Abstract B2-38: Learning the dynamic response to Ras-ERK signaling with paired reporters

Abstract

Abstract To characterize key cell signaling pathways in cancer, we employ a modular strategy to examine how signals propagate between two members in a pathway, using paired time series measurements in live cells. The two pathway members chosen bracket a sub-network for which we measure input and output signals. In this study, we characterize models for the response of FRA1 (Fos related antigen 1) to ERK (extracelluar regulated kinase) activity. Mathematical response models are constrained by the paired signal data to produce a predictive module for the ERK-to-FRA1 sub-network, complementing existing modeling efforts as a building block for complete models. Background: Many mutations in cancer impair cellular information processing by hyperactivating signaling pathways associated with proliferation and migration. However, targeted inhibitors for hyperactivated elements have not yielded consistent responses, and acquired resistance hinders their use in therapy. Cell signaling can be a complex dynamic process [1], the detailed understanding of which will support development of new research directions and clinical therapies. The Ras-ERK signaling pathway regulates cell proliferation and migration, and is often mutated in cancer. ERK activity (the nominal pathway output) regulates major drivers of cell decision making, such as the transcription factor FRA1 which is involved in migration and invasion [2]. FRA1 is observed being regulated by ERK at three levels: (1) transcription via ERK-dependent phosphorylation of transcription factors; (2) translation via promotion of mTORC1 (mammalian target of rapamycin complex 1), and (3) stability via ERK-dependent phosphorylation of a c-terminal destabilization motif. The interplay of multiple regulation points in this feed-forward manner may lead to complex responses, including behavior as a ‘persistence detector’ for ERK activity, where only long term activity produces strong responses because short pulses fail to stabilize the protein. Experimental method: To make paired measurements, we generate a cell line expressing both a fluorescent reporter for ERK activity, and a fluorescent FRA1 fusion protein on the genomic locus. Cells are treated with varied levels of extracellular ligands and therapeutic drugs to elicit varied ERK responses, and time series images taken for up to 72 hours. Segmentation and particle tracking yield individual cell time courses over the length of the experiment. Modeling: Models of the FRA1 response to ERK activity are considered at several levels of abstraction to evaluate the complexity of the response against a mechanistic model. The most parsimonious model, that which satisfies data with the least complexity, can then predict FRA1 levels based on ERK activity, with minimal bias from overfitting. Fitting individual cell time courses produces the distribution of cell responses over populations. Conclusions: The characterized response models demonstrate the predictive capacity for FRA1 from ERK activity, revealing that the responsiveness to ERK varies from cell to cell. Over the cell populations observed, the models capture the distribution of the response. These models augment the existing body of work describing the ERK activity resulting from receptor stimulation, and may be further combined with future modular models toward a predictive global model.