Characterization of physiological responses to 22 gene knockouts in Escherichia coli central carbon metabolism

Abstract

Understanding the impact of gene knockouts on cellular physiology, and metabolism in particular, is centrally important to quantitative systems biology and metabolic engineering. Here, we present a comprehensive physiological characterization of wild-type Escherichia coli and 22 knockouts of enzymes in the upper part of central carbon metabolism, including the PTS system, glycolysis, pentose phosphate pathway and Entner-Doudoroff pathway. Our results reveal significant metabolic changes that are affected by specific gene knockouts. Analysis of collective trends and correlations in the data using principal component analysis (PCA) provide new, and sometimes surprising, insights into E. coli physiology. Additionally, by comparing the data-to-model predictions from constraint-based approaches such as FBA, MOMA and RELATCH we demonstrate the important role of less well-understood kinetic and regulatory effects in central carbon metabolism.

Keywords: COBRA modeling; Cell physiology; Escherichia coli; Gene knockout; Metabolism.

Figure 1

Upper central carbon metabolism, with all genes studied here shown in their metabolic contexts (fbaA and rpiA were not included in this study, shown in red, see Methods section).

Figure 2

Measured physiological parameters. Bar colors reflect pathway assignment (wild-type: gray; transporters and phosphoglucomutase: blue; oxidative pentose phosphate pathway: red; non-oxidative pentose phosphate pathway: green; Entner-Doudoroff pathway: orange; upper EMP pathway: purple). Error bars indicate standard errors of the mean for growth rate (n=3) and cell density per OD (n=2), and standard errors attributable to regression and measurement error for biomass and acetate yield.

Figure 3

Calculated uptake rates of glucose and oxygen. Bar colors reflect pathway assignment (wild-type: gray; transporters and phosphoglucomutase: blue; oxidative pentose phosphate pathway: red; non-oxidative pentose phosphate pathway: green; Entner-Doudoroff pathway: orange; upper EMP pathway: purple). Error bars reflect the propagation of measurement errors.

Figure 4

Biomass composition analysis. Bar colors reflect pathway assignment (wild-type: gray; transporters and phosphoglucomutase: blue; oxidative pentose phosphate pathway: red; non-oxidative pentose phosphate pathway: green; Entner-Doudoroff pathway: orange; upper EMP pathway: purple). Error bars represent standard errors of the mean (n=4; 2 biological replicates with 2 technical replicates each).

Figure 5

Pairwise correlation analysis of all measured data. The coefficients are given in the lower triangle, and the quality and direction of the correlation is represented visually by ellipsoids in the upper triangle (more elongated ellipsoid = higher quality correlation). The coloring is scaled to reflect value from −1.0 (red) to 0 (white) to 1.0 (blue). The included data sets (left to right) are: growth rate (h⁻¹), dry weight per OD (g/L/OD₆₀₀), biomass yield (g/g), acetate yield (mol/mol), percentages of the four major biomass components, and the relative fatty acid contents (mmol/g(lipid)). All coefficients greater than 0.4 indicate a significant nonzero correlation at 95% confidence.

Figure 6

Scatter plots of correlated data. Marker colors reflect pathway assignment (wild-type: gray; transporters and phosphoglucomutase: blue; oxidative pentose phosphate pathway: red; non-oxidative pentose phosphate pathway: green; Entner-Doudoroff pathway: orange; upper EMP pathway: purple).

Figure 7

Principal component analysis (PCA) plot showing the first two components, which together account for more than half of the total variation in the data. The coefficients mapping these components to the original (normalized and standardized) data are shown in the table to the right.

Figure 8

Comparison of experimental growth rates, biomass yields, and acetate yields to those predicted by three constraint-based modeling approaches: FBA (flux balance analysis), MOMA (minimization of metabolic adjustment), and RELATCH (relative optimality in metabolic networks). Marker colors reflect pathway assignment (wild-type: gray; transporters and phosphoglucomutase: blue; oxidative pentose phosphate pathway: red; non-oxidative pentose phosphate pathway: green; Entner-Doudoroff pathway: orange; upper EMP pathway: purple). Pearson correlation coefficients (ρ) describe the agreement between prediction and measurement. Wild-type data were excluded from this correlation.