Steady state analysis of the genetic regulatory network incorporating underlying molecular mechanisms for anaerobic metabolism in Escherichia coli.

A Gene Regulatory Network (GRN) represents complex connections between genes in a cell which interact with each other through their RNA and protein expression products, thereby determining the expression levels of mRNA and proteins required for functioning of the cell. Microarray experiments yield the log fold change in mRNA abundance and quantify the expression levels for a GRN at the genome level. While Boolean or Bayesian modeling along with expression and location data are useful in analyzing microarray data, they lack underlying mechanistic details present in GRNs. Our objective is to understand the role of molecular mechanisms in quantifying a GRN. To that effect, we analyze under steady state, the complete GRN for the central metabolic pathway during anaerobiosis in Escherichia coli. We simulate the microarray experiments using a steady state gene expression simulator (SSGES) that models molecular mechanistic details such as dimerization, multiple-site binding, auto-regulation and feedback. Given a GRN, the SSGES provided the log fold change in mRNA expression values as the output, which can be compared to data from microarray experiments. We predict the log fold changes for mutants obtained by knocking out crucial transcriptional regulators such as FNR (F), ArcA (A), IHFA-B (I) and DpiA (D) and observe a high degree of correlation with previously reported experimental data. We also predict the microarray expression values for hitherto unknown combinations of deletion mutants. We hierarchically cluster the predicted log fold change values for these mutants and postulate that E. coli has evolved from a predominantly lactate secreting (FAID mutant) into a mixed acid secreting phenotype as seen in the wild type (WT) during anaerobiosis. Upon simulating a model without incorporating the mechanistic details, not only the correlation with the experimental data reduced considerably, but also the clustering of expression data indicated WT to be closer to the quadruple mutant FAID. This clearly demonstrates the significance of incorporating mechanistic data while quantifying the expression profile of a GRN which can help in predicting the effect of a gene mutant and understanding the evolution of transcriptional control.