(188p) Critical Analysis of Methodologies Based on Fluxomics for Identifying Active Elementary Flux Modes

Elementary flux modes (EFM) are a stoichiometric model that calculates all direct pathways that keep a cellular metabolism at steady state. They are mostly used for identifying pathways, assessing the structural robustness of the metabolic network, identifying the pathways with theoretical optimal product yield, verifying the importance of reactions, and computing minimal cut sets. Considering all these applications, EFM are an important tool to aid determining genetic engineering targets. They are considered an unbiased method, since they can describe the complete space of possible pathways. This characteristic can be seen as an advantage when compared to biased methods, like flux balance analysis, but it can also complicate their analysis. The number of EFM of a metabolic network increases exponentially with the number of reactions and, for very large metabolism, their calculation may not be possible due to computational limitation. For example, a metabolic network with approximately 100 reactions can reach hundreds of thousands of elementary flux modes. Hence, it is very difficult to analyze and interpret the complete set of EFM. However, for a certain physiological state, only a few elementary flux modes are active, so concentrating the analysis on them can be a way to overcome this issue. Since any flux distribution of a metabolic network at steady state can be described as a linear combination of elementary flux modes, some authors have proposed several methodologies to determine these active EFM from reaction flux data. The aim of this work is to evaluate them by analyzing their performance and general applicability. The studied methodologies were chosen based on their relevance, assessed by the year of publication and number of citations, and on what type of information they require, as only methods that use just EFM and a flux vector were selected. After a critical screening of their mathematical formulation, five methodologies were chosen to be applied on the central metabolism of wild-type Escherichia coli. The metabolic network and the complete flux vector were obtained from literature. As this problem has a large number of degrees of freedom, most methodologies formulate an optimization problem and use the relationship between EFM and the flux vector as restriction to estimate the contribution of each EFM to the flux distribution. At first, these methodologies were applied using only the flux vector of external metabolites. All of them provided different answers and, although they were mathematically consistent, none of them had a biological explanation. When an estimated flux vector calculated using the estimated contributions were compared to the complete flux distribution, it was clear that all methodologies performed poorly, indicating that the results were arbitrary, i.e., they were chosen among infinite possibilities following a mathematical assumption and have no biological meaning. When the complete flux vector was used to identify active elementary flux modes, those methodologies resulted again in different answers, showing that this problem is still under-determined. This implies that flux data alone is not sufficient to actually identify which pathways are being used. Also, this result can be seen as evidence of the robustness of metabolic networks.

ACKNOWLEDGEMENT: J. G. C. Gomez would like to express his acknowledgement to FAPESP for its financial support through grant number 2013/50357-2