1] Department of Physiology and Medical Physics, Royal College of Surgeons in Ireland, Dublin 0002, Ireland [2] Centre for Systems Medicine, Royal College of Surgeons in Ireland, Dublin 0002, Ireland.
Cell Death Differ. 2013 Nov;20(11):1521-31. doi: 10.1038/cdd.2013.106. Epub 2013 Aug 9.
Many cancer entities and their associated cell line models are highly heterogeneous in their responsiveness to apoptosis inducers and, despite a detailed understanding of the underlying signaling networks, cell death susceptibility currently cannot be predicted reliably from protein expression profiles. Here, we demonstrate that an integration of quantitative apoptosis protein expression data with pathway knowledge can predict the cell death responsiveness of melanoma cell lines. By a total of 612 measurements, we determined the absolute expression (nM) of 17 core apoptosis regulators in a panel of 11 melanoma cell lines, and enriched these data with systems-level information on apoptosis pathway topology. By applying multivariate statistical analysis and multi-dimensional pattern recognition algorithms, the responsiveness of individual cell lines to tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) or dacarbazine (DTIC) could be predicted with very high accuracy (91 and 82% correct predictions), and the most effective treatment option for individual cell lines could be pre-determined in silico. In contrast, cell death responsiveness was poorly predicted when not taking knowledge on protein-protein interactions into account (55 and 36% correct predictions). We also generated mathematical predictions on whether anti-apoptotic Bcl-2 family members or x-linked inhibitor of apoptosis protein (XIAP) can be targeted to enhance TRAIL responsiveness in individual cell lines. Subsequent experiments, making use of pharmacological Bcl-2/Bcl-xL inhibition or siRNA-based XIAP depletion, confirmed the accuracy of these predictions. We therefore demonstrate that cell death responsiveness to TRAIL or DTIC can be predicted reliably in a large number of melanoma cell lines when investigating expression patterns of apoptosis regulators in the context of their network-level interplay. The capacity to predict responsiveness at the cellular level may contribute to personalizing anti-cancer treatments in the future.
许多癌症实体及其相关的细胞系模型在对凋亡诱导剂的反应性方面具有高度异质性，尽管对潜在信号网络有详细的了解，但目前仍不能根据蛋白质表达谱可靠地预测细胞死亡易感性。在这里，我们证明了将定量凋亡蛋白表达数据与途径知识相结合可以预测黑色素瘤细胞系的细胞死亡反应性。通过总共 612 次测量，我们确定了 11 个黑色素瘤细胞系中 17 个核心凋亡调节剂的绝对表达（nM），并通过细胞凋亡途径拓扑的系统水平信息对这些数据进行了丰富。通过应用多变量统计分析和多维模式识别算法，可以非常准确地预测单个细胞系对肿瘤坏死因子相关凋亡诱导配体（TRAIL）或达卡巴嗪（DTIC）的反应性（91%和 82%的正确预测），并且可以在计算机中预先确定个别细胞系的最有效治疗方案。相比之下，如果不考虑蛋白质-蛋白质相互作用的知识，则细胞死亡反应性的预测效果较差（55%和 36%的正确预测）。我们还生成了关于是否可以针对抗凋亡 Bcl-2 家族成员或 X 连锁凋亡抑制剂蛋白（XIAP）进行数学预测，以增强个别细胞系中 TRAIL 的反应性。随后的实验利用药理学 Bcl-2/Bcl-xL 抑制或基于 siRNA 的 XIAP 耗竭，证实了这些预测的准确性。因此，当在细胞凋亡调节剂的网络水平相互作用的背景下研究其表达模式时，我们证明可以可靠地预测大量黑色素瘤细胞系对 TRAIL 或 DTIC 的细胞死亡反应性。在未来，预测细胞水平的反应能力可能有助于癌症治疗的个体化。
