Allele-specific non-CG DNA methylation marks domains of active chromatin in female mouse brain Christopher L. Keowna, Joel B. Berletchb, Rosa Castanonc, Joseph R. Neryc, Christine M. Distecheb,d, Joseph R. Eckerc,e, and Eran A. Mukamela,1 aDepartment of Cognitive Science, University of California, San Diego, CA 92037; bDepartment of Pathology, University of Washington, Seattle, WA 98195; cGenomic Analysis Laboratory, The Salk Institute for Biological Studies, La Jolla, CA 92037; dDepartment of Medicine, University of Washington, Seattle, WA 98195; and eHoward Hughes Medical Institute, The Salk Institute for Biological Studies, La Jolla, CA 92037 Edited by Paul D. Soloway, Cornell University, Ithaca, NY, and accepted by Editorial Board Member Andrew G. Clark December 20, 2016 (received for review July 21, 2016) DNA methylation at gene promoters in a CG context is associated correlated with high or low levels of DNA methylation at CG with transcriptional repression, including at genes silenced on the dinucleotides (mCG) in promoter regions, respectively (10). How- inactive X chromosome in females. Non-CG methylation (mCH) is a ever, different epigenetic profiles may be associated with XCI and distinct feature of the neuronal epigenome that is differentially escape from XCI in the brain because the DNA methylation distributed between males and females on the X chromosome. landscape of neurons is distinct from other cell types. In particular, However, little is known about differences in mCH on the active neurons accumulate methylation at millions of genomic cytosines in (Xa) and inactive (Xi) X chromosomes because stochastic CA and CT dinucleotides during postnatal brain development be- X-chromosome inactivation (XCI) confounds allele-specific epige- ginning at 1 wk of age in mice (6, 11). This non-CG methylation nomic profiling. We used whole-genome bisulfite sequencing in a correlates with reduced gene expression and inactivation of distal mouse model with nonrandom XCI to examine allele-specific DNA regulatory elements in a highly cell type-specific manner (12). Al- methylation in frontal cortex. Xi was largely devoid of mCH, though the functional relevance of non-CG methylation whereas Xa contained abundant mCH similar to the male X (mCH) is unclear, it is bound by the transcriptional repressor chromosome and the autosomes. In contrast to the repressive methyl-CpG binding protein 2 (MeCP2) as neurons mature, association of DNA methylation at CG dinucleotides (mCG), mCH and is enriched at genes that are up-regulated in Rett syn- accumulates on Xi in domains with transcriptional activity, includ- drome (13, 14). ing the bodies of most genes that escape XCI and at the Mosaic XCI prevents discrimination of methylation on the X-inactivation center, validating this epigenetic mark as a signature active X chromosome (Xa) and Xi alleles by conventional of transcriptional activity. Escape genes showing CH hypermethy- methylome profiling. We reasoned that understanding the allele- lation were the only genes with CG-hypomethylated promoters on specific distribution of neuronal mCH in the context of X in- Xi, a well-known mark of active transcription. Finally, we found activation and imprinting could yield new insights into this extensive allele-specific mCH and mCG at autosomal imprinted unique aspect of the brain epigenome. Therefore, we profiled regions, some with a negative correlation between methylation allele-specific DNA methylation, as well as transcription, in in the two contexts, further supporting their distinct functions. mouse frontal cortex using a Xist mouse mutant hybrid in which Our findings show that neuronal mCH functions independently the paternal allele was deterministically inactivated in all cells of mCG and is a highly dynamic epigenomic correlate of allele- (8). To assign sequencing reads to alleles, we used female specific gene regulation. DNA methylation | X-chromosome inactivation | imprinting | Non-CG | Bcor Significance n diploid mammals, the equivalence of the two parental alleles Mammalian cells contain two copies of the genome inherited Iis violated by allele-specific epigenetic regulation in a small, from the two parents. Although most genes are expressed using but critical, subset of the genome. Genomic imprinting, or par- both, a small but critical part of the genome has different levels of ent-of-origin–dependent gene regulation (1), is critical for em- expression from each copy. These parts include the X chromosome in females and imprinted genes in both genders, which play key bryonic development and plays a role in neuronal differentiation roles in brain development and cognition. We measured gene (2). In females, epigenetic inactivation of one X chromosome expression and DNA methylation, an epigenetic modification of silences transcription of most genes to equalize gene expression the genome, in the brains of mice using a technique that allowed with males (3). Both imprinting and X-chromosome inactivation us to analyze the maternal and paternal copies of the genome (XCI) are critical to healthy brain development (4, 5). Despite separately. Our findings show that a brain-specific form of DNA the importance of allele-specific gene regulation in the brain, the methylation called non-CG methylation marks regions of active epigenetic mechanisms controlling these patterns are not com- transcription within the inactive X chromosome. pletely known, in part, due to the challenge of allele-specific epi- genomic profiling. In particular, DNA methylation patterns can Author contributions: C.L.K., C.M.D., J.R.E., and E.A.M. designed research; C.L.K., J.B.B., reflect allelic asymmetries in autosomal gene regulation (6), but R.C., J.R.N., C.M.D., and E.A.M. performed research; C.L.K., J.B.B., C.M.D., and E.A.M. contributed new reagents/analytic tools; C.L.K. and E.A.M. analyzed data; and C.L.K. their correlation with XCI has not been fully addressed. and E.A.M. wrote the paper. XCI has unique advantages as a case study for the investiga- The authors declare no conflict of interest. tion of allele-specific epigenomic regulation. The inactivated allele This article is a PNAS Direct Submission. P.D.S. is a Guest Editor invited by the Editorial is selected stochastically during early development and maintained Board. through subsequent cell divisions (7), yielding a mosaic pattern Freely available online through the PNAS open access option. of allelic expression in adult female tissues. Despite extensive Data deposition: The data reported in this paper have been deposited in the Gene Ex- inactivation of one X chromosome, some genes escape silencing pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE83993). ∼ and are expressed from the inactive X chromosome (Xi): 3% of 1To whom correspondence should be addressed. Email: [email protected]. X-linked genes in mice (8) and 15% in humans (9). Analysis of This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. peripheral blood showed that XCI and escape from XCI are 1073/pnas.1611905114/-/DCSupplemental. E2882–E2890 | PNAS | Published online March 20, 2017 www.pnas.org/cgi/doi/10.1073/pnas.1611905114 Downloaded by guest on September 30, 2021 PNAS PLUS Iqsec2 Kdm5c A B Iqsec2 chrX: 152,069,000 - 152,335,000 Kantr Global mCG at promoters 25KB 75% C57/Spretus SNPs R1 69.7 69.0 Combined 50% R2 %mCG R1 25% 41.1 39.6 Xi 33.5 31.6 R2 31.4 mRNA 29.1 0 R1 Xa R2 Global mCG outside of promoters R1 2 1 75% Combined R2 85.2 85.6 85.5 85.3 84.9 84.9 75.6 75.7 R1 50% Xi %mCG R2 25% Xi hypo DMRs Xa hypo 0 mCG Methylation R1 Global mCH Xa 1.4% R2 Male X 1.26 1.24 1.23 1.0% 1.22 1.01 1.01 R1 Combined %mCH R2 0.4% R1 0.020 0 0.020 Xi R2R1 R2R1 R2R1 R2R1 R2 Maternal Paternal Xa Xi R1 Autosomes mCH Methylation Xa R2 Male X X Inactivation Center C Tsix Jpx Mir374c Cdx4 Chic1 Tsx Gm9159 Zcchc13 Xist Ftx chrX: 103,302,000 - 103,658,000 Mir374 25KB C57/Spretus SNPs R1 Combined R2 R1 Xi R2 mRNA R1 Xa R2 R1 1 23 Combined R2 R1 Xi R2 Xi Hypo DMRs Xa Hypo mCG Methylation R1 Xa R2 GENETICS Male X R1 Combined R2 R1 Xi R2 R1 Xa mCH Methylation R2 Male X Fig. 1. Ultrasparse mCH on Xi correlates with escape domains. (A) Allele-specific mCG and mCH levels on autosomes and chromosome X. Browser view of methylation and expression for the Kdm5c locus (B) and the XIC (C). Ticks show the methylation level at individual cytosine positions (CG, green; CH, blue) on the forward (upward ticks) and reverse (downward ticks) strands. Combined tracks show both alleles, whereas the Xa and Xi tracks include only reads sorted using SNPs between C57 and spretus. Monoallelically expressed genes (Iqsec2 and Chic1) and intergenic regions harbor mCH on Xa only, whereas diallelically expressed escape genes (Kdm5c) and the Xi-expressed noncoding RNA Xist contain dense mCH on Xi. Male X data are from 6-wk-old frontal cortex (11). chrX, chromosome X; R1, replicate 1; R2, replicate 2. Keown et al. PNAS | Published online March 20, 2017 | E2883 Downloaded by guest on September 30, 2021 F1 mice from crosses between C57BL/6 Xist mutant and Mus mCH on Xi presents an opposite (positive) correlation with tran- spretus wild-type mice (8), and analyzed species-specific genetic scription: Xi is remarkably void of mCH except in the gene body of variants (∼42 million single-nucleotide polymorphisms (SNPs), the escape gene, Kdm5c, where mCH is enriched (3.09%) and including 1.95 million SNPs on the X chromosome). Our data re- more abundant than on Xa (0.26%; P = 0.001).