Systems approach to refining genome annotation

Abstract

Genome-scale models of Escherichia coli K-12 MG1655 metabolism have been able to predict growth phenotypes in most, but not all, defined growth environments. Here we introduce the use of an optimization-based algorithm that predicts the missing reactions that are required to reconcile computation and experiment when they disagree. The computer-generated hypotheses for missing reactions were verified experimentally in five cases, leading to the functional assignment of eight ORFs (yjjLMN, yeaTU, dctA, idnT, and putP) with two new enzymatic activities and four transport functions. This study thus demonstrates the use of systems analysis to discover metabolic and transport functions and their genetic basis by a combination of experimental and computational approaches.

Fig. 1.

Systematic procedure for identifying missing reactions and ORF assignments. With a metabolic network reconstruction, growth predictions can be made and compared with experimental growth phenotypes (green box). For medium conditions where the microorganism grows and the model does not predict growth (red wells), reactions from a universal database can be identified by an optimization algorithm (red box) that, if added to the current reconstruction, allow for growth predictions by the model (see Materials and Methods and Supporting Text for algorithm details). Experimental testing of mutant strains can be used to identify genes responsible for the added enzymatic activities, and subsequent studies of gene-expression and biochemical characterizations can be used to further support conclusions (blue boxes). Data shown in the blue boxes is for the genes identified as being involved in l-galactonate catabolism: yjjL (putative transporter), yjjM (putative regulator), and yjjN (putative oxidoreductase). Reported gene-expression levels are relative to the internal control gene, acpP.

Fig. 2.

Utilization of d-malate as a sole carbon source. (A) The two pathways that are predicted by the algorithm and the corresponding genes whose mutants were screened for altered phenotypes (red genes indicate that corresponding mutant strains showed altered growth). (B) Growth phenotype screens for the parental strain (BW25113), three mutants with altered phenotypes (ΔyeaU, ΔdctA, and ΔyeaT), and other screened mutants for growth on d-malate. (C) Expression levels were determined for BW25113 grown on l-lactate and d-malate by using RT-PCR. The values reported are the expression levels relative to the internal control gene acpP and the fold expression changes (d-malate to l-lactate) across the two conditions. (D) Results from kinetic assays of overexpressed YeaU protein show enzymatic activity levels as a function of increasing concentrations of d-malate.