(314d) Using Multi-Omics Data to Estimate Parameters In Enzymatic Rate Laws | AIChE

(314d) Using Multi-Omics Data to Estimate Parameters In Enzymatic Rate Laws


Cotten, C. - Presenter, University of Wisconsin-Madison
Reed, J. L. - Presenter, University of Wisconsin-Madison

In recent years, studies have gathered global measurements of the components of cells, yielding concentrations of RNA, protein, and metabolites, as well as estimated metabolic fluxes. However, few studies have used a metabolic model to analyze these metabolic measurements; instead, statistical methods and other techniques were used to draw conclusions. In this study, we examined fluxomic, proteomic, and metabolomic data measured by Ishii et. al. [1] for Escherichia coli mutants using a constraint-based model at steady state with kinetic constraints. Functional forms for central metabolic enzyme rate laws from a previous study by Chassagnole et. al. [2] were used as constraints on the flux through enzymes.

Kinetic parameters and their confidence intervals for the included rate laws were first determined using maximum likelihood parameter estimation. Measurements from 25 distinct growth conditions were used for estimating parameters, including varied growth rates and gene deletion strains with defective metabolic pathways. The newly-parameterized rate laws provide a more effective fit to the experimental data than the previously-reported parameters estimated by Chassagnole et. al. [2]. In addition, we found that an equivalent fit to the data can be achieved by a much simpler model using fewer than half of the parameters from the original rate laws proposed by Chassagnole et al. This simplified model requires fewer computational resources to solve than the previous model, and we found that the parameter confidence intervals are two orders of magnitude tighter than parameter confidence intervals for the previously-reported parameters [3]. The simpler model is preferred for its equivalent predictions, reduced complexity, and better-resolved parameters.

The resulting kinetic model was used to draw conclusions about the state of reactions in E. coli central metabolism during aerobic growth on glucose. Proximity to equilibrium was examined for reactions in the central metabolism. Triose phosphate isomerase, phosphoglucomutase, ribulose phosphate 3-epimerase, and the transketolase reactions were found to be near equilibrium in all conditions, and the remainder of the reactions in the core metabolism were not uniformly near equilibrium. Reactions that are not at or near equilibrium are generally considered to be limited by kinetics and regulation, and are potential targets for metabolic engineering strategies. In addition to estimating proximity to equilibrium, the degree of enzyme saturation was also determined for metabolic reactions where kinetic parameters were estimated. Eight out of 31 substrates were above concentration levels needed to saturate their respective enzymes, while four others had concentrations near their respective binding coefficients, indicating that these substrates exert fine and measurable control on their respective enzymes. Two enzymes also appeared to be significantly influenced by concentrations of their activators. The implications and potential uses for the employed model construction strategy and the constructed E. coli central metabolic model will be discussed.


[1] N. Ishii, K. Nakahigashi, T. Baba, et. al. "Multiple high-throughput analyses monitor the response of E. coli to perturbations," Science, vol. 316, Apr. 2007, pp. 593-7.

[2] C. Chassagnole, N. Noisommit-Rizzi, J.W. Schmid, K. Mauch, and M. Reuss, "Dynamc modeling of the central carbon metabolism of Escherichia coli," Biotechnology and Bioengineering, vol. 79, Jul. 2002, pp. 53-73.

[3] R.N. Gutenkunst, J.J. Waterfall, F.P. Casey, K.S. Brown, C.R. Myers, and J.P. Sethna, "Universally sloppy parameter sensitivities in systems biology models," PLoS computational biology, vol. 3, Oct. 2007, pp. 1871-78.