Novel enzyme-based reduced representation method for DNA methylation profiling with low inputs

DNA methylation at cytosine-phospho-guanine (CpG) residues is a vital biological process that regulates cell identity and function. Although widely used, bisulfite-based cytosine conversion procedures for DNA methylation sequencing require high temperature and extreme pH, which leads to DNA degradation, especially among unmethylated cytosines. This disproportionate damage to unmethylated cytosines contributes to inaccuracies in GC content representation. EM-seq, an enzyme-based cytosine conversion method, has been proposed as a less biased alternative to methylation profiling. Compared to bisulfite-based methods, EM-seq boasts greater genome coverage with less GC bias and has the potential to cover more CpGs with the same number of reads (i.e., higher signal-to-noise ratio). Reduced representation approaches enrich samples for CpG-rich genomic regions, thereby enhancing throughput and cost effectiveness. We hypothesized that enzyme-based technology could be adapted for reduced representation methylation sequencing to enable high-resolution DNA methylation profiling on low inputs samples. We leveraged the well-established differences in methylation profile between mouse CD4+ T cell populations to compare reduced representation EM-seq (RREM-seq) performance against our previously published modified reduced representation bisulfite sequencing (mRRBS). While the mRRBS method failed to generate reliable DNA libraries when using <2-ng inputs (equivalent to DNA from around 350 cells), the RREM-seq method successfully generated DNA libraries from 1–25 ng of mouse and human genomic DNA. These libraries fell within the expected size range, and primer contamination was not observed. Low-input (<2-ng) RREM-seq libraries’ final concentration, regulatory genomic element coverage, and methylation status within the lineage-defining Treg cell-specific super-enhancers were comparable to mRRBS libraries with more than 10-fold higher DNA input. RREM-seq libraries also successfully detected the methylation differences between alveolar Tconv and Treg cells in mechanically ventilated patients with severe SARS-CoV-2 pneumonia. Our results suggest that the RREM-seq method can generate reliable libraries for single-nucleotide resolution methylation profiling using low input clinical samples. DNA libraries for methylation sequencing were generated using flow cytometry-sorted splenic CD4+ conventional T cells (Tconv cells) and CD4+CD25+FOXP3+ regulatory T cells (Treg cells) from healthy mice and paired alveolar CD4+ Tconv and Treg cells from mechanically ventilated patients with severe SARS-CoV-2 pneumonia. *************************************************************** Submitter states that missing raw data are being made available for controlled access in dbGaP. ***************************************************************

胞嘧啶-磷酸-鸟嘌呤（cytosine-phospho-guanine，CpG）位点的DNA甲基化是调控细胞身份与功能的关键生物学过程。尽管基于亚硫酸氢盐的胞嘧啶转化流程已被广泛应用于DNA甲基化测序，但该方法需要高温与极端pH环境，会导致DNA降解，尤其是未甲基化的胞嘧啶。这种对未甲基化胞嘧啶的选择性损伤，会导致GC含量表征出现偏差。EM-seq作为一种酶法胞嘧啶转化方法，已被提出作为甲基化谱分析的低偏差替代方案。相较于基于亚硫酸氢盐的方法，EM-seq具备更高的基因组覆盖度与更低的GC偏好性，且在相同测序读段数下可覆盖更多CpG位点（即更高的信噪比）。简化代表性测序策略可对富含CpG的基因组区域进行样本富集，从而提升测序通量并降低成本。我们提出假设：酶法技术可适配简化代表性甲基化测序，从而实现低起始量样本的高分辨率DNA甲基化谱分析。我们利用已被充分验证的小鼠CD4+ T细胞群体间甲基化谱差异，对比了简化代表性EM-seq（reduced representation EM-seq，RREM-seq）与此前发表的改良型简化代表性亚硫酸氢盐测序（modified reduced representation bisulfite sequencing，mRRBS）的性能。当起始DNA量低于2 ng（约相当于350个细胞的DNA含量）时，mRRBS法无法构建合格的DNA文库；而RREM-seq法则可成功从1~25 ng的小鼠及人类基因组DNA中构建文库。所构建的文库均处于预期的片段大小范围内，且未检测到引物污染。低起始量（<2 ng）的RREM-seq文库的最终浓度、调控基因组元件覆盖度，以及谱系定义性Treg细胞特异性超级增强子区域的甲基化状态，均与DNA起始量高出10倍以上的mRRBS文库相当。RREM-seq文库还成功检测了重症SARS-CoV-2肺炎机械通气患者肺泡常规T细胞（conventional T cells，Tconv细胞）与调节性T细胞（regulatory T cells，Treg细胞）间的甲基化差异。本研究结果表明，RREM-seq法可利用低起始量临床样本构建适用于单核苷酸分辨率甲基化谱分析的合格文库。本研究通过流式细胞术分选获取健康小鼠的脾脏CD4+ 常规T细胞与CD4+CD25+FOXP3+ 调节性T细胞，以及重症SARS-CoV-2肺炎机械通气患者的配对肺泡CD4+ 常规T细胞与调节性T细胞，用于构建甲基化测序DNA文库。
****************************************************************
提交者说明：缺失的原始数据将通过dbGaP（Database of Genotypes and Phenotypes，基因型与表型数据库）进行受控访问共享。
****************************************************************