Multiscale Modeling to Optimize Inhibition of Oncogenic ERK Pathway Signaling

项目总结/摘要 靶向RAF/MEK/ERK通路的小分子抑制剂已成为精准医学的有效工具， 但它们的临床疗效在RAS和BRAF突变癌症的多样性中高度可变。即使在 对于易感的癌症，这些抑制剂很少产生持久的反应。研究抵抗的原因， 包括RAF抑制剂对ERK通路“反常”激活，揭示了 由RAF和ERK途径激酶组成的复杂网络。这些复杂性限制了我们 了解和预测靶向治疗的有效性，特别是联合治疗-尽管几十年来， 密集的研究，包括数学建模。准确的预测不仅需要了解 蛋白激酶调节的分子复杂性和网络的复杂系统水平行为， 激酶的组成，而且这两个水平的控制是如何耦合。准确预测的挑战 因此，靶向治疗及其组合的有效性需要分子生物学的融合， 和系统生物学方法。这里提出的系统生物学项目旨在确定最佳的 激酶抑制剂的组合，通过机制模型，整合了对两者的理解： 1)激酶抑制剂的构象选择性--影响其结构、热力学和动力学性质 靶向激酶;和2）系统级网络性质，包括反馈环、突变和 激酶/支架丰度，其可以修饰反馈环并允许通常无关紧要的激酶 同种型以补偿同种型特异性激酶抑制。结合这些特征， 直接将分子/结构和网络方面联系起来的细胞信号建模方法， 抑制剂和它们的组合可以有效地抑制致癌信号传导， 恢复、生长和耐药性。我们建议开发这样的下一代多尺度模型， 致癌ERK信号传导和药物反应，并建立一个新的概念基础，发现 通过整合结构、热力学和动力学信息， 短时间尺度的分子动力学（MD）与长时间尺度的系统级动力学建模。我们将测试 通过整合和迭代建模和实验研究，我们的模型预测严格。实验 研究将开始于具有确定的突变差异的成对的同基因癌细胞系中。一旦模型预测 是适当稳健的，我们将进展到癌细胞系组，然后到细胞系衍生的体内异种移植物， 然后是患者来源的异种移植物和黑色素瘤的基因工程小鼠模型（GEMM）， 这是临床综合预测的前兆。我们将确定是否通过两个（或更多） 具有不同构象选择性的抑制剂-在我们的初步数据中似乎很有希望-通常 适用，可以与抑制途径内的不同靶点组合，并且可以在一个特定的水平上理解。 详细的机械水平使用我们的多尺度模型。

PROJECT SUMMARY/ABSTRACT Small molecule inhibitors targeting the RAF/MEK/ERK pathway have become potent tools in precision medicine, but their clinical efficacy is highly variable across the diversity of RAS- and BRAF-mutated cancers. Even in susceptible cancers, these inhibitors rarely give durable responses. Studying the causes of resistance, which include ‘paradoxical’ ERK pathway activation by RAF inhibitors, has revealed complex molecular adaptations in the complicated networks comprised of RAF and ERK pathway kinases. These complexities limit our ability to understand and predict effectiveness of targeted therapies, especially in combination – despite decades of intense study, including mathematical modeling. Accurate predictions require understanding not only of the molecular complexities of protein kinase regulation and the intricate systems-level behavior of the networks that kinase constitute, but also of how these two levels of control are coupled. The challenge of accurately predicting effectiveness of targeted therapies and their combinations therefore demands an amalgamation of molecular and systems biology approaches. The systems biology project proposed here aims to identify optimal combinations of kinase inhibitors through mechanistic models that integrate understanding of both: 1) Conformation selectivity of kinase inhibitors – affecting structural, thermodynamic and kinetic properties of the targeted kinase(s); and 2) Systems-level network properties, including feedback loops, mutations and kinase/scaffold abundances, which can modify feedback loops and allow normally inconsequential kinase isoforms to compensate for isoform-specific kinase inhibition. Combining these features necessitates novel approaches to modeling cell signaling that directly link molecular/structural and network facets to predict which inhibitors and their combinations can efficiently suppress oncogenic signaling while disabling or delaying signal recovery, growth, and drug resistance. We propose to develop such next-generation multiscale models of oncogenic ERK signaling and drug responses, and to establish a new conceptual foundation for discovering effective drug combinations by integrating structural, thermodynamic and kinetic information – and combining short time-scale molecular dynamics (MD) with long time-scale modeling of systems-level dynamics. We will test our model predictions rigorously by integrating and iterating modeling and experimental studies. Experimental studies will begin in paired isogenic cancer cell lines with defined mutational differences. Once model predictions are suitably robust, we will progress to panels of cancer cell lines, then to cell line-derived xenografts in vivo, and then to patient-derived xenografts and genetically engineered mouse models (GEMMs) of melanoma – as a presage to clinically integrated predictions. We will determine if the strategy of hitting a kinase by two (or more) inhibitors with distinct conformation selectivity – as appears promising in our preliminary data – is generally applicable, can be combined with inhibition of different targets within a pathway, and can be understood at a detailed mechanistic level using our multiscale models.