Fundamental Research in Cellular Metabolism
Historically, metabolism has been extensively studied at the level of differentiated organs (brain, liver, muscle), but less so at the level of single cells, particularly during proliferation. Cells undergoing replicative division must duplicate their biomass in order to produce daughter cells, and this presents a number of challenges not faced by post-mitotic organs. These challenges are met through the combined activities of numerous metabolic pathways acting in concert to convert simple nutrients (sugars, amino acids, etc.) into macromolecules (proteins, lipids, and nucleic acids). These activities are orchestrated in part by growth factor-stimulated signal transduction pathways, which direct cells to take up abundant nutrients and allocate them into the proper metabolic pathways, and other factors yet to be defined. We want to know which metabolic pathways are necessary to support physiological states of proliferation such as embryogenesis, tissue remodeling and activation of the immune system, and whether/how these pathways become dysregulated in pathological growth states like cancer. We use a combination of techniques in molecular biology, cell biology and biochemistry, coupled with metabolomics and metabolic flux analysis. By identifying the crucial metabolic activities that propel pathological cell growth, we hope to develop novel methods to treat cancer and other diseases.
A major goal of the laboratory is to define the full complement of metabolic activities that can support tumor cell growth. This information will provide a necessary foundation for developing therapeutic strategies to capitalize on cancer metabolism. More fundamentally, it will provide a framework to understand new mechanisms of metabolic regulation. For example, we previously identified four major modes through which the tricarboxylic acid cycle can operate in support of cancer cell survival and growth (Fig. 1). To generate a more systematic analysis of cancer cell metabolism, we analyzed flux from glucose through de novo serine synthesis in ~80 lung cancer cell lines. This demonstrated a nearly 50-fold range in the fraction of serine that was synthesized de novo as opposed to imported from the extracellular serine pool (Fig. 2). Molecular characteristics of these cell lines (mutational status, gene expression, growth rate, sensitivity to chemotherapy, etc) can now be compared among groups with high/low serine biosynthesis to identify correlations that will stimulate novel areas of research. The same type of analysis will be applied to additional metabolites from other informative branches of metabolism.
A related area of research aims to minimize some of the assumptions and over-simplifications associated with culture-based models of cell growth. Typically, metabolism is analyzed in unsynchronized populations of cells growing in a two dimensional monolayer attached to plastic. Implicit in this approach is the incorrect assumption that all of the cells on the dish are biologically equivalent. One obvious problem is that cells are distributed into distinct phases of the cell cycle, and that activity of various metabolic pathways may fluctuate as cells progress around the cycle. We have developed methods to monitor metabolic activities in precise stages of the cell cycle, and are now addressing which activities change dramatically from stage to stage. Second, tumorigenesis involves the formation of three-dimensional structures in which not all cells have equal access to oxygen and other nutrients. We developed a system to detect and quantify changes in metabolic flux associated with three-dimensional growth, and to identify novel metabolic vulnerabilities elicited by this state.
Methods to Analyze Tumor Metabolism in Vivo
A major challenge in cancer research is to understand which metabolic pathways operate in live tumors growing in a native microenvironment. We and our collaborators at UT-Southwestern have developed methods to address this problem in both mice and human patients. This involves introducing isotope-labeled nutrients (e.g. 13C-glucose) intravenously for a few hours prior to resection of the tumor tissue. Metabolites are then extracted and analyzed for abundance and 13C enrichment, enabling models of bona fide tumor metabolic flux to be generated. Applying this approach to human patients, we determined that high-grade gliomas avidly oxidize glucose in the mitochondria (Fig. 3). They also use a number of other glucose-dependent metabolic activities, some of which could not have been predicted simply by studying the metabolism of glioma cell lines in culture. The use of mouse models enables metabolic preferences to be compared between the tumor and surrounding tissue; in gliomas, this revealed the unexpected finding that tumors contain a large glutamine pool and small glutamate pool relative to the normal brain. Work is underway to apply this same approach to other types of cancer, both in mice and in humans.
Novel Methods to Detect and Monitor Abnormal Metabolic States
We are interested in the possibility that metabolism can be monitored non-invasively using techniques similar to MRI. There are at least two approaches to tackle this challenge. In the first, a static but quantitative view of a target metabolite is generated through proton magnetic resonance spectroscopy (MRS), in effect imaging the pool of a metabolite of interest. Conventional MRS has been used for many years to quantify a small handful of abundant metabolites (lactate, choline, N-acetyl-aspartate, etc). Recently, higher-field MRI has made it possible to image many more molecules. A broader and more quantitative view of the metabolome should make it possible to predict disease states with better precision. One application of this technology at UT-Southwestern has been to quantify the oncometabolite 2-hydroxyglutarate (2HG) within gliomas (Choi et al, Nat Med 18: 624-629 (2012)). Noninvasive 2HG detection demonstrates essentially 100% correlation with mutations in the genes encoding isocitrate dehydrogenase 1 or 2 (IDH1/2).
A second approach involves dynamic nuclear polarization (hyerpolarization) of metabolites labeled with stable isotopes, particularly 13C. Hyperpolarization involves a temporary redistribution of the populations of available energy levels into a non-equilibrium state, enabling a massive gain in magnetic resonance signal. For 13C, this gain can exceed 10,000-fold, making it possible to image both the 13C-labeled substrate and its metabolic products within seconds (Fig. 4). Because the abundance of both the substrate and the products can be quantified in this technique, it is possible to determine definitive flux measurements from intact tissue (Merritt et al, PNAS USA 104:19773-7 (2007)). We are using hyperpolarization to probe metabolism in cancer cells, tumors, and mouse models of metabolic disease.