Cultured cell lines are the workhorse of cancer research, but the extent to which they recapitulate the heterogeneity observed among malignant cells in tumors is unclear. Here we used multiplexed single-cell RNA-seq to profile 198 cancer cell lines from 22 cancer types. We identified 12 expression programs that are recurrently heterogeneous within multiple cancer cell lines. These programs are associated with diverse biological processes, including cell cycle, senescence, stress and interferon responses, epithelial–mesenchymal transition and protein metabolism. Most of these programs recapitulate those recently identified as heterogeneous within human tumors. We prioritized specific cell lines as models of cellular heterogeneity and used them to study subpopulations of senescence-related cells, demonstrating their dynamics, regulation and unique drug sensitivities, which were predictive of clinical response. Our work describes the landscape of heterogeneity within diverse cancer cell lines and identifies recurrent patterns of heterogeneity that are shared between tumors and specific cell lines.

培养的细胞系是癌症研究的主力，但它们在多大程度上概括了肿瘤中恶性细胞中观察到的异质性尚不清楚。在这里，我们使用多重单细胞 RNA-seq 来分析 22 种癌症类型的 198 个癌细胞系。我们鉴定了 12 个表达程序，这些程序在多个癌细胞系中经常存在异质性。这些程序与多种生物过程相关，包括细胞周期、衰老、应激和干扰素反应、上皮-间质转化和蛋白质代谢。大多数这些程序都概括了最近在人类肿瘤中发现的异质性。我们优先考虑特定细胞系作为细胞异质性模型，并用它们来研究衰老相关细胞的亚群，展示它们的动态、调节和独特的药物敏感性，这些可预测临床反应。我们的工作描述了不同癌细胞系内的异质性景观，并确定了肿瘤和特定细胞系之间共有的异质性的重复模式。