• genetic interaction;
  • metabolic flux;
  • metabolomics;
  • system biology;
  • transaldolase

Central carbon metabolism is a basic and exhaustively analyzed pathway. However, the intrinsic robustness of the pathway might still conceal uncharacterized reactions. To test this hypothesis, we constructed systematic multiple-knockout mutants involved in central carbon catabolism in Escherichia coli and tested their growth under 12 different nutrient conditions. Differences between in silico predictions and experimental growth indicated that unreported reactions existed within this extensively analyzed metabolic network. These putative reactions were then confirmed by metabolome analysis and in vitro enzymatic assays. Novel reactions regarding the breakdown of sedoheptulose-7-phosphate to erythrose-4-phosphate and dihydroxyacetone phosphate were observed in transaldolase-deficient mutants, without any noticeable changes in gene expression. These reactions, triggered by an accumulation of sedoheptulose-7-phosphate, were catalyzed by the universally conserved glycolytic enzymes ATP-dependent phosphofructokinase and aldolase. The emergence of an alternative pathway not requiring any changes in gene expression, but rather relying on the accumulation of an intermediate metabolite may be a novel mechanism mediating the robustness of these metabolic networks.


Systematic phenotype analysis of gene-deletion mutants, combined with in silico predictions from genome-scale metabolic network models, has been used to identify new genetic interactions and previously unknown gene functions in model microorganisms. As this approach depends on a predicted or observed phenotype, genetic robustness limits the availability of gene candidates showing some phenotype under the conditions tested. Such robustness could, in part, originate from redundancy such as the presence of an isozyme or another pathway with a duplicate function. In addition, the specialized functions of many genes for specific growth conditions, such as the availability of different carbon sources, could contribute to overall robustness. Systematic deletion of two or more genes, and fitness tests of the mutants under many conditions, would be powerful systems for the discovery of new gene functions.

Using a new method employing a P1 phage derivative, we created systematic double-deletion mutants in the central carbon metabolism of E. coli. The mutants were created by combining 31 single-gene deletions (first deletion) with deletions in seven key reactions (second deletion). The seven key reactions were selected to represent each of the following pathways: glycolysis (two reactions), the pentose phosphate pathway (two reactions), the anaplerotic pathway (two reactions), and the glyoxylate shunt (one reaction). The resulting strains were then tested for growth capabilities under various nutrient conditions, including rich medium, minimal medium with 10 different carbon sources, and medium containing a combination of two carbon sources (Figure 1).

At the same time, we performed model-based prediction of the growth phenotypes of these mutants using genome-scale metabolic models. By contrasting the simulation result with the experimental result, we aimed to elucidate previously unknown reactions within this exhaustively analyzed pathway in one of the best-studied organisms.

Among 2177 double mutant experiments from which we obtained both experimental and predicted growth phenotypes, we found 39 cases in which model-based analysis predicted double mutant-specific slow-growth phenotypes, although experimental results indicated growth comparable with that of the parental single-knockout mutants.

Out of the 39 cases, we were most interested in eight cases that carried one of their deletions in transaldolase (talA talB). Xylose was used as a carbon source in five of these eight cases. Further examination of metabolic pathways indicated that transaldolase must be essential for xylose utilization through currently known reactions in central carbon metabolism (Figure 4A and B). Although one known pathway could potentially serve as a bypass for utilizing xylose in transaldolase mutants, this bypass could not explain the normal growth of several double-knockout strains such as fbp-talAB, tpiA-talAB, deoC-talAB, and deoB-talAB (Figure 4C). Thus, we decided to focus on this discrepancy to find new reaction(s).

First, we performed microarray analysis to find specifically induced genes in the transaldolase mutant growing on xylose minimal medium. However, it revealed no notable changes in the mRNA levels of genes involved in related metabolic pathways and did not suggest interesting candidates for the novel pathway. Next, we performed metabolome analysis by CE-MS, which revealed greater than 40-fold accumulation of S7P in the talAB-knockout cells and several-fold accumulation of pentose phosphates, but otherwise less than twofold change in the levels of metabolites in related pathways (Figure 4E). We also discovered accumulation of an unidentified metabolite, postulated to be S1,7P, which was previously not considered to be present in E. coli. Combined with another experimental result from the phenotype analysis that pfkA is necessary for the growth of transaldolase mutants on xylose, we hypothesized that S7P is utilized through S1,7P and degraded to DHAP and E4P in transaldolase mutants (red reactions in Figure 4F).

To test this hypothesis, we performed MFA of wild-type and talAB-knockout strains using 1-13C-xylose as the sole carbon source and measured the isotopomer distribution of intermediate metabolites by CE-MS. The wild-type and talAB-knockout strains clearly showed distinct 13C isotopomer distributions for many metabolites, and the differences were explained by the presence of the hypothesized new reactions in the talAB knockout, but not in the wild type (Figure 4F).

Finally, we validated these novel reactions at the level of enzymatic activity. Using purified recombinant PfkA, the candidate enzyme for converting S7P to S1,7P, and FbaA, the candidate enzyme for converting S1,7P to DHAP and E4P, we confirmed the conversion from S7P and ATP to (putative) S1,7P by PfkA and then to DHAP and E4P by addition of FbaA.

Thus, consistent with our hypothesis, S7P must be converted to S1,7P and then to DHAP and E4P by sequential action of the glycolytic enzyme PfkA (phosphofructokinase) and FbpA (fructose-bisphosphate-aldorase) in transaldolase-deficient cells.

The discovery of new reactions, in addition to proving the potency of a strategy combining experimental and computational phenotype analysis of large-scale multiple-knockout mutants, has two substantially important implications.

First, although the novel reactions seemed to be present only in transaldolase mutants in E. coli, other organisms might also possess these reactions. The most probable candidate organism might be another bacterium, L. lactis, which does not seem to encode transaldolase in its genome, but is known to utilize xylose through glycolysis and the pentose phosphate pathway. In higher eukaryote, some mammalian tissues known to lack transaldolase, and associated with liver cirrhosis, represent another possible candidate having the novel reactions.

Second, emergence of these alternative reactions does not require any change in gene expression, but rather relies on the accumulation of an intermediate metabolite, S7P. The emergence of an alternative pathway that does not require any change in gene expression, but rather relies on the accumulation of an intermediate metabolite, may be a novel mechanism that mediates the robustness of metabolic networks.