Genomes to Life Contractor-Grantee Workshop III
February 6-9, 2005, Washington, D.C.
Genomics:GTL Program Projects
University of Massachusetts, Amherst
30
Integrating Phenotypic and Expression Data to Characterize Metabolism in G. sulfurreducens
R. Mahadevan1, C. H. Schilling1, D. Segura2, B. Yan3, J. Krushkal3, and D. R. Lovley2* (dlovley@microbio.umass.edu)
1Genomatica, Inc., San Diego, CA;2University of Massachusetts, Amherst, MA; and 3University of Tennessee, Memphis, TN
Geobacteraceae have been shown to be important in bioremediation of uranium contaminated subsurface environments, and in harvesting electricity from waste organic matter. These applications are intricately linked to cellular metabolism, and hence, motivating the need to understand metabolism in these metal reducing bacteria. An iterative approach of mathematical modeling followed by experimentation was adopted to understand metabolism in these organisms.
A genome-scale metabolic model has been developed using the constraint-based modeling approach. Model-based analysis has revealed significant insights on the effect of global proton balance on the physiology of G. sulfurreducens and has provided explanation for the reduced yields during Fe (III) reduction. In addition, the comparison of the model predictions of the flux distributions with gene expression data was valuable in elucidating the function of genes putatively annotated as encoding for NADPH dehydrogenase. The in silico analysis of the energetics of menaquinone secretion indicated a substantial reduction in the growth rate and suggested an explanation for why Geobacteraceae predominate over other bacteria that require such electron shuttles. The initial metabolic model provided important physiological and ecological insights on the metabolism of Geobacteraceae. However, the analysis of metabolism revealed several redundant pathways in central metabolism around acetate utilization and pyruvate metabolism.
In order to further understand the role of these redundant pathways and their contribution to the overall robustness of metabolism, a combined computational and experimental approach was utilized to unravel the activity of the redundant pathways under different environmental conditions. The computational analysis of the metabolic network identified all the conditionally dependent metabolic pathways. A series of metabolic mutants in pyruvate oxidoreductase (POR), malate dehydrogenase, phosphoenol pyruvate carboxykinase (PPCK), phosphotransacetylase, was designed based on the computational analysis to resolve the activity of the redundant pathways. These mutants were characterized phenotypically under different growth conditions and the experimental data was compared with model predictions. This comparison revealed several interesting aspects of how central metabolism is regulated: POR is the only mechanism for the synthesis of pyruvate from acetate, PPCK is essential for growth on Fe(III) suggesting a potential for this enzyme to be regulated during Fe(III) reduction. These studies indicated the importance of incorporating the mechanism corresponding to the regulation of metabolism to refine the conceptual and in silico model.
Gene expression data corresponding to several environmental and genetic perturbations in G. sulfurreducens represents information that captures the activity of the regulatory network. Hence, gene expression data derived from several experiments were processed and assembled for further analysis. This expression data was filtered and then clustered based on expression similarities to identify co-expressed genes across the different perturbations. This was followed by sequence analysis including the searching the upstream regions of these co-expressed genes and operons for known transcription factor binding sites, and aligning the upstream regions to identify motifs that correspond to novel sites. This analysis revealed several potential regulatory interactions including a mechanism for regulating heat shock response, and motifs for regulation of sulfate metabolism. Further analysis with additional expression data that incorporates metabolic perturbations is expected to derive regulatory constraints for the metabolic model.
These studies reveal the potential of a combined computational and experimental strategy to iteratively characterize metabolism and the associated regulatory network. Such highly refined conceptual and in silico models of cellular metabolism will be important to design and optimize efficient strategies for bioremediation and harvesting energy from organic substrates.
* Presenting author
