Dynamical systems biology approach to identify mediators of the Epithelial-Hybrid-Mesenchymal spectrum
Cancer metastasis – the spread of cancer cells from one organ to another – remains the major cause of cancer-related deaths. A hallmark of metastasizing cells is their ability to adapt quickly and reversibly to their dynamic microenvironment. This ability to switch among different cell-states is called as phenotypic plasticity. A well-studied example of phenotypic plasticity in carcinomas (cancers originating in epithelial tissues) is Epithelial-Mesenchymal Transition (EMT) and its reverse Mesenchymal-Epithelial Transition (MET). EMT is characterized by cancer cells losing their cell-cell adhesion and gaining migration and invasion traits, enabling cancer cell dissemination. During MET, the disseminating cells, upon reaching distant organs, regain the epithelial traits, facilitating metastatic colonization. Initially, EMT and MET were considered as binary processes, but recent studies have discovered that cancer cells can acquire one or more hybrid epithelial/mesenchymal (E/M) phenotypes that can be highly aggressive and are associated with worse patient outcomes. While the molecular drivers of EMT have been extensively investigated, the molecular factors that can stabilize hybrid E/M phenotypes or drive MET are ill understood. In my work, I have used dynamical systems approach to identify two transcription factors that can stabilize hybrid E/M phenotype(s) – NFATc and SLUG – and two transcription factors that can drive MET – ELF3 and KLF4. Our computational model predictions are validated by extensive transcriptomic data analysis at both bulk and single-cell analysis levels. Further, modeling results collected over an ensemble of kinetic parameters – using a tool called RACIPE (RAndom CIrcuit PErturbation) – suggest that the role of these players in stabilizing hybrid E/M phenotype or driving MET emerges from underlying network topology rather than specific parameter values. First, I incorporated experimentally reported interactions of NFATc with key molecules influencing EMT dynamics, such as E-cadherin, SNAIL and ZEB, in a mechanism-based model. Bifurcation analysis reveals that NFATc can prevent the progression towards a full EMT and expand the parameter region for the existence of hybrid E/M phenotype. Further, RACIPE analysis demonstrated the role of NFATc in augmenting the co-existence of epithelial, hybrid E/M and mesenchymal phenotypes. Knockdown of NFATc in H1975 cells (lung cancer cells exhibiting a stable hybrid E/M state) drove them towards a complete EMT, thus validating the role of NFATc as a stabilizer of hybrid E/M state. Next, via dynamical modeling and transcriptomic data analysis, I examined the role of EMT-inducing transcription factor SLUG in mediating EMT/MET. I found that SLUG, unlike its family member SNAIL, drove a weak EMT and stabilized cells in hybrid E/M state. Overexpression of SLUG led to an enrichment of hybrid E/M state, highlighting its role in maintaining this phenotype. Second, I investigated the role of KLF4 and ELF3 through expanding abovementioned regulatory networks governing EMT/MET. Mechanism-based modeling suggested that both KLF4 and ELF3 can delay the onset of EMT, and their overexpression can drive MET, with ELF3 being a relatively more potent inducer. In both cell lines and primary tumors, KLF4 and ELF3 correlate negatively with EMT-inducing factors, and their expression is inhibited during EMT. Thus, both ELF3 and KLF4 are associated with an epithelial phenotype and are putative drivers of MET. Finally, I observed that while high levels of NFATc, SLUG, KLF4 and ELF3 associated with worse patient survival in some solid tumors, the trends were tissue specific. These observations reveal the complex association of EMT/MET with patient survival. Overall, my research showcases how a dynamical systems biology approach can help identify potent regulators implicated in phenotypic plasticity, thus suggesting putative therapeutic targets to be considered for curtailing metastasis.