Combined CRISPR and proteomics screening reveal a cohesin-CTCF-bound allele contributing to increased expression of RUVBL1 and prostate cancer progression [gRNA]. Combined CRISPR and proteomics screening reveal a cohesin-CTCF-bound allele contributing to increased expression of RUVBL1 and prostate cancer progression [gRNA]

Introduction Genome-wide association studies along with expression quantitative trait loci (eQTL) have identified hundreds of single nucleotide polymorphisms (SNPs) and their target genes in prostate cancer (PrCa). Although these genetic associations to PrCa have been widely reported, functional characterization of these risk loci remains challenging. Methods To screen for regulatory SNPs, we designed a library containing 9133 guide RNAs (gRNAs) to target 2,166 candidate SNP sites implicated in PrCa. We performed negative screening in dCas9-KRAB stable prostate cell lines and applied the RIGOR program to identify the essential SNPs for cell proliferation. We further characterized the regulatory role of a selected single nucleotide polymorphim (SNP, rs60464856) using luciferase reporter assay, ChIP-qPCR, and xCas9 base editing in prostate cells. Finally, we investigated the biological impact of the SNP-regulated gene RUVBL1 on cell proliferation and tumor growth via gene knockdown using in vitro and in vivo assays. Results From interference of 2,166 candidate SNPs via CRISPR interference screening, the RIGOR program identified 117 SNPs that could regulate genes to promote growth advantage in prostate cancer cell lines. Compared to unselected SNPs, the 117 candidates tended to reside near 5 kb flanking the transcription start sites (p = 0.01). To characterize the regulatory role of these SNPs, we selected one SNP (rs60464856) for detailed analysis. This SNP was covered by multiple gRNAs significantly depleted in the screening (FDR<0.05). Pooled SNP association analysis in the PRACTICAL cohort showed significantly higher PrCa risk for the G allele (pvalue=1.2E-16). eQTL analysis showed that the G allele is associated with an increased expression of RUVBL1 in multiple datasets. To further validate the CRISPR interference effect, we transfected a gRNA targeting the rs60464856 site in the dCas9 stable cell lines and observed significant inhibition of the RUVBL1 expression. We also applied the xCas9 adenine base editor to convert the rs60464856 A into G allele and observed an increased RUVBL1 expression in subclones carrying the rs60464856 G allele in prostate cell lines. To test if any protein showed allele-specific binding at rs60464856, we used SILAC-based DNA pull-down proteomics and observed that cohesin subunits (including SMC3) preferred the A allele. ChIP qPCR assays showed significant enrichment of CTCF and SMC3 signals at the rs60464856 site with preferential binding to the A allele. To evaluate the potential role of this locus in maintaining long-range chromatin structure, we analyzed a HiC dataset and found that the rs60464856 locus enriched consistent chromatin interactions in prostate cell lines. To determine the potential role of the rs60464856 target gene, we knocked down the RUVBL1 via shRNA and observed significant proliferation inhibition in prostate cell lines. We also tested PC3 cells with RUVBL1 knockdown in nude mice xenografts and observed reduced tumorigenesis. Gene set enrichment analysis showed that RUVBL1 expression was associated with the enrichment of cell cycle related pathways in both cell line and TCGA prostate cancer cohorts. Lastly, we showed that an increased RUVBL1 expression and its relevant pathway activation were associated with poor survival. Conclusion We applied the CRISPR interference screening at selected prostate cancer risk loci and identified over a hundred regulatory SNPs essential for prostate cell proliferation. Further analysis confirmed the important role of rs60464856 and its target gene RUVBL1 in prostate cell growth and tumorigenesis. To discover proliferative essential SNP loci, we used CRISPRi screening technology and DNA-seq to determine those functional candidate from previously published eQTL variants Overall design: Guide RNA (gRNA) read count were compared between screening ending and start in two replicates with RIGER program. Differential gene expression analysis were performed to compare the RUVBL1 induced changes

引言
全基因组关联研究结合表达数量性状位点（expression quantitative trait loci, eQTL）已在前列腺癌（prostate cancer, PrCa）中鉴定出数百个单核苷酸多态性（single nucleotide polymorphisms, SNPs）及其靶基因。尽管这些与PrCa相关的遗传关联已被广泛报道，但对这些风险位点的功能表征仍颇具挑战。

方法
为筛选调控型SNPs，我们设计了包含9133条向导RNA（guide RNAs, gRNAs）的文库，以靶向与PrCa相关的2166个候选SNP位点。我们在dCas9-KRAB稳定前列腺细胞系中开展了负向筛选，并应用RIGOR程序鉴定对细胞增殖至关重要的SNPs。我们进一步在前列腺细胞中利用荧光素酶报告基因测定、染色质免疫共沉淀定量PCR（ChIP-qPCR）及xCas9碱基编辑技术，对选定的单核苷酸多态性（SNP, rs60464856）的调控功能进行了表征。最后，我们通过体外与体内实验开展基因敲低（gene knockdown），探究了该SNP调控的基因RUVBL1对细胞增殖及肿瘤生长的生物学影响。

结果
借助CRISPR干扰（CRISPR interference, CRISPRi）筛选对2166个候选SNPs进行干预后，RIGOR程序鉴定出117个可调控基因并赋予前列腺癌细胞系生长优势的SNPs。与未筛选的SNPs相比，这117个候选位点多位于转录起始位点上下游5kb范围内（p=0.01）。
为表征这些SNPs的调控功能，我们选取其中一个SNP（rs60464856）进行详细分析。该位点被多条在筛选中显著耗竭的gRNAs覆盖（错误发现率<0.05）。对PRACTICAL队列开展的合并SNP关联分析显示，G等位基因与更高的PrCa风险显著相关（p值=1.2×10^-16）。eQTL分析显示，在多个数据集里，G等位基因与RUVBL1的表达上调显著相关。
为进一步验证CRISPR干扰的效应，我们在dCas9稳定细胞系中转染了靶向rs60464856位点的gRNA，并观察到RUVBL1的表达受到显著抑制。我们还应用xCas9腺嘌呤碱基编辑器将rs60464856位点的A等位基因转化为G等位基因，并在携带rs60464856 G等位基因的前列腺细胞系亚克隆中观察到RUVBL1的表达上调。
为检测是否存在蛋白质在rs60464856位点表现出等位基因特异性结合，我们采用基于细胞培养中氨基酸稳定同位素标记（stable isotope labeling with amino acids in cell culture, SILAC）的DNA pull-down蛋白质组学技术，发现黏连蛋白复合物亚基（包括SMC3）偏好结合A等位基因。ChIP-qPCR实验显示，CTCF与SMC3在rs60464856位点存在显著富集，且优先结合A等位基因。
为评估该位点在维持长距离染色质结构中的潜在作用，我们分析了一个高通量染色体构象捕获（Hi-C）数据集，发现rs60464856位点在前列腺细胞系中存在稳定的染色质相互作用富集。
为明确rs60464856靶基因的潜在功能，我们通过短发夹RNA（short hairpin RNA, shRNA）敲低RUVBL1的表达，并观察到前列腺细胞系的增殖受到显著抑制。我们还在裸鼠异种移植模型中对RUVBL1敲低的PC3细胞进行了实验，观察到肿瘤发生能力减弱。基因集富集分析显示，无论是在细胞系还是癌症基因组图谱（The Cancer Genome Atlas, TCGA）前列腺癌队列中，RUVBL1的表达均与细胞周期相关通路的富集显著相关。最后，我们证实RUVBL1的表达上调及其相关通路激活与不良预后显著相关。

结论
我们针对选定的前列腺癌风险位点开展了CRISPR干扰筛选，鉴定出百余对前列腺细胞增殖至关重要的调控型SNPs。进一步分析证实了rs60464856及其靶基因RUVBL1在前列腺细胞生长及肿瘤发生中的重要作用。

为鉴定对增殖至关重要的SNP位点，我们采用CRISPRi筛选技术与DNA测序（DNA-seq）技术，从已发表的eQTL变异体中筛选功能性候选位点。
总体实验设计：利用RIGER程序对两个生物学重复的筛选终点与起始点的gRNA读段计数进行比较。开展差异基因表达分析以对比RUVBL1敲低诱导的基因表达变化。