| We have used a systems biology approach to investigate regulatory networks within Escherichia coli, with a focus on relating network structure and network behavior. The global response networks governing the response to the reactive nitrogen species (RNOS) nitric oxide (NO) and S-nitrosoglutathione (GSNO) were elucidated based on measured network behavior. The analysis was multi-pronged, involving transcriptome, phenotype, chemoinformatic and biochemical analysis. The identified NO response network consists of multiple levels of interactions, with the primary level dominated by the interaction of NO with metal centers. Specifically, we determined that NO inactivates the Fe-S clusters of dihydroxy-acid dehydratase and isopropylmalate isomerase, enzymes essential for the synthesis of branched chain amino acids (BCAA). This targeting of BCAA biosynthesis is the mechanism underlying NO-mediated bacteriostasis and has not been previously recognized. The primary NO sensors are the response regulators NsrR, NorR, IscR and the proteins IlvD, LeuCD and cytochrome bo. BCAA depletion perturbs biosynthesis regulators IlvY, MetJ and the leucine transcription attenuator as well as the virtual "stringent factor". ArgR, PurR and GatR are perturbed as part of the stringent response. Decreased respiratory activity activates ArcA. In contrast to the NO-metal interactions, the major mode of GSNO attack is metabolite depletion by transnitrosation. Transnitrosation of cysteine and homocysteine results in perturbation of the MetJ and CysB regulators. Methionine starvation activates the stringent response, perturbing PurR and FlhDC. ArcA is repressed directly by GSNO. Thus, in addition to determining the structure of the response networks and novel RNOS targets, we have identified two regulators, MetJ and ArcA, which respond differentially to different types of RNOS. We have also studied prediction of network behavior from known network structure with the pap regulatory network. We used a stochastic model to integrate known biochemical data and predict the network behavior. This model reproduces existing experimental data and has been used to investigate the role of the individual network components and the expression data of virtual mutants. Specifically, the model predicts that pap expression varies according to growth rate and that PapI feedback fine tunes the growth rate sensitivity. |
