What this research found
Flux balance analysis says an engineered E. coli strain could make lycopene at 1.056568 mmol/gDW/h, but enzyme kinetics put the unmodified pathway's real capacity at 0.163 µM/s — a gap of roughly 6,000-fold. Combining the two model types located the enzymes that actually limit the pathway, then searched for the expression ratio that relieves them, landing on a CrtE : CrtB : CrtI ratio of 1.0 : 6.8 : 7.0 and a predicted 70.7-fold flux gain.
- The stoichiometric and kinetic pictures disagree by about 6,000-fold. Flux balance analysis puts the theoretical ceiling at 1.056568 mmol/gDW/h, while the kinetic model gives the baseline pathway a capacity of only 0.163 µM/s.
- All three bottlenecks sit in the lycopene branch rather than the upstream methylerythritol phosphate pathway. Phytoene desaturase (CrtI) scores 6480, phytoene synthase (CrtB) 6339, and geranylgeranyl diphosphate synthase (CrtE) 1849, on a scale where any score above 1 means metabolic demand exceeds enzyme capacity.
- Slow turnover is the underlying cause. CrtB has a kcat of 0.05 per second and CrtI 0.25 per second, against a median of 1.80 per second across the 11 enzymes parameterised and 25.00 for the fastest of them.
- The best resource split barely moves as the expression budget grows: about 6.7 to 6.8 percent to CrtE, 46.1 percent to CrtB, and 47.1 to 47.2 percent to CrtI at every budget tested.
- Predicted flux rises close to linearly with total expression — 9.4-fold at a 20x budget, 23.6-fold at 50x, 47.1-fold at 100x, and 70.7-fold at 150x, reaching 11.525 µM/s.
How it was done
Flux balance analysis on iJO1366, the genome-scale model of E. coli K-12 MG1655 with 2,587 reactions and 1,808 metabolites, fixed the theoretical flux ceiling and the per-reaction demand. Turnover numbers and Michaelis constants for 11 methylerythritol phosphate and lycopene pathway enzymes were then assembled from BRENDA and the published literature, together with estimated intracellular concentrations for 15 metabolites, giving each enzyme a capacity of kcat times enzyme concentration times substrate saturation at 5 µM enzyme. Dividing stoichiometric demand by kinetic capacity ranked the bottlenecks. A genetic algorithm in pymoo, run with a population of 100 over 200 generations, then searched expression fold-changes from 1x to 100x for the allocation that maximises the pathway's slowest step under four total-budget caps, and the results were written up as a 17-page manuscript with a proposed promoter-and-plasmid design for experimental validation.
Data sources
- iJO1366 genome-scale metabolic model of E. coli K-12 MG1655 — 2,587 reactions, 1,808 metabolites, 1,367 genes
- BRENDA enzyme database and published literature — kcat and Km for 11 enzymes
- Estimated intracellular concentrations for 15 metabolites spanning central metabolism, the methylerythritol phosphate pathway, and the lycopene branch
Limitations
The predictions are entirely computational, and kinetic constants measured in vitro may not hold inside a living cell. The model also treats enzyme activities as independent and ignores metabolic burden from overexpression, protein solubility limits, competition for ribosomes and ATP, and feedback inhibition.
Figures from this analysis
How this research was produced
K-Dense Web planned and ran this synthetic biology investigation end to end — gathering the sources, carrying out the analysis, producing the figures, and drafting the report. The full session transcript, including every intermediate step, is available to view.
