Understanding global reprogramming of central carbon metabolism in cancer.


Dennis Vitkup, Matthew Vander Heiden


In Progress


Metabolic reprograming is now considered to be a hallmark of neoplastic transformation. It is clear that in order to spur uncontrolled proliferation, cancer uses multiple mechanisms to effectively hijack cellular metabolism. Therefore, understanding the major molecular mechanisms of metabolic transformation across diverse tumors is of primary importance for advancing our knowledge of cancer biology and expanding recently rediscovered potential of anticancer therapeutics aimed at metabolic targets. Notably, multiple studies independently converged on serine and glycine metabolism as an important driver of cancer growth. Unfortunately, existing studies usually focus on individual genes, and it is currently unclear why these specific pathways are essential for tumor growth and exactly how they support cell proliferation. Thoroughly validated simulation methods, such as flux-balance analysis, are now available, as well as complementary metabolomics and fluxomics approaches that can be used to constrain and test the models. 

Considering the clear importance of the problem, now is the right time to perform the proposed research. Briefly, we propose to measure key metabolic fluxes for a collection of cancer cell lines and use this information to develop constraint-based models of the major metabolic transformations that drive tumor growth. The main hypothesis we want to test is that metabolism of serine and glycine, in cooperation with glutamine metabolism, acts as an engine of cancer growth by feeding folate cycles and associated nucleotide metabolism, which is essential for rapid cell proliferation. Specifically, we propose to: 

Aim 1Use a combination of mass spectrometry and classical biochemical approaches to experimentally measure a set of key exchange and internal metabolic fluxes for a diverse collection of cancer cells lines.

Aim 2. Use the measured fluxes to build global flux-balanced (FBA) models of cancer metabolism by performing constraint-based sampling of metabolic flux distributions. We will use the models to predict the effects of all possible single and double metabolic gene deletions on cancer growth. 

Aim 3. Test and validate the models by performing knockdown experiments of key metabolic genes in the serine, glycine, and folate pathways. To close the experimental-computational loop we will adjust the models according to experimental results by identifying flux distributions that are consistent with the knockdown experiments.


We are working on the construction of accurate flux balance models of cancer metabolism. Encouragingly, preliminary results from computational modeling agree well with several key observations in cell lines and model systems. Specifically, the FBA modeling and analysis of expression data [1] predicted the observed high rate of glutamine uptake very well. The glutamine consumption rate exceeds nucleotide biosynthesis by almost an order of magnitude, and thus the donation of its γ-nitrogen to nucleotides accounts for only a small fraction of total glutamine consumption. It is likely that a major role of glutamine is to donate its α-nitrogen or carbon skeleton in the form of glutamate. Notably, we found that the conversion of glutamine to glutamate is mostly catalyzed by glutaminase and that this reaction primarily occurred in cytoplasm. 

The model also predicted the observed alanine excretion and de novo synthesis of aspartate. Alanine is produced from pyruvate through the action of alanine transaminase (GPT), the most important function of which is to convert glutamate into α-ketoglutarate. We also found that the complete oxidation of glutamine carbon involved exit from the TCA cycle as malate, conversion to pyruvate and then acetyl-CoA, and then re-entry into the cycle. Glutamine also contributed carbon to lipogenic acetyl-CoA through a second pathway, the reductive metabolism of α-ketoglutarate by cytoplasmic NADP+ dependent isocitrate dehydrogenase (IDH1), as proposed by recent studies for tumor cells under hypoxia or with defective mitochondria. In fact, this reductive glutamine metabolism was used as the primary route through which glutamine-derived α-ketoglutarate was converted to lipogenic acetyl-CoA (~85%), in agreement with previously measured values for multiple cell lines.


Hu J, Locasale JW, Bielas JH, O'Sullivan J, Sheahan K, Cantley LC, Vander Heiden MG, Vitkup D. Heterogeneity of tumor-induced gene expression changes in the human metabolic network. Nat Biotechnol. 2013 Jun;31(6):522-9.