Fig. 4.

Model predictions and experimental data. a Single-cell measurements reveal that metabolic enzymes are as variable as other members of the proteome¹⁷. Data correspond to ~80% of the E. coli proteome, including 268 enzymes involved in various metabolic functions⁵⁵ (coloured circles). The coefficient of variation (CV) is defined as the standard deviation over the mean of measured distributions. Distributions with CV > 1 (dashed line) are long-tailed and peak at zero, which resemble the distributions required for catalytically-induced bimodality and multimodality (Figs. 2b and 3); shown are the distributions of three representative enzymes computed from fitted Gamma distributions¹⁷. b Predictions of the Poisson Mixture Model for combinations of parameters λ_∞ and K in Eq. (5). For a lowly abundant enzyme (distribution in inset), the model predicts unimodality, catalytically-induced bimodality and multimodality in large regions of the (λ_∞, K) space. The red line represents a constant ratio λ_∞/K for the median k_cat ≈ 16.5 s⁻¹ across 752 enzymes in E. coli, see k_cat distribution displayed in the inset⁴⁹. Shaded red corresponds to the lines obtained for k_cat values within a range 0.5- and 2-fold of the median (highlighted in green in the inset). The grey lines and shaded areas correspond to perturbations to the consumption rate constant (k_c) and enzyme saturation $\left(ϵ\right)$; more details in Methods