A Mechanistic Modeling Framework Reveals the Key Principles Underlying Tumor Metabolism

While aerobic glycolysis, or the Warburg effect, has for a long time been considered a hallmark of tumor metabolism, recent studies have revealed a far more complex picture. Tumor cells exhibit widespread metabolic heterogeneity, not only in their presentation of the Warburg effect but also in the nutrients and the metabolic pathways they are dependent on. Moreover, tumor cells can switch between different metabolic phenotypes in response to environmental cues and therapeutic interventions. A framework to analyze the observed metabolic heterogeneity and plasticity is, however, lacking. Using a mechanistic model that includes the key metabolic pathways active in tumor cells, we show that the inhibition of phosphofructokinase by excess ATP in the cytoplasm can drive a preference for aerobic glycolysis in fast-proliferating tumor cells. The differing rates of ATP utilization by tumor cells can therefore drive heterogeneity with respect to the presentation of the Warburg effect. Building upon this idea, we couple the metabolic phenotype of tumor cells to their migratory phenotype, and show that our model predictions are in agreement with previous experiments. Next, we report that the reliance of proliferating cells on different anaplerotic pathways depends on the relative availability of glucose and glutamine, and can further drive metabolic heterogeneity. Finally, using treatment of melanoma cells with a BRAF inhibitor as an example, we show that our model can be used to predict the metabolic and gene expression changes in cancer cells in response to drug treatment. By making predictions that are far more generalizable and interpretable as compared to previous tumor metabolism modeling approaches, our framework identifies key principles that govern tumor cell metabolism, and the reported heterogeneity and plasticity. These principles could be key to targeting the metabolic vulnerabilities of cancer. Author SummaryTumor cells exhibit heterogeneity and plasticity in their metabolic behavior, relying on distinct nutrients and metabolic pathways, and switching to reliance on different pathways when challenged by an environmental change or a drug. While multiple previous studies have focused on identifying metabolic signatures that can distinguish tumor cells from non-tumorigenic ones, frameworks to analyze the metabolic heterogeneity in tumors have been lacking. Here, we present a mechanistic mathematical model of some of the key metabolic pathways active in tumor cells and analyze the steady state behaviors the model can exhibit. We find that the rate of ATP use by tumor cells can be a key determinant of the metabolic pathway via which tumor cells utilize glucose. We further show that tumor cells can utilize different pathways for satisfying the same metabolic requirements, and explore the implications of such behavior for the response of tumor cells to drugs targeting tumor metabolism. At each step, we discuss how our model predictions fit within the context of experimental observations made across tumor types. The present modeling framework represents an important step towards reconciling the wide array of experimental observations concerning tumor metabolism, and towards a more methodical approach to targeting tumors metabolic vulnerabilities.