Inclusive Design of Flux-Determining 13C NMR Experiments

Wednesday, October 19, 2011
Exhibit Hall B (Minneapolis Convention Center)
Jonathan Meade, Chemical Engineering, Carnegie Mellon, Pittsburgh, PA, Saleem A. Khan, Microbiology and Molecular Genetics, University of Pittsburgh, Pittsburgh, PA, Mohammad M. Ataai, Chemical and Petroleum Engineering, University of Pittsburgh, Pittsburgh, PA and Michael M. Domach, Chemical Engineering, Carnegie Mellon University, Pittsburgh, PA

NMR is supposed to fill in the flux value gaps that result when extracellular measurements are insufficient for contending with the degrees of freedom and redundancy present in a metabolic network.  Many methods use several analytes and randomly search for a flux set that minimizes the error between the calculated and experimental NMR data.  While this approach can be effective, one problem that this presentation will introduce is multiple solutions have been found for the same cell/mutant under similar growth conditions. This is problematic because NMR is supposed to be the resolver of flux possibilities; hence, finding different answers in the literature for the same cell/mutant is unsettling.  We suggest that this problem can originate due to (i) coupling flux constraints to iterative error minimization rather than making full use of MFA capabilities, (ii) assuming no error in external measurements, (iii) failing to distinguish between gross error minimization and contrast between the alternate flux set solutions, (iv) extrapolating wild-type physiology to mutants, and (v) not recognizing spectrum uniqueness in terms of fluxes.  This presentation will describe a maximally inclusive formulation for the NMR experimental design problem.  The flux bounds and thus the possible flux sets are developed based on factors (i) – (iv).  Label design is then optimized for an analyte or set.  A test for whether the relative and nondiscarded information in spectra will lead to a unique flux distribution is then performed. Finally, for experimental design and then problem closure, we will show that for some problems, quasi-linearity exists where the isotopomer distribution vectors for analytes at flux extreme points can be combined in a linear fashion, which then makes the NMR data-to-flux problem a directed and bounded flux-determining problem.  The example used to illustrate the approach and computations will be determining the fluxes in an E. coli mutant that has been shown to provide increased yield of therapeutic DNA.  The yield evidence illustrates that the mutation “works”, but understanding exactly why in terms of fluxes is unresolved because two flux solutions have been reported for the mutant by NMR, where only one is consistent with flux optimization analysis.  Thus, this case study has practical and basic utility.

Extended Abstract: File Not Uploaded
See more of this Session: Poster Session: Bioengineering
See more of this Group/Topical: Food, Pharmaceutical & Bioengineering Division