Figure 4.

Microenvironment and multi-level selection model extensions. Allowing the microenvironment to determine tumor-propagating cells still resulted in the evolution of (A) high tumor-propagating cell levels and (B) high tumor-propagating self-renewing cell division probabilities. As a control for the microenvironment model, we ran simulations in which the probability for tumor-propagating cell self-renewal could not evolve. (C) The final percent of tumor-propagating cells is smaller when transient cells contribute to the fitness of clonally related tumor-propagating cells than when they do not, and (D) the final proportion of symmetric divisions does not increase (initial value: 0.05, final value: mean = 0.502, SEM = 0.002). Error bars show standard error of the mean, and 50 simulations were initialized per condition (n = 50 for all conditions).