Spinosa Phillip C, Kinnunen Patrick C, Humphries Brock A, Luker Gary D, Luker Kathryn E, Linderman Jennifer J
Department of Chemical Engineering, University of Michigan, 2800 Plymouth Road, Ann Arbor, MI 48109-2800 USA.
Department of Radiology Center for Molecular Imaging, University of Michigan Medical School, Ann Arbor, MI 48109 USA.
Cell Mol Bioeng. 2020 Jul 27;14(1):49-64. doi: 10.1007/s12195-020-00640-1. eCollection 2021 Feb.
CXCR4 and epidermal growth factor receptor (EGFR) represent two major families of receptors, G-protein coupled receptors and receptor tyrosine kinases, with central functions in cancer. While utilizing different upstream signaling molecules, both CXCR4 and EGFR activate kinases ERK and Akt, although single-cell activation of these kinases is markedly heterogeneous. One hypothesis regarding the origin of signaling heterogeneity proposes that intercellular variations arise from differences in pre-existing intracellular states set by extrinsic noise. While pre-existing cell states vary among cells, each pre-existing state defines deterministic signaling outputs to downstream effectors. Understanding causes of signaling heterogeneity will inform treatment of cancers with drugs targeting drivers of oncogenic signaling.
We built a single-cell computational model to predict Akt and ERK responses to CXCR4- and EGFR-mediated stimulation. We investigated signaling heterogeneity through these receptors and tested model predictions using quantitative, live-cell time-lapse imaging.
We show that the pre-existing cell state predicts single-cell signaling through both CXCR4 and EGFR. Computational modeling reveals that the same set of pre-existing cell states explains signaling heterogeneity through both EGFR and CXCR4 at multiple doses of ligands and in two different breast cancer cell lines. The model also predicts how phosphatidylinositol-3-kinase (PI3K) targeted therapies potentiate ERK signaling in certain breast cancer cells and that low level, combined inhibition of MEK and PI3K ablates potentiated ERK signaling.
Our data demonstrate that a conserved motif exists for EGFR and CXCR4 signaling and suggest potential clinical utility of the computational model to optimize therapy.
CXCR4和表皮生长因子受体（EGFR）代表了两类主要的受体家族，即G蛋白偶联受体和受体酪氨酸激酶，它们在癌症中发挥着核心作用。虽然CXCR4和EGFR利用不同的上游信号分子，但二者均能激活激酶ERK和Akt，尽管这些激酶的单细胞激活具有明显的异质性。关于信号异质性起源的一种假说是，细胞间的差异源于外在噪声所设定的预先存在的细胞内状态的不同。虽然预先存在的细胞状态在不同细胞间存在差异，但每种预先存在的状态都决定了对下游效应器的确定性信号输出。了解信号异质性的原因将为使用靶向致癌信号驱动因子的药物治疗癌症提供依据。
我们构建了一个单细胞计算模型，以预测Akt和ERK对CXCR4和EGFR介导的刺激的反应。我们通过这些受体研究了信号异质性，并使用定量的活细胞延时成像对模型预测进行了测试。
我们发现预先存在的细胞状态可预测通过CXCR4和EGFR的单细胞信号传导。计算模型显示，同一组预先存在的细胞状态解释了在多种配体剂量下以及在两种不同的乳腺癌细胞系中通过EGFR和CXCR4的信号异质性。该模型还预测了磷脂酰肌醇-3-激酶（PI3K）靶向疗法如何增强某些乳腺癌细胞中的ERK信号传导，以及低水平联合抑制MEK和PI3K如何消除增强的ERK信号传导。
我们的数据表明，EGFR和CXCR4信号传导存在一个保守基序，并表明该计算模型在优化治疗方面具有潜在的临床应用价值。
