【摘要】 目的 整合结直肠癌单核转录组测序(single nucleus RNA sequencing, snRNA-seq)和单细胞染色质可及性测序(single cell assay for transposase-accessible chromatin sequencing, scATAC-seq)数据，识别结直肠癌恶性上皮细胞中的关键转录因子，并评估其功能、预后价值及潜在治疗相关性。方法 下载GEO数据库中结直肠癌snRNA-seq数据集GSE201348和scATAC-seq数据集GSE201336,纳入结直肠癌组织和正常组织样本进行单细胞图谱构建及细胞注释。基于inferCNV分析识别恶性上皮细胞，并采用DoRothEA/decoupleR推断恶性上皮细胞的转录因子活性。通过整合转录因子活性及染色质可及性特征，筛选恶性上皮细胞相关优先候选转录调控因子。进一步基于候选转录因子靶基因进行基因本体论(gene ontology, GO)和京都基因与基因组百科全书(kyoto encyclopedia of genes and genomes, KEGG)富集分析和转录调控网络构建，并利用TCGA-COAD/READ及5套Bulk转录组队列评估候选因子与总生存期(overall survival, OS)的关系。最后，基于CellMiner数据库分析候选转录因子表达水平与DTP NCI-60药物活性指标之间的相关性，并结合公开药物证据评估候选药物的优先级。结果 质控后共获得53 426个snRNA-seq高质量细胞和41 829个scATAC-seq高质量细胞，其中snRNA-seq中识别出11 872个结直肠癌恶性上皮细胞。综合单细胞转录因子活性、染色质可及性证据及外部队列支持，最终筛得8个结直肠癌恶性上皮细胞关键转录因子：MYC、HIF1A、KLF5、TEAD1、SRF、BACH1、TCF7L2和HNF4A,其中SRF结合既往结直肠癌相关研究证据及Kaplan-Meier生存分析结果，仍作为补充候选因子纳入后续分析。GO功能富集分析显示，这些转录因子主要参与细胞增殖、缺氧与代谢适应、上皮生长维持、侵袭迁移、干性维持及分化调控等过程，并基于其靶基因关系构建了转录调控网络。6套Bulk队列预后分析以OS为主要终点。Kaplan-Meier生存分析显示，SRF高表达组在4个外部队列中较低表达组表现出更差OS(Log-rank P<0.05);BACH1和TEAD1高表达组在部分队列中亦与较差OS相关(Log-rank P<0.05),而HNF4A高表达组在部分队列中与较好OS相关。CellMiner相关性分析显示，不同候选转录因子表达水平与DTP NCI-60药物活性指标之间的Pearson相关系数和Spearman相关系数存在方向和强度差异，其中TEAD1、MYC和HIF1A相关候选药物组合的综合优先级相对较高。结论 本研究识别出结直肠癌恶性上皮细胞相关的8个关键转录因子(MYC、HIF1A、KLF5、TEAD1、SRF、BACH1、TCF7L2和HNF4A),涉及参与增殖、缺氧与代谢适应、侵袭迁移、干性维持及分化调控等过程，同时部分关键转录因子具有一定的预后评估价值和潜在治疗意义，可为结直肠癌恶性上皮细胞状态调控机制研究及候选生物标志物筛选提供参考。更多还原

【Abstract】 Objective The aim of this study was to integrate single-nucleus RNA sequencing(snRNA-seq)and single-cell assay for transposase-accessible chromatin sequencing(scATAC-seq)data of colorectal cancer, identify key transcription factors in malignant epithelial cells of colorectal cancer, and evaluate their function, prognostic value, and potential therapeutic relevance.Methods The snRNA-seq dataset GSE201348 and scATAC-seq dataset GSE201336 of colorectal cancer from the GEO database were downloaded, and colorectal cancer and normal tissue samples were included for single-cell atlas construction and cell annotation.Malignant epithelial cells were identified based on inferCNV analysis, and transcription factor activity of malignant epithelial cells was inferred using DoRothEA/decoupleR.Priority candidate transcriptional regulators associated with malignant epithelial cells were screened by integrating transcription factor activity and chromatin accessibility characteristics.Gene ontology(GO)and Kyoto Encyclopedia of Genes and Genomes(KEGG)enrichment analysis and transcription regulatory network construction were further performed based on candidate transcription factor target genes, and the relationship between candidate factors and overall survival(OS)was evaluated using TCGA-COAD/READ and five bulk transcriptomic cohorts.Finally, based on the CellMiner database, the correlation between the expression of candidate transcription factors and DTP NCI-60 drug activity indicators was analyzed, and the priority of candidate drugs was evaluated in combination with public drug evidence.Results After quality control, a total of 53,426 high-quality snRNA-seq cells and 41,829 high-quality scATAC-seq cells were obtained, including 11,872 malignant epithelial cells of colorectal cancer identified in the snRNA-seq dataset.By integrating single-cell transcription factor activity, chromatin accessibility evidence, and support from external cohorts, eight key transcription factors in malignant epithelial cells of colorectal cancer were finally screened: MYC,HIF1A,KLF5,TEAD1,SRF,BACH1,TCF7L2,and HNF4A.Among them, SRF was still included as a supplementary candidate factor in the subsequent analysis based on previous evidence from colorectal cancer-related research and Kaplan-Meier survival analysis results.GO functional enrichment analysis showed that these candidate transcription factors were primarily involved in processes such as cell proliferation, hypoxia and metabolic adaptation, epithelial growth maintenance, invasion and migration, stemness maintenance, and differentiation regulation.A transcriptional regulatory network was constructed based on their target gene relationships.Prognostic analysis of six bulk cohorts used OS as the primary endpoint.Kaplan-Meier survival analysis showed that OS in the high SRF expression group exhibited poorer than that in the low expression group in four external cohorts(Log-rank P<0.05).High expression of BACH1 and TEAD1 was also associated with poorer OS in some cohorts(Log-rank P<0.05),whereas high expression of HNF4A was associated with better OS in some cohorts.CellMiner correlation analysis showed differences in the direction and intensity of Pearson and Spearman correlation coefficients between the expression of different candidate transcription factor and DTP NCI-60 drug activity indicators, with TEAD1,MYC,and HIF1A-related candidate drug combinations having relatively high comprehensive priorities.Conclusions This study identified eight key transcription factors(MYC,HIF1A,KLF5,TEAD1,SRF,BACH1,TCF7L2,and HNF4A)associated with malignant epithelial cells in colorectal cancer, involved in processes such as proliferation, hypoxia and metabolic adaptation, invasion and migration, stemness maintenance, and differentiation regulation.Additionally, some key transcription factors possess prognostic evaluation value and potential therapeutic significance, providing a reference for studying the regulatory mechanisms of malignant epithelial cell states in colorectal cancer and screening candidate biomarkers.更多还原