Designing Phenotypes
In silico design of over producing bacterial strains.
 

In Silico Discovery of Gene Targets through Natural Fruition

Zachary L. Fowler

Genetic engineering has enabled the purposeful direction of genetic changes to produce microbial strains capable of high produce yields for a fast array of metabolic products, both natural and engineered. Developing phenotypes for enhanced production traditionally is done through intuitive engineering from know metabolic functions and/or networks, or random mutagenesis. More recent efforts have focus on system’s biology approaches to retrofit microbial metabolism to a desired end. However, in many of these approaches the complexity of the metabolic networks, specifically the governing mass and energy balances along with their kinetics and regulation, as well as the lack of relevant information on all metabolic channels has proven to be challenging barriers to overcome.

Yet in the absence of detailed information on regulation and kinetics it has been shown feasible to predict, at least in part, the behavior of global cellular metabolism through steady state analysis using genome-scale stoichiometric models. These constraint-based models provide a means for assessing the integrated metabolic potential of all cellular phenotypes, represented by flux-balance constraints inferred from the mass balances of metabolites across the whole known metabolic network. Using the principle of Darwinian evolution, our approach evolves strains in silico though mutation and crossover for the purposes of finding a global optimum, the most fit phenotype, in the space of all possible phenotypes. Once identified, in silico gene deletions can be reproduced in the laboratory for model comparison as well as the generation of a specifically tuned metabolic phenotype for flavonoid production.

A       B
 

Figure: (A) Adapted from Dansenko and Warner, 2000. Deletion strategy to be used. (B) Productivity-Viability
solution space for primary gene deletion using an expanded genome scale model of E. coli metabolism.

Optimization of Anthocyanin Biosynthesis by Application of Phenotype Design

Yajun Yan and Zachary L. Fowler

Optimization strategies for developing recombinant strains generally rely on trial-and-error methods that require numerous rounds of isolation, mutation, and phenotype assessment, often times not improving metabolic yields greater than 5% for each round of mutation. As such, the prior outline systematic designing of phenotypes using linear programming allows for the identification of excess metabolic channels with the greatest impact on flavonoid yields. We have isolated a plethora of gene deletion targets to investigate that along with selected over expression that have predicted fold increase of at least 100%. While such large improvements are not expected in vivo due to the lack of kinetics in the model, increase of 20% and greater are more than likely due to simply increase carbon flow in the direction of the required metabolites UDP-glucose and UTP.