The non-CG methylation is cell-type specific and conserved between human and mouse [Third-party reanalysis]

The non-CG methylations are abundant in several mammalian cell types, while their biological functions are largely unknown. We gathered 51 DNA methylomes in both human and mouse, covering brain neuron cells, embryonic stem (ES) and induced pluripotent stem (iPS) cells, primordial germ cells (PGC) and oocytes. Then a two-dimensional comparison on non-CG methylation was carried out across cell types and species. We employed an unbiased sub-motif prediction method and identified CW (W can be A or T) as the representative non-CG methylation context. We further showed mCW was significantly distinct from CC methylation (mCC) in both context and genomic distribution. Our studies on sequence preferences and genomic region enrichment revealed that mCW was cell-type specific and conserved between human and mouse. We reported young long interspersed nuclear elements-1 (LINE-1) elements tended to be deprived of mCW in brain for both species. Interestingly, both human Alu elements and mouse B1 elements were found to prefer high mCW at specific loci during evolution. Additionally, we reported a peak of CG methylation (mCG) at the promoter of young LINE-1 element in PGC, in contrast to the global CG demethylation. Lastly, we confirmed strand-skewed non-CG methylation in mouse ES intronic region, and showed the strand-specific distributions of mCW in LINE elements are also cell-type specific and conserved. Our findings indicate that mCW is conserved in species, dynamically regulated during development, and likely to guide the evolution of transposon elements. In this study, 41 samples were re-aligned from raw sequencing data using BS-Seeker2. The raw data available from the samples associated with the following Series: GSE11034 GSE30199 GSE30206 GSE34864 GSE37202 GSE42923 GSE46644 GSE46710 GSE49828 GSE51239 GSE52331 GSE56650 GSE61330 GSE61457 GSE63394 GSE63818 The accession numbers of the re-analyzed GEO Samples (GSMnnnnn; associated with each processed data file) are indicated in the 'complete_metadata.xls'. Please note that several raw data files are downloaded directly from SRA, DDBJ (https://trace.ddbj.nig.ac.jp), European Nucleotide Archive, and the public accession numbers for the raw data associated with each processed data file are indicated in the 'complete_metadata.xls' as well. The 'DATA PROCESSING PIPELINE' section (in the 'complete_metadata.xls') describes the re-analysis details.

非CG甲基化在多种哺乳动物细胞类型中广泛存在，但其生物学功能在很大程度上仍未明确。我们收集了人类与小鼠共51份DNA甲基化组样本，涵盖大脑神经元细胞、胚胎干细胞（embryonic stem, ES）、诱导多能干细胞（induced pluripotent stem, iPS）、原始生殖细胞（primordial germ cell, PGC）以及卵母细胞。随后我们针对不同细胞类型与物种开展了非CG甲基化的二维比较分析。我们采用无偏亚基序预测方法，确定了CW（W可代表A或T）为代表性非CG甲基化序列环境。我们进一步证实，mCW在序列环境与基因组分布两方面均显著区别于CC甲基化（mCC）。我们对序列偏好性与基因组区域富集性的研究显示，mCW具有细胞类型特异性，且在人类与小鼠中保守存在。我们发现，在两个物种的大脑中，年轻的长散在核元件-1（long interspersed nuclear elements-1, LINE-1）区域往往缺乏mCW修饰。有趣的是，人类Alu元件与小鼠B1元件在进化过程中，均倾向于在特定位点呈现高mCW修饰水平。此外，我们还报道了原始生殖细胞中年轻LINE-1元件启动子区域存在CG甲基化（mCG）峰值，这与全局CG去甲基化现象形成鲜明对比。最后，我们验证了小鼠胚胎干细胞内含子区域存在链偏好性非CG甲基化，并证实LINE元件中mCW的链特异性分布同样具有细胞类型特异性且在物种间保守。我们的研究结果表明，mCW在物种间保守存在，在发育过程中受到动态调控，且可能指导转座子元件的进化。本研究中，41份样本通过BS-Seeker2工具从原始测序数据中完成重新比对。原始数据可从以下系列（Series）获取：GSE11034、GSE30199、GSE30206、GSE34864、GSE37202、GSE42923、GSE46644、GSE46710、GSE49828、GSE51239、GSE52331、GSE56650、GSE61330、GSE61457、GSE63394、GSE63818。重新分析的GEO样本（GSMnnnnn，对应每份处理后的数据文件）的登录号详见"complete_metadata.xls"文件。请注意，部分原始数据文件直接从SRA、日本DNA数据库（DDBJ, https://trace.ddbj.nig.ac.jp）、欧洲核苷酸档案馆下载获取，每份处理后数据文件对应的原始数据公共登录号同样详见"complete_metadata.xls"文件。"complete_metadata.xls"中的"DATA PROCESSING PIPELINE"（数据处理流程）部分详细描述了本次重新分析的具体步骤。