Purpose: With projects sequencing the genomes of >100,000 important foodborne pathogens currently underway, the objective of this study was to generate GEMs for numerous foodborne outbreak strains of Salmonella spp. and Listeria monocytogenes, and to compare nutrient utilization predictions of these models to in vitro results.
Methods: Genomes of six L. monocytogenes and five Salmonella spp. strains were taken from the NCBI database and uploaded to KBase, a semi-automated computational resource used to generate the GEMs. These models were then used to generate strain-specific nutrient utilization predictions for 95 sources of carbon under aerobic and anaerobic conditions using General Algebraic Modeling System (GAMS) software. In silico predictions were then compared to in vitro experiments performed using Biolog phenotypic microarray plates. Following this comparison, GEMs were manually curated to increase agreement between in silico predictions and in vitro results.
Results: Carbon source utilization agreement between in silico predictions and in vitro results was strong and significant (Pearson correlation test statistic yields p<0.001) and ranged from 80% to 90% for L. monocytogenes and from 90% to 98% for Salmonella strains. Once validated, L. monocytogenes and Salmonella GEMs were used to simulate environments of numerous food matrices and host-niches to identify differentiating metabolic pathway capabilities and 100’s of essential metabolic reactions required for growth and viability for the strains from each genus.
Significance: Since many of these pathogenic bacteria continue to emerge in foods and have large global impacts to human health and the economy, this research has demonstrated new post-genomic era approaches to identify new targets to treat human disease and make foods safer from Salmonella spp. and Listeria monocytogenes.