LAUSR

Epigenetics reflects complex interactions between genes and the environment and plays a role in long-lasting gene expression changes. Therefore, unraveling epigenetic landscape of brain cells from patients with psychiatric disorders will contribute to understanding the pathophysiology of these disorders. We performed promoter-wide DNA methylation analysis of the prefrontal cortex of patients with schizophrenia (N = 35, Table S1). Brain cells were separated into neuronal and nonneuronal nuclei by NeuN-based nuclei sorting. Methylated DNA was collected using the MBD2B/3L, which does not bind hydroxymethylcytosine and enables the analysis of densely methylated regions in the genome. DNA methylation profiles were obtained with promoter tiling arrays covering 25 500 human promoters. Differentially methylated regions (DMRs) were identified by using a deposited control dataset (N = 35, GSE137921), which had been obtained by the same experimental procedure. This study was approved by the ethics committees of the participating institutes (the Research Ethics Committee of Kumamoto University, the Research Ethics Committee of the Faculty of Medicine of The University of Tokyo, the Ethical Review Board of Juntendo University, and the Wako 1st Research Ethics Committee of RIKEN). This study was conformed to the provisions of the Declaration of Helsinki. The experimental procedures are described in Appendix S1. We identified 91 DMRs and 69 DMR-associated genes in nonneurons and 74 DMRs and 59 DMR-associated genes in neurons (Tables S2 and S3). We found that 59.4% of nonneuronal and 69.5% of neuronal DMR-associated genes were shared between the two cell types (Fig. 1a). The DMR with the most significant change in neurons was located in the CHGA promoter region and showed hypermethylation. This DMR was also identified and showed the most significant change in nonneurons. CHGA encodes chromogranin A, a neuroendocrine protein located in the vesicles of neurons. The level of chromogranin A was reported to be reduced in the cerebrospinal fluid and prefrontal cortex in schizophrenia. The DMR with the second highest score in neurons was found in the LINGO1 promoter region, and showed hypermethylation. This DMR was also identified and showed the fifth highest change in nonneurons. LINGO1 plays a role in myelination and neurite outgrowth, and the disturbance of LINGO1 signaling has been implicated in the pathophysiology of schizophrenia. Gene ontology analysis using both neuronal and nonneuronal DMR-associated genes revealed the enrichment of potassium channelrelated terms (FDR corrected P < 0.05) (Fig. 1b). In addition, enrichment of the Ras/Rho signaling and glucocorticoid receptor signaling pathways was detected (nominal P < 0.05). We previously performed a microarray-based gene expression analysis of the same brain region in the same subjects. By utilizing the dataset we assessed the gene expression status of DMR-associated genes. We found that the expression levels of 25 probes, which covered 19 DMRassociated genes, showed reliable gene expression values. Among them, five probes for three genes (ATP2B2, NCOA2, and PEG10) showed significantly altered expression (Welch’s t-test, nominal P < 0.05) (Fig. 1c). ATP2B2 encodes plasma membrane calcium ATPase isoform 2 and de novo damaging variants in this gene were identified in autism. NCOA2 encodes nuclear receptor coactivator 2 and has a histone acetyltransferase activity. PEG10 encodes paternally imprinted gene 10 and contains two overlapping open reading frames, RF1 and RF2. PEG10 was also identified as a differentially expressed gene in schizophrenia in a large-scale transcriptomics study. The hypermethylated DMRs in schizophrenia are located in the 3’-UTR of PEG10 and may affect the regulation of isoform variations. We then assessed whether the DMRs were enriched in genome-wide association study (GWAS) loci of psychiatric disorders by promoterbased random sampling analysis (Appendix S1). We previously found that neuronal DMRs in bipolar disorder (BD) were significantly enriched in the GWAS loci of BD, but nonneuronal DMRs were depleted from the GWAS loci of schizophrenia. In this analysis, we found no significant enrichment or depletion of the DMRs in the GWAS loci of any psychiatric disorders (P > 0.05). To increase sensitivity, we defined the DMRs with a relaxed threshold (P < 10 ), yielding 525 neuronal and 958 nonneuronal DMRs. Similar to the DMRs in BD, we found significant depletion of the nonneuronal DMRs from the schizophrenia GWAS loci (P = 0.0354), whereas no enrichment or depletion in the neuronal DMRs (Figs 1d and S1). These results suggest that the DMRs, especially the nonneuronal DMRs, and the GWAS loci of schizophrenia may have different spatiotemporal roles in the brain. In summary, we identified DMRs of the prefrontal cortex of patients with schizophrenia. The DMRs reported in this study will be useful for understanding the pathophysiology of schizophrenia.