Whole-cell modeling in yeast predicts compartment-specific proteome constraints that drive metabolic strategies

Abstract

When conditions change, unicellular organisms rewire their metabolism to sustain cell maintenance and cellular growth. Such rewiring may be understood as resource re-allocation under cellular constraints. Eukaryal cells contain metabolically active organelles such as mitochondria, competing for cytosolic space and resources, and the nature of the relevant cellular constraints remain to be determined for such cells. Here, we present a comprehensive metabolic model of the yeast cell, based on its full metabolic reaction network extended with protein synthesis and degradation reactions. The model predicts metabolic fluxes and corresponding protein expression by constraining compartment-specific protein pools and maximising growth rate. Comparing model predictions with quantitative experimental data suggests that under glucose limitation, a mitochondrial constraint limits growth at the onset of ethanol formation-known as the Crabtree effect. Under sugar excess, however, a constraint on total cytosolic volume dictates overflow metabolism. Our comprehensive model thus identifies condition-dependent and compartment-specific constraints that can explain metabolic strategies and protein expression profiles from growth rate optimisation, providing a framework to understand metabolic adaptation in eukaryal cells.

Fig. 1. pcYeast model formulation and calibration of protein synthesis parameters.

a A schematic overview of reactions in the model, their interdependence and constraints. Metabolic reactions $v_i$ are proportional to enzyme concentrations $e_i$ that are synthesised at rate $v_{syn,i}$ by the ribosomes $R$. Each protein can be degraded with rate $v_{deg,i}=k_{deg}\cdot e_i$ or diluted by growth rate $v_{dil,i}=\mu \cdot e_i$. Compartment-specific constraints are indicated in the light-blue boxes. b Optimisation problem with the key constraints, including (1) steady-state mass balances; (2) production of biomass components such as DNA, lipids, cell wall and polysaccharides. Proteins are excluded as their synthesis rates are optimisation variables (3) enzyme capacity constraints that couple metabolic flux to catalytic rate $k_{cat,i}$ and the enzyme level, whose value at steady state is determined by its synthesis rate. Note we use equalities and hence enzymes work at their maximal rate and minimal required protein levels are computed; (4) ribosome capacity that defines an upper bound for protein synthesis rate; (5) compartment-specific proteome constraints that define the maximal concentration of proteins that can be contained in that compartment, with $w_i$ the specific volume or area of protein $i$; (6) a cytosolic protein density constraint that has the same function as that of proteome constraints, but whose equality forces the cell to fill up any vacant proteome space with unspecified protein UP. c Growth rate was varied through sugar type (trehalose, galactose, maltose, glucose) or glucose concentration, and ribosomal protein fraction was determined by proteomics. The translation rate was calibrated on the literature data (Supplementary Notes 6). d Impact of mCherry protein overexpression on growth rate. Symbols show experimental data, solid lines show model predictions based on glucose minimal (SD) medium or rich SC/YPD media. Model predictions were obtained by varying the proteome mass fraction, occupied by mCherry, and determining the maximal predicted growth rate at each value of the mass fraction. The relative growth fitness represents the ratio between the growth rate at certain mCherry expression level vs. the unperturbed state (no mCherry expression). Source data for panels c and d are provided as a Source Data file.

Fig. 2. Predicted and measured physiological response of S. cerevisiae CEN.PK as a function of glucose availability.

Predicted (lines) and measured (symbols) physiological parameters and fluxes of S. cerevisiae CEN.PK strain a Measured (symbols) and predicted (line) residual glucose concentrations as a function of growth rate. The latter was calculated based on published affinity for glucose and assuming negligible intracellular glucose under these conditions. Note that this resembles a Monod growth curve but with the dependent and independent axis swapped, as we control growth rate in a chemostat. b Maximal feasible growth rates of the model as a function of the glucose transporter saturation. c Measured (symbols) and predicted biomass yield on glucose. d Experimental fluxes from glucose-limited chemostats at different dilution rates and from two batch experiments: excess trehalose (which mimicks glucose limitation at low dilution rate) and excess glucose at the highest growth rate. The lines are model predictions; e Computed proteome occupancy of different constrained protein pools. A fraction of 1 means that the compartment is full with metabolically actively proteins and constrains the growth rate at that condition. The shading of the different growth regimes is based on the (latest) constraint, actively limiting growth, referring to Panel (e). Source data are provided as a Source Data file.

Fig. 3. Proteomics data of selected pathways as a function of glucose availability.

Blue symbols are glucose-limited chemostat data; orange symbols are controlled batch experiments with excess trehalose (lowest growth rate) or glucose (highest growth rate) a Comparison of predicted minimal proteome fractions to sustain growth with the experimentally determined proteome fraction for four pathways. The ratio between the two represents an estimate of the saturation level of the constituent enzymes. Lines represent the model; experimental data are symbols. b Decay of steady-state mitochondrial protein fraction with growth rate at onset of ethanol formation suggests a maximal rate of mitochondrial biosynthesis $v_{syn,max}$. The shading of the different growth regimes is based on the (latest) constraint, actively limiting growth, referring to Fig. 2e. Individual proteins in panel a were mapped to metabolic pathways using a manually-curated pathway annotation file (Supplementary Data 5). Source data are provided as a Source Data file.

Fig. 4. Predicted and measured physiology of S. cerevisiae CEN.PK strain in sugar batch cultivations.

Model predictions, fluxes and protein levels plotted as a function of growth rate during hexose sugar excess conditions (in the order: trehalose, galactose, maltose, glucose) a Fluxes of sugar consumption, oxygen consumption and ethanol production. Circles are experimental data, bar plots indicate model predictions (of both the growth rate and fluxes); b Predicted active constraints under the different sugar excess conditions as predicted by the mode (see legend of Fig. 2 for details). c Comparison of predicted minimally needed proteome fractions with experimentally determined ones suggests differences in saturation level between pathways. Lines represent the model, experimental data are circles; d Linearity of the expression of individual enzymes in glycolysis (right) and respiration (left) with growth rate suggests trading in of respiratory protein for fermentative protein. Asterixes indicate aggregated proteome fractions instead of fractions of individual proteins. The respiratory proteins converge at 0.474 ± 0.0002 h⁻¹. Shading in top plots of panel (d) highlights the common trend of individual protein abundance, corresponding to the end-products in the scheme on the bottom. Individual proteins in panels c and d were mapped to metabolic pathways using a manually-curated pathway annotation file (Supplementary Data 5). Source data are provided as a Source Data file.

Fig. 5. The effect of translation inhibition by cyclohexamide on growth rate, fluxes and proteome fractions in controlled aerobic batch fermentations on glucose.

a Dependency of culture optical density (OD) on the time post-inoculation to the medium supplemented with cycloheximide. Lines are values of consecutive OD measurements, points represent the times when cultures were sampled. b–d Comparison of pcYeast predictions and experimental data: lines are model predictions; symbols are experimental data points. b Main catabolic fluxes as a function of the growth rate. c Ribosomal proteome fractions. Data from Fig. 1c are included for comparison. d Proteome fractions measured for key metabolic pathways, and the minimal proteome fractions predicted by pcYeast. Individual proteins in panel d were mapped to metabolic pathways using a manually-curated pathway annotation file (Supplementary Data 5). Source data are provided as a Source Data file.