The role of coupled positive feedback in the expression of the SPI1 type three secretion system in Salmonella

Abstract

Salmonella enterica serovar Typhimurium is a common food-borne pathogen that induces inflammatory diarrhea and invades intestinal epithelial cells using a type three secretion system (T3SS) encoded within Salmonella pathogenicity island 1 (SPI1). The genes encoding the SPI1 T3SS are tightly regulated by a network of interacting transcriptional regulators involving three coupled positive feedback loops. While the core architecture of the SPI1 gene circuit has been determined, the relative roles of these interacting regulators and associated feedback loops are still unknown. To determine the function of this circuit, we measured gene expression dynamics at both population and single-cell resolution in a number of SPI1 regulatory mutants. Using these data, we constructed a mathematical model of the SPI1 gene circuit. Analysis of the model predicted that the circuit serves two functions. The first is to place a threshold on SPI1 activation, ensuring that the genes encoding the T3SS are expressed only in response to the appropriate combination of environmental and cellular cues. The second is to amplify SPI1 gene expression. To experimentally test these predictions, we rewired the SPI1 genetic circuit by changing its regulatory architecture. This enabled us to directly test our predictions regarding the function of the circuit by varying the strength and dynamics of the activating signal. Collectively, our experimental and computational results enable us to deconstruct this complex circuit and determine the role of its individual components in regulating SPI1 gene expression dynamics.

Figure 1. SPI1 gene expression is hierarchical and exhibits a switch-like transition from the “off” to the “on” state.

(A) Diagram of the SPI1 gene circuit. HilA is the master SPI1 regulator as it activates the expression of the genes encoding the T3SS. HilA, in turn, is regulated by HilC, HilD, and RtsA. These three regulators can independently activate HilA expression. They can also activate their own expression and that of each other's. HilE represses the activity of HilD by binding to it and preventing it from activating its target promoters. (B) Time-course dynamics of P_hilD (pSS074), P_hilC (pSS075), P_rtsA (pSS076), and P_hilA (pSS077) promoter activities in wild-type cells as determined using luciferase transcriptional reporters. To induce SPI1 gene expression, cells were first grown overnight in LB/no salt and then sub-cultured into fresh LB/1% NaCl conditions to an OD of 0.05 and grown statically. Luminescence values were normalized with the OD₆₀₀ absorbance to account for cell density. Average promoter activities from three independent experiments on separate days are reported. For each experiment, six samples were tested. Error-bars indicate standard deviation. (C) Dynamics of P_hilA (pSS055) promoter activity in wild-type cells as determined using green fluorescent protein (GFP) transcriptional fusions and flow cytometry. The SPI1 gene expression was induced as described above. Samples were collected at the indicated times and arrested in their respective state by adding chloramphenicol. Approximately 30,000 cell measurements were used to construct each histogram. As a control, we expressed GFP from a constitutive promoter and observed continuous, rheostatic-like expression dynamics and a homogenous response in the population (Figure S1E). Strain genotypes and plasmid descriptions are provided in Tables 1 and 2 .

Figure 2. SPI1 gene expression is induced by a step increase in P_hilD promoter activity.

(A) Comparison of time-course dynamics for P_hilD (pSS074) promoter activities in wild type (black) and a ΔhilD mutant (JS253, red) as determined using luciferase transcriptional reporters. (B) Comparison of P_hilD (pSS072) promoter activity in wild type (black) and a ΔhilD mutant (JS253, grey) as determined using GFP transcriptional reporters and flow cytometry. Note that the activation of the P_hilD promoter is switch-like both in wild type and the ΔhilD mutant. Experiments were performed as described in Figure 1 .

Figure 3. HilC and RtsA amplify SPI1 gene expression.

(A) Comparison of time-course dynamics for P_hilD (pSS074, black) and P_hilA (pSS077, red) promoter activities in wild type (solid lines) and a ΔhilC ΔrtsA mutant (CR350, dashed lines) as determined using luciferase transcriptional reporters. (B and C) Comparison of P_hilA (pSS055, B) and P_hilD (pSS072, C) promoter activities in wild type (black) and a ΔhilC ΔrtsA mutant (CR350, grey) as determined using GFP transcriptional reporters and flow cytometry. Note that the loss of HilC and RtsA causes both a delay and decrease in P_hilA promoter activity whereas it causes only a decrease in activity in the case of the P_hilD promoter. Experiments were performed as described in Figure 1 .

Figure 4. HilE dampens SPI1 gene expression.

(A) Comparison of time-course dynamics for P_hilD (pSS074, black) and P_hilA (pSS077, red) promoter activities in wild type (solid lines) and a ΔhilE mutant (CR361, dashed lines) as determined using luciferase transcriptional reporters. (B) Comparison of P_hilA (pSS055) promoter activities in wild type (black) and a ΔhilE mutant (CR361, grey) as determined using GFP transcriptional reporters and flow cytometry. Similar results are also observed with the P_hilD promoter, though the phenotypic effect is much larger (Figure S4). Experiments were performed as described in Figure 1 .

Figure 5. Mathematical model is able to accurately capture SPI1 gene expression dynamics both for wild type and key mutants.

(A) Time-course simulation of HilD, HilC, RtsA, and HilA expression dynamics in wild-type cells. These results are the average of 1000 simulations. These simulations are meant to capture the population-level behavior of the circuit. (B) Time-course simulation of HilA expression at single-cell resolution. The expression values are normalized to one and plotted on a log scale. The expression values are given in relative log units (R.L.U.). Similar expression dynamics are also seen for HilD, HilC, and RtsA (see Matlab code provided as supplementary material). (C) Same results provided as a two-dimension heat plot, where the color intensity denotes the density of events. Note that the model captures the transient heterogeneity observed in our flow cytometry data where cells in both the “off” and “on” states are found at intermediate times. Panels A–C were generated from the same set of simulation runs. (D and E) Time-course simulation of HilD (D) and HilA (E) expression dynamics in wild type and ΔhilD, ΔhilC ΔrtsA, and ΔhilE mutants at population resolution. The results for each mutant were obtained from the average of 1000 simulations. Similar behavior is also seen at single-cell resolution. Mutants were simulated by setting the activity of the respective gene to zero in the model. A detailed description of the model is provided in the Materials and Methods.

Figure 6. Parametric analysis of model predicts that SPI1 gene circuit functions as an amplifier and encodes an activation threshold.

(A) Effect of positive feedback on HilD expression. Plot shows steady-state concentration of HilD as a function of the parameters and . The parameter specifies the degree by which the SPI1 regulators - HilC, HilD, and RtsA - can activate HilD expression, effectively the strength of positive feedback on HilD expression. The parameter specifies the strength of the signal activating HilD expression. (B) Effect of HilC and RtsA on HilD expression. Plot shows the steady-state concentration of HilD as a function of the parameters , , and . The parameters and specify the degree by which the SPI1 regulators - HilC, HilD, and RtsA - can activate HilC and RtsA expression, respectively. In other words, these parameters set the strength of feedback on HilC and RtsA expression. In these simulations, the parameters and were both varied in tandem: the numerical values for the two are the same. (C) Effect of HilE on HilD expression. Plot shows the steady-state concentration of HilD as a function of the parameters and . The parameter specifies the rate of HilE expression. Results for HilA are shown in Figures S6A–C. The black lines in the plots are used to denote the results obtained using the nominal parameters (aside from ). A detailed description of the model is provided in the Materials and Methods.

Figure 7. Rewiring SPI1 gene circuits demonstrates that HilE imposes threshold on activation.

(A) Comparison of time-course dynamics for P_hilA (pSS077) promoter activities in wild type (black), CR355 (ΔP_hilD::P_hilC ΔhilC, red) and CR356 (ΔP_hilD::P_hilC ΔhilC ΔhilE, blue) as determined using luciferase transcriptional reporters. In strain CR355, the P_hilD promoter was replaced with the P_hilC promoter in an otherwise ΔhilC background. In this strain, hilD is transcriptionally regulated in a manner identical to hilC. Strain CR356 is the same as CR355 except that it lacks HilE. (B) Dynamics of P_hilA promoter activity in CR356 as determined using green fluorescent protein (GFP) transcriptional fusions and flow cytometry. Note that the activation of the P_hilA promoter in CR356 is no longer switch-like but rather rheostatic in nature. Similar dynamics are seen with the P_hilC promoter (Figure S2C). Experiments were performed as described in Figure 1 .