, A. Shabalin1, G. Kumar1, S. Clark1, J. L. McClay1, L. Y. Xie1, R. Chan1, S. wedish Schizophrenia Consort.1,3,4, V. Vladimirov1,2, C. Hultman3, P. F. Sullivan3,4, P. K. E. Magnusson3, K. A. Aberg1 1) Center Biomarker Research and Personalized Med, Virginia Commonwealth Univ, Richmond, VA; 2) Virginia Institute for Psychiatric and Behavioral Genetics, Virginia Commonwealth University, Richmond ,VA, USA; 3) Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden; 4) Department of Genetics, University of North Carolina at Chapel Hill, NC, USA.

   We performed joint analyses of data from methylome- (MWAS) and genome-wide association studies (GWAS) to identify schizophrenia (SZ) disease mechanisms. Methylation measurements for 28 million CpGs were obtained with methyl-binding domain enrichment followed by sequencing (MBD-seq) in post-mortem brain tissue (prefrontal cortex, N=75) and whole blood (N=1,459) of SZ patients and controls. Of the 67 million reads/sample, 47% were used after alignment and QC. After 1000 genomes imputation and selection on imputation quality and MAF > 0.05, 5 million SNPs were available. A variety of mechanisms were tested such as effects CpGs created or destroyed by SNPs (CpG-SNPs) or whether effects of cis/trans methylation quantitative trait loci (meQTLs) were altered in cases versus controls. Analyses were performed with a specifically designed analysis pipeline that was bundled in an ultra-fast and memory efficient software called RaMWAS. Critical findings were replicated using targeted (bisulfite) pyrosequencing in independent SZ case-control samples from post-mortem brain tissue (N=50) and blood (N=400-1,100). Permutation tests showed partial but significant overlap between MWAS and GWAS top findings with two SZ GWAS meta-analyses (N=32,143 and 21,953; p < 0.031 and 710-4). Overlapping findings frequently involved CpG-SNPs. One example is a CpG-SNP in interleukin receptor, IL1RAP, which was associated with the same direction of effects in both blood and brain and replicated in independent samples (p < 1.610-4 N=368). The majority (~68%) of CpG-SNPs were likely methylated with 94% of these sites being methylated in both brain and blood. Because CpG-SNPs show individual variation in both sequence and methylation and may have similar methylation statuses across multiple tissues, we also test how relevant there sites are for other diseases. Using the NHGRI GWAS catalogue finding for all disease, we observed substantial enrichment (odds ratio = 3). Other overlapping SZ MWAS and GWAS findings implicated transcription factor binding sites where both genotype and methylation status could potentially inhibit the binding of transcription factor to their recognition elements (e.g. CREB1, replication p < 110-10, N=1,086). The MWAS findings that did not overlap with GWAS tended to reflect environmental insults (e.g. hypoxia and inflammation). In summary, our joint analyses identified replicating sites that implicated specific hypotheses about SZ disease mechanisms.

You may contact the first author (during and after the meeting) at