The Severity of an Enzyme Mutation Strongly Influences Pleiotropy, Often Mediated through Currency Metabolite Production

We study how the severity of a mutation affects pleiotropy in genome-scale metabolic networks. We measure pleiotropy as the number of biomass components whose maximal production is reduced by the mutation [Szappanos et al, 2011]. For each biomass component, we added a new exchange reaction and maximized the corresponding secretion [Shlomi et al, 2007]. For each gene, we simulated mutations of different severity by restricting the flux through all reactions catalyzed by the gene to a fixed fraction of the wildtype flux, which we reduced in 10%-steps [Xu et al, 2012]. We then examined how the maximal production of each biomass component differs between mutant and wildtype. For a genome-scale model of E. coli metabolism [Feist et al, 2007] in minimal medium (MM) and rich medium (RM), we found that mutations of genes for 78% in MM and 82% in RM show no fitness effect. 21% and 17% of genes show a linear relationship between flux reduction and biomass production in MM and RM, respectively, while 1% shows a piece-wise linear relationship, with a steeper decline of fitness at larger mutations for both media. Mutations to the majority of genes affect no biomass components, independent of mutation severity. However, 54% of genes with fitness effects in MM (66.1% in RM) show a stepwise increase of pleiotropy with increasing mutation severity. A further 41% (MM) of genes have constant pleiotropy (33.5% in RM). Another 5% (MM) and 0.4% (RM) of genes with fitness effects affect no individual biomass components. A fraction of pleiotropy in MM is associated with the generation of currency metabolites: when we make CTP, NADH, NADPH, Ubiquinol-8 and L-Glutamate freely available, we found reduced pleiotropy for 5%, 10%, 3%, 5%, and 3% of previously pleiotropic genes, respectively. However, making other currency metabolites available (ATP, GTP, UTP, Inosine triphosphate (ITP), Flavin adenine dinucleotide reduced (FADH2), Reduced flavin mononucleotide (FMNH2), Menaquinol-8, 2-Demethylmenaquinol 8 and Acetyl-CoA) has no effect on pleiotropy. An even stronger influence of many currency metabolites is found in RM: when we make ATP, CTP, NADH, NADPH, Ubiquinol-8, L-Glutamate and FADH2 freely available, 14%, 8%, 33%, 30%, 33%, 23%, and 1% of previously pleiotropic genes show reduced pleiotropy, respectively. Again, other currency metabolites (ITP, UTP, GTP, FMNH2, Menaquinol-8, Acetyl-CoA and 2-Demethylmenaquinol 8) have no effect on pleiotropy. We conclude that a fraction of pleiotropic effects is mediated through currency metabolites, but that the majority of pleiotropy is pathway specific.