ORIGINAL

Abstract

The solid tumor and its surrounding tumor microenvironment (TME) form an intricate dynamical system characterized by both resource and population heterogeneities. It is widely recognized that the spatiotemporal dynamics of the TME contribute to the progression of cancer and the evolution of therapeutic resistance. Nevertheless, realistic in vitro models that recapitulate the fundamental features of the TME are lacking in preclinical biomedical research. Here, we present the death galaxy, an experimental microfluidic platform that establishes a heterogeneous stress landscape over diverse cellular populations. Designed in the form of connected microhabitats based on evolutionary principles, the technology achieves a controllable chemical gradient via a static diffusion mechanism. The platform provides means for quantification of long-term cellular dynamics on the time scale of weeks at single-cell resolution. We further describe the application of the death galaxy to the study of prostate cancer evolutionary and population dynamics in response to the cytotoxic chemotherapeutic docetaxel. From an evolutionary lens, the phenotypic emergence of polyploidal giant cancer cells (PGCCs) acts as a reservoir for drug resistance and a cancer persistence strategy. We find that the death galaxy facilitates the robust survival and enrichment of PGCCs in contrast to well-mixed systems, and that a novel cancer cell clustering phenomenon within the death galaxy indicates enhanced survival under high stress. On the population level, we explore a 3-week co-culture of prostate cancer cells and human bone marrow stromal cells in the death galaxy, revealing a transition in cancer cell survival as a function of drug concentration. Employing the differential equations of evolutionary game theory to model local population interactions within microhabitats, we establish an empirical payoff parameter landscape of cancer-stroma dynamics. In search for a more representative class of models to depict these population interactions, the stochastic interacting particle systems (IPS) perspective is discussed and further extended to incorporate arbitrary landscapes and mutant evolution. Overall, this work illustrates the potential of biologically-representative in vitro microfabricated platforms combined with theoretical modeling in understanding cancer dynamics.