CpG island turnover is associated with evolutionary changes in enhancer activity

Changes in the activity of transcriptional enhancers have been shown to influence phenotypic variation across species. However, it remains challenging to identify the sequence features associated with species differences in enhancer activity. Here we assessed the contribution of orphan CpG islands (oCGIs), which are regions enriched in CpG dinucleotides that are known to recruit chromatin modifiers and components of the transcriptional machinery. We integrated oCGIs in nine mammals with previously ascertained maps of chromatin modifications across multiple tissues, and found that species-specific oCGIs significantly co-localize with species-specific histone modification peaks. We focused on one enhancer of interest (hs754) which contains an oCGI in human but not in mouse, and overlaps an H3K27ac peak in the developing human brain, but not in mouse. We generated a humanized mouse model and profiled the locus using ChIP-seq, finding increased deposition of the histone modifications H3K27ac and H3K4me3 and gained CTCF binding sites in the humanized enhancer. ChIP-seq data for H3K27me3 was also generated and was not included in the final study, but is deposited here. We also measured gene expression in the humanized mouse and did not find any changes associated with the humanized enhancer, although using the datasets described above we found that genes near enhancers with species-specific oCGIs and species-specific activity show a species bias in expression. This work identifies oCGI turnover as a mechanism affecting mammalian enhancer evolution. ChIP-seq (H3K27ac, H3K4me3, H3K27me3, and CTCF) and RNA-seq were performed using developing diencephalon tissue (E11.5 and E17.5) from wild type and hs754-humanized mouse lines. The number of replicates for each genotype are as follows: 3 replicates for E17.5 ChIP-seq; 2 replicates for E11.5 ChIP-seq; 4 replicates for E11.5 and E17.5 RNA-seq.研究表明，转录增强子（transcriptional enhancers）的活性改变可影响跨物种的表型变异。然而，目前仍难以精准识别与增强子活性的物种差异相关的序列特征。本研究评估了孤儿CpG岛（orphan CpG islands, oCGIs）的作用——这类区域富含CpG二核苷酸，已知可招募染色质修饰因子与转录机器组分。我们整合了九种哺乳动物的oCGI数据与此前已绘制完成的多组织染色质修饰图谱，发现物种特异性oCGI可显著与物种特异性组蛋白修饰峰共定位。我们聚焦于目标增强子hs754：该增强子在人类中携带oCGI，但小鼠中不存在；其在发育中人脑中与H3K27ac峰重叠，但小鼠中无此现象。我们构建了人源化小鼠模型，并通过染色质免疫共沉淀测序（ChIP-seq）对该基因座进行分析，发现人源化增强子中H3K27ac与H3K4me3这两种组蛋白修饰的富集水平显著提升，且新增了CTCF结合位点。此外，我们还生成了H3K27me3的ChIP-seq数据，该数据未纳入最终研究，但已随本数据集公开提供。我们还检测了人源化小鼠的基因表达水平，未发现与人源化增强子相关的表达变化；不过通过上述数据集分析，我们发现携带物种特异性oCGI且具有物种特异性活性的增强子邻近基因，其表达存在物种偏好性。本研究明确oCGI周转是影响哺乳动物增强子演化的核心机制之一。本数据集使用野生型与hs754人源化小鼠品系的发育期间脑组织（胚胎日龄E11.5与E17.5）开展了染色质免疫共沉淀测序（ChIP-seq，检测靶点包括H3K27ac、H3K4me3、H3K27me3与CTCF）与RNA测序（RNA-seq）实验。各实验组的生物学重复数量如下：E17.5样本的ChIP-seq实验设3次重复；E11.5样本的ChIP-seq实验设2次重复；E11.5与E17.5样本的RNA-seq实验均设4次重复。