Additional file 1 of DNA hydroxymethylation is associated with disease severity and persists at enhancers of oncogenic regions in multiple myeloma

Additional file 1. Figure S1: Agilent BioAnalyzer profile of a 5hmC-seq Illumina library. Figure S2: MS quantification of 5mC in genomic plasma cell DNA is independent of age and sex. (A) Dot plot of 5mC global quantification by MS in normal plasma cells from healthy donors (N=5), and of myeloma cells of patients at diagnosis (N=40). (B) Dot plot of 5mC global quantification by MS by disease stage (ISS I N=9; ISS II N=17; ISS III N=13; NA=1). 5mC (C) and 5hmC (D) dot plot of MS quantification depending on the sex of the patients. 5mC (E) and 5hmC (F) dot plot of MS quantification depending on the age of the patients. Figure S3: Survival course depending on DNA methylation (5mC) level-based separation of two risk groups of NDMM (n=20 and 20). Figure S4: 5hmC association with expression and criteria of 5hmC peaks to merge in 5hmC peak clusters. (A) Average level of 5hmC in all genes normalized to the same body length. Red line stands for average 5hmC in genes with high expression level (greater than 100 Reads per Kilobase Million (RPKM)). Orange line stands for medium expression level (between 10 and 100 RPKM). Green line represents lowly expressed genes (between 1 and 10 RPKM), and blue line stands for very lowly expressed genes (below 1 RPKM). (B) Stitching of 5hmC into 5hmC-enriched domains. The y-axis represents the number of peaks left after merging. The x-axis represents the distance between peaks to merge. Each 5hmC sample was analyzed (one color per patient). The distance 12.5 kb was chosen to stitch 5hmC peaks into the 5hmC-enriched domains that we describe in this study. (C) Fraction of overlap between 5hmC-enriched domains of this study and CpG from the Illumina 450K chip. The red bar represents overlap with hypermethylated CpGs in B cell-specific enhancers that were described by Agirre and colleagues (see Additional file 3: Methods). Blue bars represent random CpGs from the same chip. Figure S5: 5hmC allows the identification of a putative CCND2 enhancer. (A) Correlation between CCND2 expression, 5hmC at CCND2 gene body and 5hmC at the putative 5hmC enhancer across the 40 MM patients. (B) Hi-C contact map in lymphoblastoid cells (GM12878 cell line) at the CCDN2 locus showing the spatial interaction between CCND2 gene and its putative enhancer. (C) Expression of core transcription factors predicted to orchestrate core regulatory circuitries with 5hmC and RNA expression genomic data. Figure S6: MM 5hmC-enriched domains associate with H3K27ac super-enhancers. Rank ordering of the 100 strongest 5hmC-enriched domains on average in the cohort (A), in the MMSET group (B), in the CCND1 group (C) and in the hyperdiploid group (D). Color highlights domains present in only one of the ROSE plots by group. Figure S7: 5hmC signal levels at WNT5B-associated domain are increased at relapse in MM07. (A) Normalized 5hmC enrichment at WNT5B-associated domain. Point shapes match replicates. Fold change=1.3, p=0.003, FDR>0.1. (B) Gene expression levels in RPKM measured by RNA-seq at diagnosis and relapse for three genes surrounding the WNT5B-associated domain. (C) 5hmC genomic signal around WNT5B-associated domain. Colors match those of (A) and (B). 5hmC domain is depicted under signal tracks (hg38: chr12:1,517,750-1,621,200).

附加文件1 图S1：5-羟甲基胞嘧啶测序（5hmC-seq）Illumina文库的安捷伦生物分析仪分析图谱。图S2：基因组浆细胞DNA中5-甲基胞嘧啶（5mC）的质谱（MS）定量结果与患者年龄和性别无关。(A) 健康供者来源正常浆细胞（N=5）及初诊多发性骨髓瘤（MM）患者骨髓瘤细胞（N=40）的5mC全局质谱定量散点图；(B) 按疾病分期绘制的5mC全局质谱定量散点图：国际分期系统（ISS）I期（N=9）、ISS II期（N=17）、ISS III期（N=13），未知分期（NA=1）。(C) 基于患者性别的5mC质谱定量散点图，(D) 基于患者性别的5hmC质谱定量散点图；(E) 基于患者年龄的5mC质谱定量散点图，(F) 基于患者年龄的5hmC质谱定量散点图。图S3：基于DNA甲基化（5mC）水平划分的新诊断多发性骨髓瘤（NDMM）两个风险组（n=20和n=20）的生存曲线。图S4：5-羟甲基胞嘧啶（5hmC）与基因表达的关联及5hmC峰聚类的合并标准。(A) 所有基因经标准化后的平均5hmC水平（以基因体长度归一化）。红色线代表高表达基因（表达量＞100每千碱基百万读段（RPKM））的平均5hmC水平，橙色线代表中等表达基因（10~100 RPKM）的平均5hmC水平，绿色线代表低表达基因（1~10 RPKM）的平均5hmC水平，蓝色线代表极低表达基因（＜1 RPKM）的平均5hmC水平。(B) 5hmC峰向5hmC富集结构域的聚类合并。纵轴为合并后剩余的峰数量，横轴为待合并峰之间的间距。本研究对所有5hmC样本进行分析（每位患者的样本以单一颜色标识），最终选取12.5 kb作为峰合并间距，用于构建本研究提及的5hmC富集结构域。(C) 本研究的5hmC富集结构域与Illumina 450K芯片CpG位点的重叠比例。红色柱代表与Agirre等学者报道的B细胞特异性增强子中超甲基化CpG位点的重叠比例（详见附加文件3：方法），蓝色柱代表同一芯片上随机选取的CpG位点的重叠比例。图S5：5hmC可用于鉴定潜在的CCND2增强子。(A) 40例MM患者中CCND2表达量、CCND2基因体5hmC水平与潜在增强子区域5hmC水平的相关性分析；(B) 淋巴母细胞系（GM12878细胞系）中CCDN2基因座的Hi-C相互作用图谱，展示了CCND2基因与其潜在增强子之间的空间互作；(C) 结合5hmC与RNA表达组学数据预测的核心调控环路相关核心转录因子的表达情况。图S6：MM的5hmC富集结构域与H3K27ac超级增强子相关。按平均信号强度对全队列(A)、MMSET组(B)、CCND1组(C)及超二倍体组(D)中排名前100的5hmC富集结构域进行排序，不同组仅在单张ROSE图中出现的结构域以颜色标注区分。图S7：MM07患者复发时，WNT5B相关结构域的5hmC信号水平升高。(A) WNT5B相关结构域的标准化5hmC富集水平。点形状对应生物学重复，倍数变化（Fold Change）=1.3，p=0.003，错误发现率（FDR）＞0.1；(B) WNT5B相关结构域上下游三个基因在初诊及复发时的RNA测序（RNA-seq）定量表达水平（以RPKM为单位）；(C) WNT5B相关结构域周边的基因组5hmC信号谱，颜色与(A)(B)一致。5hmC结构域在信号轨道下方标注（hg38：chr12:1,517,750-1,621,200）。