BindDB: Bivalent genes Case Study

Validate and Expand Established Epigenetic Profiles In order to validate the ability of BindDB to detect an already well-established epigenetic profile, we began with the large group of 'bivalent' genes coined on the premise of the epigenetic characteristics of their promoters. These 3913 genes in human and 2984 genes in mouse(Azuara et al., 2006; Bernstein et al., 2006; Li et al., 2013) have the H3K4me3 active promoter hallmark as well as the H3K27me3 repressive histone modification in close proximity to their transcriptional start sites and are lowly expressed in ESCs(Bernstein et al., 2006). The repressed nature of the bivalent promoters in ESCs is conferred by the Polycomb repressive complexes (PRC), which interact with the H3K27me3 mark. We uploaded the list of bivalent genes by gene name to BindDB, selected the 'proximal promoter (-1000, +1000) as the gene portion to query and within less than 5 minutes, could confirm all of the above. Enrichment score analysis (the ratio of factor binding to the queried gene promoters divided by the ratio of factor binding to all gene promoters) clearly indicates an enrichment in the characteristic "K4-K27" (Figure S1A, blue arrows) histone marking at bivalent gene promoters. In addition, the depletion of H3K36me3 indicates that these genes, although exhibiting an active chromatin mark, are not expressed at high levels. H3K36me3 is also depleted across the bivalent gene bodies when compared to all gene bodies (data not shown).This finding complies with the hypothesis that PHF19, a component of the Polycomb complex, recruits histone-lysine demethylase NO66 in order to reduce levels of H3K36me3 at bivalent genes(Brien et al., 2012) and is strengthened by the enrichment of H3K36me2 instead, and PHF19 itself at these promoters. In addition to PHF19, many other components of the Polycomb complexes can be found as well, including EZH1 and EZH2, SUZ12, JARID2, and RING1 proteins. The presence of KDM2A and KDM2B may imply that these proteins are involved in the demethylation of either lysine 4 or lysine 27 during cell fate determination. Alternatively, KDM2B (FBXL10) along with CBX7, which also shows enrichment in the BindDB analysis (Figure S1A, left), have been shown to be involved in the recruitment of the PRC1 complex to the H3K27me3 histone mark (He et al., 2013; Morey et al., 2012). Strikingly, hardly any significant enrichment of transcription factors could be detected at bivalent genes in ESCs (Figure S1B, filtered heatmap), except for OCT4 and SOX2, in line with previous findings that these key ESC transcription factors bind approximately one third of PRC2-occupied genes that also encode developmental transcription factors (Boyer et al., 2006; Lee et al., 2006). This phenomenon is unique to this group of genes (see subsequent examples) and signifies the important role of epigenetic regulation on these 'poised', yet inactive genes in ESCs.验证并拓展已确立的表观遗传谱 为验证BindDB检测已被广泛证实的表观遗传特征的能力，我们以基于启动子表观遗传特征所定义的“二价”基因大类作为研究起点。人类的3913个基因与小鼠的2984个基因(Azuara等人，2006；Bernstein等人，2006；Li等人，2013)，在其转录起始位点附近同时带有H3K4me3活性启动子标记与H3K27me3抑制性组蛋白修饰，且在胚胎干细胞(ESCs)中呈低表达水平(Bernstein等人，2006)。胚胎干细胞中二价启动子的抑制特性由多梳抑制复合体(Polycomb repressive complexes, PRC)赋予，该复合体可与H3K27me3标记结合。我们将按基因名称整理的二价基因列表上传至BindDB，选择“近端启动子(-1000, +1000)作为查询基因区域，在5分钟内即可验证上述全部结论。富集得分分析（即查询基因启动子的因子结合比例除以所有基因启动子的因子结合比例）清晰表明，二价基因启动子处存在特征性的"K4-K27"组蛋白标记富集（图S1A，蓝色箭头）。此外，H3K36me3的缺失表明，尽管这些基因带有活性染色质标记，但并未呈高表达水平。与所有基因体相比，二价基因体中的H3K36me3同样存在缺失（未展示数据）。这一发现与PHF19的相关假说相符：PHF19作为多梳复合体的组分之一，可招募组蛋白赖氨酸去甲基化酶NO66，以降低二价基因处的H3K36me3水平(Brien等人，2012)，而H3K36me2以及PHF19本身在这些启动子处的富集进一步验证了该假说。除PHF19外，还可检测到多梳复合体的诸多其他组分，包括EZH1、EZH2、SUZ12、JARID2与RING1蛋白。KDM2A与KDM2B的存在或许意味着，这些蛋白在细胞命运决定过程中参与赖氨酸4或赖氨酸27的去甲基化。或者，KDM2B(FBXL10)与同样在BindDB分析中呈现富集的CBX7（图S1A，左侧），已被证实可参与将PRC1复合体招募至H3K27me3组蛋白标记处(He等人，2013；Morey等人，2012)。值得注意的是，在胚胎干细胞的二价基因中几乎未检测到显著富集的转录因子（图S1B，过滤后的热图），仅OCT4与SOX2除外，这与此前的研究结论一致：这些关键的胚胎干细胞转录因子可结合约三分之一同时编码发育相关转录因子且被PRC2占据的基因(Boyer等人，2006；Lee等人，2006)。该现象是此类基因所特有的（详见后续示例），其表明表观遗传调控在胚胎干细胞中这些“待命”却无活性的基因上发挥着重要作用。