Abstract
BackgroundMicrobial hosts offer a number of unique advantages when used as production systems for both native and heterologous small-molecules. These advantages include high selectivity and benign environmental impact; however, a principal drawback is low yield and/or productivity, which limits economic viability. Therefore a major challenge in developing a microbial production system is to maximize formation of a specific product while sustaining cell growth. Tools to rationally reconfigure microbial metabolism for these potentially conflicting objectives remain limited. Exhaustively exploring combinations of genetic modifications is both experimentally and computationally inefficient, and can become intractable when multiple gene deletions or insertions need to be considered. Alternatively, the search for desirable gene modifications may be solved heuristically as an evolutionary optimization problem. In this study, we combine a genetic algorithm and elementary mode analysis to develop an optimization framework for evolving metabolic networks with energetically favorable pathways for production of both biomass and a compound of interest.ResultsUtilization of thermodynamically-weighted elementary modes for flux reconstruction of E. coli central metabolism revealed two clusters of EMs with respect to their ΔGp°. For proof of principle testing, the algorithm was applied to ethanol and lycopene production in E. coli. The algorithm was used to optimize product formation, biomass formation, and product and biomass formation simultaneously. Predicted knockouts often matched those that have previously been implemented experimentally for improved product formation. The performance of a multi-objective genetic algorithm showed that it is better to couple the two objectives in a single objective genetic algorithm.ConclusionA computationally tractable framework is presented for the redesign of metabolic networks for maximal product formation combining elementary mode analysis (a form of convex analysis), pathway thermodynamics, and a genetic algorithm to optimize the production of two industrially-relevant products, ethanol and lycopene, from E. coli. The designed algorithm can be applied to any small-scale model of cellular metabolism theoretically utilizing any substrate and applied towards the production of any product.
Highlights
Microbial hosts offer a number of unique advantages when used as production systems for both native and heterologous small-molecules
There exists a variety of methods for studying metabolic networks in both quantitative and qualitative manners: flux balance analysis (FBA) [29,30,31], 13C-labeling based metabolic flux analysis (13C-MFA) [32], metabolic control analysis [33], elementary mode analysis (EMA) [34], extreme pathway analysis [35], cybernetic modeling [36,37], and biochemical systems theory [3840]
A genetic algorithm (GA) is used for the optimization, by evolving a population of potential solutions towards the global optimum solution
Summary
Microbial hosts offer a number of unique advantages when used as production systems for both native and heterologous small-molecules. There exists a variety of methods for studying metabolic networks in both quantitative and qualitative manners: flux balance analysis (FBA) [29,30,31], 13C-labeling based metabolic flux analysis (13C-MFA) [32], metabolic control analysis [33], elementary mode analysis (EMA) [34], extreme pathway analysis [35], cybernetic modeling [36,37], and biochemical systems theory [3840] Many of these methods do not necessarily identify experimentally tractable metabolic engineering targets such as gene deletions. Some algorithms based on the aforementioned methods can be used to identify such targets including minimization of metabolic adjustment (MoMA) [41], regulatory on/off minimization (ROOM) [42], OptKnock [43], OptStrain [44], OptReg [45], and OptGene [46] All six of these methods require solving an optimization problem to determine flux distributions as a means of evaluating the strain's (or mutant strain's) metabolic capabilities. Exhaustively searching genomic space for knockout candidates is computationally intractable even on small-scale metabolic models (less than 100 reactions), much less on current genome-scale metabolic models (greater than 1000 reactions) due to prohibitive computation time
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
