The evolution of genome-scale models of cancer metabolism

Abstract

The importance of metabolism in cancer is becoming increasingly apparent with the identification of metabolic enzyme mutations and the growing awareness of the influence of metabolism on signaling, epigenetic markers, and transcription. However, the complexity of these processes has challenged our ability to make sense of the metabolic changes in cancer. Fortunately, constraint-based modeling, a systems biology approach, now enables one to study the entirety of cancer metabolism and simulate basic phenotypes. With the newness of this field, there has been a rapid evolution of both the scope of these models and their applications. Here we review the various constraint-based models built for cancer metabolism and how their predictions are shedding new light on basic cancer phenotypes, elucidating pathway differences between tumors, and dicovering putative anti-cancer targets. As the field continues to evolve, the scope of these genome-scale cancer models must expand beyond central metabolism to address questions related to the diverse processes contributing to tumor development and metastasis.

Keywords: cancer; constraint-based modeling; data analysis; metabolism; modeling and simulation; omics; systems biology; warburg effect.

Figure 1

The conceptual evolution of constraint-based models of cancer metabolism. (A) Clonal evolution commonly occurs in a developing tumor as new mutations are acquired that confer increased growth capabilities to new mutants. At the time of diagnosis, a tumor often consists of a mixed population of cancerous cells. (B) Similarly, the scope and specificity of cancer metabolic models have rapidly evolved over the past few years. Genome-scale metabolic network reconstructions have provided a valuable resource, since they contain thousands of known human metabolic reactions (Duarte et al., ; Ma et al., ; Thiele et al., 2013). This knowledge enabled the first two cancer-specific metabolic models, which focused on core metabolic pathways(Resendis-Antonio et al., ; Vazquez et al., 2010). In 2011, the first genome scale model of general cancer metabolism was used to provide insights into the Warburg effect(Shlomi et al., 2011). Shortly thereafter, transcriptomic data from the NCI-60 cell lines were used to build a general genome-scale model of cancer metabolism, which was used to assess metabolic drug targets(Folger et al., 2011). Now, numerous additional models have been built using data from specific cell lines and tumors. These models have elucidated pathways that differ between tumors(Agren et al., ; Wang et al., 2012), identified pathways that are differentially regulated with changes in estrogen receptor or p53 expression(Jerby et al., ; Goldstein et al., 2013), and predicted potential anti-cancer drug targets(Folger et al., ; Frezza et al., 2011b).

Figure 2

Expanding the reach of genome-scale metabolic models for studying cancer. (A) Genome-scale metabolic models serve as biochemically-supported data-integration platforms. In the metabolic network, metabolomic data can be associated with metabolites, while genomic, transcriptomic, proteomic, and related data types can be associated with metabolic reactions. Phenotypic measurements can be used to constrain properties of the network such as growth rate under certain experimental conditions. (B) The various Hallmarks of Cancer either affect metabolism or are modulated by metabolic changes. Therefore, modeling techniques are needed to account for these interactions between metabolism and other cell processes. Panel B adapted from (Hanahan and Weinberg, 2011) and (Kroemer and Pouyssegur, 2008) with permission.