CoChat

Improving the phenotype predictions of a yeast genome‐scale metabolic model by incorporating enzymatic constraints

Open access

The DOI matches its Crossref record; 68 of the 68 references it cites were found in a registry or an index; 18 of the 60 sources its claims cite were checked for support.

Abstract

Genome‐scale metabolic models (GEMs) are widely used to calculate metabolic phenotypes. They rely on defining a set of constraints, the most common of which is that the production of metabolites and/or growth are limited by the carbon source uptake rate. However, enzyme abundances and kinetics, which act as limitations on metabolic fluxes, are not taken into account. Here, we present GECKO, a method that enhances a GEM to account for enzymes as part of reactions, thereby ensuring that each metabolic flux does not exceed its maximum capacity, equal to the product of the enzyme's abundance and turnover number. We applied GECKO to a Saccharomyces cerevisiae GEM and demonstrated that the new model could correctly describe phenotypes that the previous model could not, particularly under high enzymatic pressure conditions, such as yeast growing on different carbon sources in excess, coping with stress, or overexpressing a specific pathway. GECKO also allows to directly integrate quantitative proteomics data; by doing so, we significantly reduced flux variability of the model, in over 60% of metabolic reactions. Additionally, the model gives insight into the distribution of enzyme usage between and within metabolic pathways. The developed method and model are expected to increase the use of model‐based design in metabolic engineering.

Overview

CoChat Summary

TL;DR

GECKO enhances genome-scale metabolic models by adding enzyme usage constraints based on turnover numbers and enzyme abundances, applied to *Saccharomyces cerevisiae* as ecYeast7. The enzyme-constrained model improves phenotype predictions, reduces flux variability, and enables direct integration of quantitative proteomics.

Key findings

  • ecYeast7 contains 6,741 reactions, 3,388 metabolites, 764 enzymes, and 404 pseudo-metabolites, including 226 complexes, 373 reactions with isozymes, and 315 promiscuous enzymes; a standard growth-maximization FBA solves in 135 ± 7.3 ms versus 138 ± 9.7 ms for Yeast7.
  • With a total enzyme mass constraint and 51% average enzyme saturation, ecYeast7 reproduced the Crabtree effect at a critical specific growth rate of approximately 0.3 h−1, whereas Yeast7 did not; across 12 carbon sources, ecYeast7 had an average relative error of 8% versus 100% for Yeast7.
  • Integrating absolute proteomics matched 453 enzymes accounting for 0.283 g gDW−1, or 88.7% of enzyme mass in the model; FVA showed reduced variability in 3,177 of 4,972 fluxes, with mean reduced variability of 87.7%, and only 1.5% of variable ecYeast7 reactions had complete variability versus 25.3% in Yeast7.
  • In metabolic engineering case studies, ecYeast7 captured the NDI1 knockout shift in critical specific growth rate and refined succinate and farnesene yield predictions; flux control coefficients identified HMG-CoA reductase and farnesene synthase as the enzymes with highest control on farnesene production.

Methods

  • GECKO expands the stoichiometric matrix with enzyme rows and enzyme-usage columns; kcat values convert metabolic fluxes to enzyme usage, enzyme abundances set upper bounds, and formalisms handle reversible reactions, isozymes, promiscuous enzymes, and complexes. When proteomic data are absent, a total enzyme mass constraint is used.
  • The method was applied to the Yeast7 consensus yeast GEM; kcat values were retrieved from BRENDA with flexible matching and manual curation, and simulations used FBA, FVA, random sampling, PCA, chemostat, batch, knockout, and yield analyses.
  • Proteomic integration used absolute quantitative data from *S. cerevisiae* CEN.PK113-7D grown aerobically in glucose-limited minimal media at 0.1 h−1 in triplicate; 453 enzymes were directly matched, and unmeasured enzymes were constrained by a remaining total mass term.

Limitations

  • Stated by the paper: GECKO imposes only upper-bound soft constraints, so regulation of enzyme activity and under-saturation due to substrate levels are not accounted for, meaning the true solution space is smaller than predicted. Simulation results rely heavily on kcat values, which may be incorrectly measured or annotated, and manual curation is advised for specific pathways.
  • Stated by the paper: kcat values from BRENDA are typically measured in vitro and may differ from in vivo values; average enzyme saturation was assumed due to lack of enzyme-specific saturation data; trehalose growth was the worst prediction, likely because trehalose’s hydrated volume reduces protein activity in a way not captured by the model.
  • Evident from the text: proteomic integration used a single growth condition and strain, and the model assumes no splice variants for Yeast7 genes; these factors may limit generalization across conditions or eukaryal organisms with limited kinetic data.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.15252/msb.20167411

Cites

68 of 68 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Framework for including enzymes as metabolites in a genome‐scale model GECKO uses a genome‐scale model and includes enzymes as part of reactions.Expansion of the stoichiometric matrix. M refers to metabolites, E to enzymes, v to metabolic fluxes, and e to enzyme usage. Note that 4 submatrices appear inside the new stoichiometric matrix: The upper left submatrix is equivalent to the original stoichiometric matrix, the upper right submatrix has only zeros, the lower left submatrix has the kinetic information, and the lower right submatrix is the identity matrix.
Figure 1 Framework for including enzymes as metabolites in a genome‐scale model GECKO uses a genome‐scale model and includes enzymes as part of reactions.Expansion of the stoichiometric matrix. M refers to metabolites, E to enzymes, v to metabolic fluxes, and e to enzyme usage. Note that 4 submatrices appear inside the new stoichiometric matrix: The upper left submatrix is equivalent to the original stoichiometric matrix, the upper right submatrix has only zeros, the lower left submatrix has the kinetic information, and the lower right submatrix is the identity matrix.CC BY · PubMed Central
Figure 2 Visualization of the Saccharomyces cerevisiae model expanded to account for enzymes ANetwork with metabolites and enzymes, color‐coded to show the location of enzymes (green dots) and metabolites (color‐coded by compartment).B, CCumulative distributions of (B) k cat values and (C) molecular weights for three different metabolic groups. All distributions were significantly different with P < 0.05 using a non‐parametric Wilcoxon rank‐sum test (Appendix Table S4).
Figure 2 Visualization of the Saccharomyces cerevisiae model expanded to account for enzymes ANetwork with metabolites and enzymes, color‐coded to show the location of enzymes (green dots) and metabolites (color‐coded by compartment).B, CCumulative distributions of (B) k cat values and (C) molecular weights for three different metabolic groups. All distributions were significantly different with P < 0.05 using a non‐parametric Wilcoxon rank‐sum test (Appendix Table S4).CC BY · PubMed Central
Figure 3 The limited enzyme supply defines yeast metabolic strategies at high energy demand in glucose‐limited cultures Chemostat aerobic data and model simulations at increasing dilution rate.Predicted top 6 pathways used in terms of mass at increasing dilution rate in aerobic conditions.Reduced flux variability compared with purely metabolic GEM simulations at increasing dilution rate in aerobic conditions.Chemostat anaerobic data and model simulations at increasing dilution rate.The model chooses to ferment when energy requirements for non‐growth maintenance are high. Experimental data (mean ± SD) were taken from biological triplicates of S. cerevisiae grown at high temperature (Lahtvee et al, 2016).Data information: The light blue areas in (A–D) denote the dual‐limitation (Janusian) region.
Figure 3 The limited enzyme supply defines yeast metabolic strategies at high energy demand in glucose‐limited cultures Chemostat aerobic data and model simulations at increasing dilution rate.Predicted top 6 pathways used in terms of mass at increasing dilution rate in aerobic conditions.Reduced flux variability compared with purely metabolic GEM simulations at increasing dilution rate in aerobic conditions.Chemostat anaerobic data and model simulations at increasing dilution rate.The model chooses to ferment when energy requirements for non‐growth maintenance are high. Experimental data (mean ± SD) were taken from biological triplicates of S. cerevisiae grown at high temperature (Lahtvee et al, 2016).Data information: The light blue areas in (A–D) denote the dual‐limitation (Janusian) region.CC BY · PubMed Central
Figure 4 The limited enzyme supply defines yeast metabolic strategies in batch growth under different carbon sources A, BModel simulations of maximum specific growth rate under different aerobic media conditions. (A) Yeast 7 simulations, with an average relative error of 100%. (B) ecYeast7 simulations, with an average relative error of 8%.C, DFirst two components of a PCA for (C) Yeast7 and (D) ecYeast7 flux simulations.
Figure 4 The limited enzyme supply defines yeast metabolic strategies in batch growth under different carbon sources A, BModel simulations of maximum specific growth rate under different aerobic media conditions. (A) Yeast 7 simulations, with an average relative error of 100%. (B) ecYeast7 simulations, with an average relative error of 8%.C, DFirst two components of a PCA for (C) Yeast7 and (D) ecYeast7 flux simulations.CC BY · PubMed Central
Figure 5 Integration of proteomic data into the model PCA of fluxes from 10,000 random simulations of Yeast7 and 10,000 random simulations of ecYeast7.Flux variability for all non‐zero variable fluxes from Yeast7 (3,286 reactions, 66.1% of all reactions) and ecYeast7 (3,822 reactions, 56.7% of all reactions).Distribution of different pathways in metabolism based on the total usage predicted by ecYeast7 (mg gDW−1) and their average flux variability reduction (mmol gDWh−1). The discontinuous lines indicate averages of the log values for different metabolic groups. The light blue filled region highlights the pathways with the highest average flux variability reduction shown in (D).Breakdown of the pathways with the highest average flux variability reduction (mmol gDWh−1). Colors correspond to the metabolic groups indicated in (C).
Figure 5 Integration of proteomic data into the model PCA of fluxes from 10,000 random simulations of Yeast7 and 10,000 random simulations of ecYeast7.Flux variability for all non‐zero variable fluxes from Yeast7 (3,286 reactions, 66.1% of all reactions) and ecYeast7 (3,822 reactions, 56.7% of all reactions).Distribution of different pathways in metabolism based on the total usage predicted by ecYeast7 (mg gDW−1) and their average flux variability reduction (mmol gDWh−1). The discontinuous lines indicate averages of the log values for different metabolic groups. The light blue filled region highlights the pathways with the highest average flux variability reduction shown in (D).Breakdown of the pathways with the highest average flux variability reduction (mmol gDWh−1). Colors correspond to the metabolic groups indicated in (C).CC BY · PubMed Central
Figure 6 Using the enzyme‐constrained model for metabolic engineering applications The physiological response of knocking out NDI1 can be reproduced using ecYeast7. Wild‐type and knockout experimental data are shown, together with model simulations (continuous and segmented lines, respectively).Succinate yield (grams of succinate per grams of substrate) versus biomass yield (grams of dry weight biomass per grams of substrate) simulations under batch conditions, compared with experimental data. Both glucose and ethanol phases are shown.Farnesene yield (grams of farnesene per grams of glucose) versus biomass yield (grams of dry weight biomass per grams of glucose) simulations under fed‐batch conditions, compared with experimental data.
Figure 6 Using the enzyme‐constrained model for metabolic engineering applications The physiological response of knocking out NDI1 can be reproduced using ecYeast7. Wild‐type and knockout experimental data are shown, together with model simulations (continuous and segmented lines, respectively).Succinate yield (grams of succinate per grams of substrate) versus biomass yield (grams of dry weight biomass per grams of substrate) simulations under batch conditions, compared with experimental data. Both glucose and ethanol phases are shown.Farnesene yield (grams of farnesene per grams of glucose) versus biomass yield (grams of dry weight biomass per grams of glucose) simulations under fed‐batch conditions, compared with experimental data.CC BY · PubMed Central

Verification

Claims the paper makes

11 Valid · 1 Not valid · 32 Not checked · 102 Own results and interpretations

42 more claims

Create a free account to see all 146 claims and to verify the 42 references still open.

See all claims — free account
  • Probably not a substantial claim. 42
  • It may state two separate claims as one. 67

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Cited by

See all 478

Metadata

Authors
Sánchez, B. J., Zhang, C., Nilsson, A., Lahtvee, P., Kerkhoven, E. J., & Nielsen, J.
Affiliations
  • Chalmers University of Technology (SE)
  • East China University of Science and Technology (CN)
  • Novo Nordisk Foundation (DK)
Corresponding author
Nielsen, Jens
Venue
Molecular Systems Biology
Published
2017-08-01
Volume
13
Issue
8
Pages
935-935
ISSN
1744-4292 (ISSN-L)
OpenAlex
W2743063996
Cited by
466
References
68
Citations by year
2026: 50 · 2025: 55 · 2024: 55 · 2023: 64 · 2022: 66 · 2021: 69 · 2020: 44 · 2019: 32 · 2018: 26 · 2017: 4
FWCI
11.89
Type
journal article
Language
en
Open access
Licence
Locations
  • Molecular Systems Biology (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • KTH Publication Database DiVA (KTH Royal Institute of Technology) (submittedVersion, other-oa)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • Research Portal (King's College London) (submittedVersion, cc-by)
  • OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) (submittedVersion)
  • Chalmers Publication Library (Chalmers University of Technology) (submittedVersion)
  • Technical University of Denmark, DTU Orbit (Technical University of Denmark, DTU) (submittedVersion, cc-by)
  • Chalmers Research (Chalmers University of Technology) (submittedVersion)
  • Zenodo (CERN European Organization for Nuclear Research) (submittedVersion, cc-by)
Funders
  • European Union's Horizon 2020 research and innovation program (686070, 720824)
  • Novo Nordisk Foundation
  • Knut och Alice Wallenbergs Stiftelse
  • DOE | SC | Biological and Environmental Research (DE‐SC0008744)
  • MINEDUC | Comisión Nacional de Investigación Científica y Tecnológica (6222/2014)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Microbial Metabolic Engineering and Bioproduction
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene Regulatory Network Analysis
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Bioinformatics and Genomic Networks
Keywords
genome-scale metabolic models, enzyme abundance, enzyme turnover number, geckos, Saccharomyces cerevisiae, metabolic flux prediction, flux variability, metabolic engineering, stress response
MeSH terms
Kinetics, Models, Biological, Phenotype, Saccharomyces cerevisiae, Genome, Fungal, Systems Biology, Metabolic Networks and Pathways, Metabolic Engineering

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.