Metabolic modeling and response surface analysis of an Escherichia coli strain engineered for shikimic acid production

Juan A Martínez,Alberto Rodriguez,Francisco Bolivar,Octavio T Ramírez,Alvaro R Lara,Noemí Flores,Fabian Moreno,Guillermo Gosset

doi:10.1186/s12918-018-0632-4

Juan A Martínez, Alberto Rodriguez + Show 6 more

Open Access

https://doi.org/10.1186/s12918-018-0632-4

Copy DOI

Abstract

BackgroundClassic metabolic engineering strategies often induce significant flux imbalances to microbial metabolism, causing undesirable outcomes such as suboptimal conversion of substrates to products. Several mathematical frameworks have been developed to understand the physiological and metabolic state of production strains and to identify genetic modification targets for improved bioproduct formation. In this work, a modeling approach was applied to describe the physiological behavior and the metabolic fluxes of a shikimic acid overproducing Escherichia coli strain lacking the major glucose transport system, grown on complex media.ResultsThe obtained flux distributions indicate the presence of high fluxes through the pentose phosphate and Entner-Doudoroff pathways, which could limit the availability of erythrose-4-phosphate for shikimic acid production even with high flux redirection through the pentose phosphate pathway. In addition, highly active glyoxylate shunt fluxes and a pyruvate/acetate cycle are indicators of overflow glycolytic metabolism in the tested conditions. The analysis of the combined physiological and flux response surfaces, enabled zone allocation for different physiological outputs within variant substrate conditions. This information was then used for an improved fed-batch process designed to preserve the metabolic conditions that were found to enhance shikimic acid productivity. This resulted in a 40% increase in the shikimic acid titer (60 g/L) and 70% increase in volumetric productivity (2.45 gSA/L*h), while preserving yields, compared to the batch process.ConclusionsThe combination of dynamic metabolic modeling and experimental parameter response surfaces was a successful approach to understand and predict the behavior of a shikimic acid producing strain under variable substrate concentrations. Response surfaces were useful for allocating different physiological behavior zones with different preferential product outcomes. Both model sets provided information that could be applied to enhance shikimic acid production on an engineered shikimic acid overproducing Escherichia coli strain.

Highlights

Classic metabolic engineering strategies often induce significant flux imbalances to microbial metabolism, causing undesirable outcomes such as suboptimal conversion of substrates to products
Physiological characterization, parametrization and modeling of strain AR36 on variant substrate conditions Figure 1 depicts the results from the 9 experimental design fermentations with the central point done by triplicate along with constructed physiological models
The largest standard deviations corresponded to final shikimic acid (SA) produced ([SA]f ), final consumed GLC ( [GLC]) and the exponential consumption rate qsexp

Summary

Introduction

Classic metabolic engineering strategies often induce significant flux imbalances to microbial metabolism, causing undesirable outcomes such as suboptimal conversion of substrates to products. SA was at first produced from the seed of the Chinese star anise plant Illicium verum, employing classic extraction processes with yields of only 30 mg/Kg approximately [10,11,12] For this reason, over the past years, many studies concerning SA production have focused on recovery technologies, chemical synthesis methods and biotechnological production using different microorganisms [9, 13, 14]. Classic metabolic engineering (ME) allows flux redirection in a biochemical network into valuable compounds by genetic manipulation, it often induces significant flux imbalances to the CCM that may cause undesirable outcomes These imbalances can disrupt precursor availability and energy balances, causing the accumulation of pathway intermediates and unwanted byproducts, reducing strain fitness and product yields [16]. Mathematical models, advances on informatics and the availability of big and more precise omics data sets have proved useful to resolve and clarify the complex network interactions and system characteristics [19,20,21,22]

Methods

Results

Discussion

Conclusion