Analysis of a logical regulatory network reveals how Fe-S cluster biogenesis is controlled in the face of stress

Abstract

Iron-sulfur (Fe-S) clusters are important cofactors conserved in all domains of life, yet their synthesis and stability are compromised in stressful conditions such as iron deprivation or oxidative stress. Two conserved machineries, Isc and Suf, assemble and transfer Fe-S clusters to client proteins. The model bacterium Escherichia coli possesses both Isc and Suf, and in this bacterium utilization of these machineries is under the control of a complex regulatory network. To better understand the dynamics behind Fe-S cluster biogenesis in E. coli, we here built a logical model describing its regulatory network. This model comprises three biological processes: 1) Fe-S cluster biogenesis, containing Isc and Suf, the carriers NfuA and ErpA, and the transcription factor IscR, the main regulator of Fe-S clusters homeostasis; 2) iron homeostasis, containing the free intracellular iron regulated by the iron sensing regulator Fur and the non-coding regulatory RNA RyhB involved in iron sparing; 3) oxidative stress, representing intracellular H₂O₂ accumulation, which activates OxyR, the regulator of catalases and peroxidases that decompose H₂O₂ and limit the rate of the Fenton reaction. Analysis of this comprehensive model reveals a modular structure that displays five different types of system behaviors depending on environmental conditions, and provides a better understanding on how oxidative stress and iron homeostasis combine and control Fe-S cluster biogenesis. Using the model, we were able to predict that an iscR mutant would present growth defects in iron starvation due to partial inability to build Fe-S clusters, and we validated this prediction experimentally.

Keywords: Fe-S cluster biogenesis; IscR; bacterial regulation; logical modeling; mathematical modeling; regulatory networks.

Figure 1.

Regulatory Graph centered on the Fe-S cluster biogenesis in E. coli. Nodes represent biological components, and edges regulations: activations (green, normal arrow) or inhibitions (red, T-shaped arrows). Ellipses stand for Boolean nodes (i.e. value of the node can only be 0 or 1), rectangles for multivalued nodes. Labels on edges specify the regulatory thresholds (equal to 1 if not indicated); two labels on the same edge means the regulation happens at both thresholds. Each module is represented by a color: blue nodes to the oxidative stress module; green nodes to the iron homeostasis module; the red nodes represent the IscRSUA module. The three orange nodes are the outputs of the model.

Figure 2.

Modular description of the five asymptotic classes of the model. The two axes of the square represent the environmental conditions, set by the input nodes O₂ and Fe_ext; the triangles are used to depict values of the inputs, from low to high (i.e. 0 to 2). The grid separates the five different classes of asymptotical behaviors of the system. (A): Each area contains four circles representing the behavior of the four modules of the RG identified by a color: oxidative stress response (blue), iron homeostasis module (green), IscRSUA module (red), and the Suf node (yellow). The symbol ∼ denotes an oscillatory asymptotical behavior of the module, otherwise it is stable. (B): The number indicates the size of the cyclical attractor (number of states) for each different class of behavior represented in A.

Figure 3.

WT simulations of the model. Results of simulations for the six representative nodes in each of the 9 input combinations are shown in the form of a heatmap. The two axes of the squares represent the environmental conditions, set by the input nodes O₂ and Fe_ext, the triangles are used to depict values of the inputs, from low to high (i.e. 0 to 2). Each cell of the grid stands for one of the nine input conditions (depending on the values of O₂and Fe_ext). The qualitative asymptotical behavior of the node is indicated: oscillation (∼) or stable (no indication). Its expression level is reflected through the color graduation given by the color scale on the right of each square. In cases of oscillations, the color represents the mean of all the states of the cyclic attractor.

Figure 4.

Simulations of simple and double perturbations of the iron and oxidative stress modules on nodes H₂O₂ and Fe_free. Heatmap representation of the asymptotical behavior of nodes H₂O₂(top line) and Fe_free (bottom line) in the nine input conditions (see caption of Fig 3) for, from left to right, WT, OxyR KO mutant, Fur KO mutant and Fur KO-OxyR KO double mutant. The two axes of the squares represent the environmental conditions, set by the input nodes O₂ and Fe_ext; the triangles are used to depict values of the inputs, from low to high (i.e. 0 to 2). The qualitative asymptotical behavior of the node is indicated: oscillation (∼) or stable (no indication). Node level is reflected through the color graduation given by the color scale on the right of each square. In cases of oscillations, the color represents the mean of all the states of the cyclic attractor.

Figure 5.

Modules perturbations effects on the Fe-S cluster biogenesis machineries. Heatmap representation of the asymptotical behaviour of nodes Isc (top line) and Suf (bottom line) in the nine input conditions (see caption of Fig 3) for, from left to right, WT, OxyR KO mutant, Fur KO mutant and Fur KO-OxyR KO double mutant. The two axes of the squares represent the environmental conditions, set by the input nodes O₂ and Fe_ext; the triangles are used to depict values of the inputs, from low to high (i.e. 0 to 2). The qualitative asymptotical behavior of the node is indicated: oscillation (∼) or stable (no indication). Node level is reflected through the color graduation given by the color scale on the right of each square. In cases of oscillations, the color represents the mean of all the states of the cyclic attractor.

Figure 6.

Modules perturbations effects on the ATCs. Heatmap representation of the main nodes depending of Fe_extand O₂. The two axes of the squares represent the environmental conditions, set by the input nodes O₂ and Fe_ext; the triangles are used to depict values of the inputs, from low to high (i.e. 0 to 2). The qualitative asymptotical behavior of the node is indicated: oscillation (∼) or stable (no indication). Node level is reflected through the color graduation given by the color scale on the right of each square. In cases of oscillations, the color represents the mean of all the states of the cyclic attractor.

Figure 7.

Growth defects predictions and experimental verification. (A) Heatmap representation of situations where neither the Isc and Suf machineries are active in the model. The two axes of the squares represent the environmental conditions, set by the input nodes O₂ and Fe_ext; the triangles are used to depict values of the inputs, from low to high (i.e. 0 to 2). The qualitative asymptotical behavior of the node is indicated: oscillation (∼) or stable (no indication). Node level is reflected through the color graduation given by the color scale on the right of each square. In cases of oscillations, the color represents the mean over all the states of the cyclic attractor. Red and blue circles correspond to the experimentally tested conditions (iron starvation and iron replete conditions respectively). (B) Experimental verification of the predictions. The dashed and filled lines correspond respectively to iron replete and iron deficient conditions, respectively. Iron starvation was induced using 300 µM of dipyridyl chelator. Growth was measured at OD₆₀₀ for 14 h at 37°C using a TECAN microplate reader.