Constraint-Based Analysis of Genetic Variation Effects in Escherichia coli after Adaptive Laboratory Evolution

Genetic variation occurs naturally in microorganisms both in the environment and laboratory. The effects of genetic variation range from silent mutations with no measurable phenotypic effects to drastic changes with emergence of new cellular functions or disruption of existing ones. With next-generation sequencing technology, screening of these genetic variants can be accomplished rapidly. On the computational side, genome-scale metabolic models can be combined with constraint-based simulations to analyze the effects of phenomena observed in high-throughput data.

In this work we have combined bioinformatic tools for predicting the effects of genetic variation on protein function (PROVEAN, SIFT, I-Mutant and FoldX) with constraint-based simulations in order to understand how mutations affect the metabolic capabilities of cells. We analyzed genetic variants of E. coli adapted to 11 toxic compounds through adaptive laboratory evolution. In 242 re-sequenced strains, we identified 1948 mutations in metabolic genes (1748 single nucleotide variations, 122 small insertions or deletions, 85 large deletions and 13 transposon insertions). A constraint-based model, based on the iJO1366 genome-scale reconstruction of E. coli, was modified to represent each of the mutant strains, using predictions from the bioinformatic tools mentioned above. These strain-specific models were then used to predict the growth rates of the mutant strains. The predictions based on the four bioinformatic tools were compared to each other and different methods of integrating the mutations into the model were attempted. The highest prediction accuracy was achieved using SIFT, with a Pearson’s correlation coefficient of 0.63 between computationally predicted growth rates and experimentally measured data.