Locus-specific and genome-wide analysis of adaptations to hypoxia in iPSC-derived endothelium from a population adapted to high altitude [scRNA-Seq]

Tibetan adaptation to high-altitude hypoxia remains a classic example of Darwinian selection in humans. Amongst Tibetans, alleles in the EPAS1 gene - whose protein product, Hif-2α, is a central regulator of the hypoxia response - have repeatedly been shown to carry some of the strongest signals of positive selection in humans and to influence several adaptive phenotypes. We recently showed that the selected haplotype at this locus spans a hypoxia-dependent enhancer (ENH5) that contributes to the regulation of EPAS1 in a variety of cell types. However, selective sweep signals alone may account for only part of the phenotypes that differentiate Tibetans from closely related lowlanders. Therefore, there is a pressing need to functionally probe adaptive alleles and their impact at the genome-wide level and across cell types to uncover the full range of beneficial traits. To cast a wider net, we established a library of induced pluripotent stem cells (iPSCs) derived from Tibetan and Han Chinese individuals, a robust model system allowing precise exploration of both locus-specific and genome-wide effects on transcriptional responses. We harness this system by differentiating the iPSC library into vascular endothelium and investigating the locus-specific effects of the ENH5 enhancer in this cellular context. In addition, we use it to explore Tibetan-specific transcriptome-wide responses and find evidence that energy metabolism and immune pathways have been shaped by natural selection in Tibetans. Finally, to aid with the interpretation of the transcriptional differences between populations, we test for polygenic adaptations as a complementary in silico approach for the identification of beneficial Tibetan phenotypes. We developed a panel of sex-matched Tibetan and Han Chinese iPSCs using lymphoblastoid cell lines (LCLs) from 10 unrelated individuals of Tibetan origin and 10 Han Chinese (CHB) from the 1000 Genome Project.These 20 LCLs were reprogrammed into iPSCs in batches, balanced by population and sex and then successfully differentiated into vascular endothelium. We used CRISPR-CAS9 editing to delete the region (chr2: 46578867 to 46579857; hg19) spanning the ENH5 enhancer region of the iPSC line derived from Han individual CHB633 in order to create knockout cells carrying a low-altitude haplotype across the entire EPAS1 locus, but with complete loss of ENH5 activity. These DENH5 iPS cells (subsequently referred to as knockout, KO) and the original CHB633 (wild type, WT) iPS cells were differentiated into vascular endothelium in five independent replicates and cultured in hypoxic conditions (1% O2) for 48 hours. In order to assess the impact of the ENH5 deletion in vascular endothelial cells, we performed single cell RNA-sequencing in a targeted 10,000 cells per KO and WT replicate. We assessed the expression of 25 endothelial markers (in the 10 UMAP clusters identified by Seurat) and identified two cell clusters as being robustly enriched in all endothelial markers. Transcript data from the cells in these two clusters were pooled to compare expression differences between KO and WT in a cell-type specific manner. We utilized our complete panel of 20 cell lines to compare transcript levels between TAC and CHB iPSC-derived vascular endothelium under both normoxic and hypoxic conditions. Given the larger number of cell lines and treatment conditions, we performed bulk RNA-sequencing rather than single cell RNA-sequencing.

藏族对高海拔低氧环境的适应一直是人类达尔文选择的经典范例。在藏族人群中，EPAS1基因的等位基因——其编码的蛋白质产物Hif-2α是低氧应答的核心调控因子——多次被证实带有人类中最强的正选择信号之一，并影响多种适应性表型。我们此前的研究表明，该位点的选择单倍型跨越一个依赖于低氧的增强子（ENH5），该增强子可在多种细胞类型中调控EPAS1的表达。然而，仅靠选择性清除信号或许只能解释藏族与亲缘关系密切的低海拔人群之间部分表型差异。因此，亟需在全基因组层面以及跨细胞类型中对适应性等位基因及其影响开展功能探究，以揭示所有有益性状的全貌。
为了开展更广泛的研究，我们建立了源自藏族与汉族个体的诱导多能干细胞（induced pluripotent stem cells, iPSCs）文库，这一可靠的模型系统可精准探究位点特异性与全基因组层面的转录应答效应。我们利用该系统将iPSC文库诱导分化为血管内皮细胞，并在该细胞环境中探究ENH5增强子的位点特异性效应。此外，我们借助该系统探索藏族特异性的全转录组应答，发现能量代谢与免疫通路曾在藏族人群中受到自然选择塑造的证据。最后，为辅助解读人群间的转录组差异，我们采用多基因适应作为一种互补的计算机模拟（in silico）方法，用于识别藏族的有益表型。
我们利用来自10名无关藏族个体与10名来自千人基因组计划（1000 Genome Project）的汉族（CHB）个体的淋巴母细胞系（lymphoblastoid cell lines, LCLs），构建了性别匹配的藏族与汉族iPSC文库。这20个LCLs按人群与性别进行批次平衡后被重编程为iPSCs，并成功诱导分化为血管内皮细胞。我们使用CRISPR-Cas9编辑技术，删除源自汉族个体CHB633的iPSC系中覆盖ENH5增强子区域的序列（chr2: 46578867至46579857；hg19），以构建在整个EPAS1位点携带低海拔单倍型但完全丧失ENH5活性的敲除细胞。这些ΔENH5 iPS细胞（后续称为敲除型，KO）与原始的CHB633（野生型，WT）iPS细胞被独立重复5次诱导分化为血管内皮细胞，并在低氧条件（1% O₂）下培养48小时。
为评估ENH5缺失对血管内皮细胞的影响，我们针对每个KO与WT重复样本的10000个细胞开展了单细胞RNA测序（single cell RNA-sequencing）。我们评估了25种内皮细胞标志物在Seurat鉴定出的10个UMAP簇中的表达情况，并确定了两个在所有内皮标志物中均显著富集的细胞簇。将这两个簇中的细胞的转录组数据合并后，我们以细胞类型特异性的方式比较了KO与WT之间的表达差异。我们利用全部20个细胞系文库，比较了常氧与低氧条件下TAC与CHB iPSC来源的血管内皮细胞的转录水平。鉴于细胞系与处理条件的数量较多，我们采用了批量RNA测序（bulk RNA-sequencing）而非单细胞RNA测序。