Tumor cells reshape their metabolism to support uncontrolled proliferation, and the ways in which cells adapt metabolic networks to enable malignant behavior are determined by both genotype and cellular environment. Recent progress in genomic technologies has allowed significant advances in understanding how genetic and epigenetic changes impact proliferation across diverse tumor types. In contrast, relatively little s known about how even basic environmental factors and nutrients determine the biomass composition, metabolism, growth, and proliferation of tumor cells. Although several major pathways for nutrient usage have been identified, their relative contribution in different environments is not well understood. The overarching goal of the proposal is to address this fundamental gap in our knowledge using comprehensive experimental measurements and probabilistic modeling of metabolic fluxes. By controlling and systematically varying cell environment across multiple cancer cell lines we will investigate how different environmental factors affect metabolic phenotypes. The guiding hypotheses of the proposal, supported by our preliminary results, is that: (I) availability of key nutrients (such as glucose, lipids and amino acids) and environmental factors (such as hypoxia and pH) play major roles in determining the cellular composition and metabolic flux distributions and (II) in similar environmental conditions these major metabolic phenotypes will be mostly conserved across cells with diverse genetic backgrounds. As part of this work, we will also systematically investigate how the metabolic properties of the same tumor cells change in different environments. In addition to significantly advancing our basic understanding of cancer metabolism, the proposed research will allow improved targeting of metabolic pathways with anticancer therapies. Because of the metabolic complexity of mammalian cells, it is not currently possible to comprehensively define the metabolic state of a cancer cell by direct measurement of all internal metabolic fluxes. Therefore, we will pursue an integrated experimental-computational approach. Towards that end, we have extended the popular flux balance analysis (FBA) framework using a principled probabilistic approach based on Markov Chain Monte Carlo (MCMC) sampling of fluxes. The probabilistic FBA (pFBA) can be used to obtain correct posterior distribution for all fluxes, while imposing diverse experimental, biochemical, and energetic constraints during the sampling procedure. Importantly, pFBA provides probability distributions (i.e. estimates uncertainty) for essentially all model predictions, including those resulting from genetic or environmental perturbations. Our preliminary results demonstrate that pFBA will allow us to obtain narrowly bound estimates of major metabolic fluxes using the proposed experimental measurements.
In contrast to the wealth of genetic information pertaining to cancers, we lack even the basic understanding of how the metabolic phenotype of cancer cells depends on their environment. In the proposed research, we will obtain comprehensive experimental data from multiple cancer cell lines across diverse environmental conditions. We will integrate these data using a novel computational framework in order to build predictive models connecting the metabolic phenotype of cancer cells and their environment. This research will greatly advance our knowledge of how cancer cells rewire their metabolism in response to changes in their environment.
