One of the core tenets of our lab is that the study of metabolic fluxes and their relationship to other "omics" data is indispensible for understanding cellular physiology. Unlike metabolite, protein and mRNA measurements, which provide information about what is present inside the cell at a particular time, fluxes give insights into what the cell is actually doing with these molecules. They give us a traffic report that describes the flow of material throughout metabolism. This understanding can in turn provide clues as to how the cells should be manipulated in order to elicit a desired phenotype.
The flux maps we generate rely upon the marriage of precise experimental techniques and sophisticated computational algorithms for subsequent data analysis. Tracer molecules that incorporate stable isotopes (e.g. 13C or 2H) at specific positions are used to probe the flux state of the cell. By analyzing the rearrangement of these heavy atoms in excreted products and intracellular metabolites, we can extract information about the fluxes that gave rise to the observed labeling patterns. Our method of choice for measurement of isotopic labeling is GC-MS due to its high sensitivity and versatility. Through clever choices of chemical derivatizations and GC conditions, it is possible to selectively analyze specific classes of molecules or to canvass a wide range of metabolites simultaneously. Furthermore, we can control the type and extent of molecular fragmentation that occurs in the MS by varying the derivatization method and MS parameters, thus offering a wealth of information on labeling patterns and a high degree of redundancy for detecting inconsistent measurements.
The GC-MS data alone can be used to infer precursor-product relationships and to draw qualitative conclusions about the distribution of flux in the metabolic network. This is not a new idea and has in fact been applied within the biological community for many decades. However, recent advances spearheaded by current and past members of our research group have led to the development of numerical methods that process the mass spectral data in order to determine quanititative flux estimates. The methods rely upon specification of the network stoichiometry and atom transitions to simulate the GC-MS measurements that would arise from a particular flux distribution. By systematically adjusting these flux estimates until the mismatch between simulated and observed measurements is minimized, the computed solution eventually converges to a set of fluxes that best fit the available data. The software package Metran developed within our lab incorporates robust algorithms for solving the flux estimation problem. In addition, it provides routines for characterizing the optimal solution, including nonlinear statistical methods for assessing the goodness-of-fit and for computing rigorous confidence intervals on estimated fluxes. Optimal experiment design is also facilitated by using the software to simulate the outcome of hypothetical isotope labeling experiments.
The computational and experimental techniques developed within our group have been applied to estimate fluxes in a variety of biological contexts, including amino acid production in Corynebacterium glutamicum and 1,3-propanediol production in genetically modified E. coli strains. More recently, we have extended these methods to study model systems that are relevant to understanding human physiology and pathophysiology, such as lipogenesis in adipocyte cell lines and gluconeogenesis in primary hepatocyte cultures. Future work will focus on whole animal tracer experiments investigating gluconeogenesis, insulin resistance and type-2 diabetes in mice.
- Jose Aleman
- Jamey Joung
- Jason Walther