Understanding Emergent Biological Behaviors: Agent Based Simulations of In vitro Epithelial Morphogenesis in Multiple Environments
C. Anthony Hunt
University of California, San Francisco
Last modified: May 22, 2006
In vitro cultures of MDCK cells are model systems for the study of Epithelial cell growth, morphogenesis, differentiation, and transition to cancer-like forms. Environment composition and cell orientation influence the development of characteristic multicellular structures and growth features such as stable cyst formation in embedded culture, inverted cyst formation in suspension culture, and lumen formation in overlay culture. We have constructed a low-resolution, agent-based simulation model of these systems and report on its use to explore how the observed morphogenetic phenomena, under those different growth conditions might be generated and controlled. A decision tree inside each cell object provides the logic governing At each step during simulated growth, each cell object, acting independently, uses an environment-focused embedded decision tree to provide the logic that governs simulated cell behavior. Several in silico growth characteristics, including stable “cyst” formation, mimic those observed in vitro in four different simulated environments, under different operating conditions. We studied how changes in the logic used might cause abnormal growth. Simulation results confirm the importance of a polarized response to the environment to the generation of a normal epithelial phenotype. Some simulated behaviors are sensitive to rule and environment changes (the in silico analogue of epigenetic factors). In several cases, rule changes cause cancer-like growth patterns to emerge. The model just described simulates absolute, error-free mechanistic control and regulation through many generations. Error-free control is thought not to be characteristic of biological systems. We relaxed that control by providing “cells” a stochastically controlled option to follow (or not) one or more rules each time the option arose. Even modest relaxation of rule application can lead to aberrant simulated growth characteristics that, in some cases, are similar to in vitro phenotypes that are characteristic of transitions to pre-cancerous states. The results demonstrate that the in silico experimental approach used can provide a powerful means of gaining deeper insight into how system level phenotype emerges from orchestrations of lower and molecular level events.