Expression of SPI-2 T3SS is bimodal

To detect the expression of the SPI-2 T3SS in Salmonella, we examined the production of the SPI-2 T3SS needle protein SsaG using GFP reporter systems. We constructed the pPssaG-GFP plasmid encoding GFP under the control of the ssaG promoter (PssaG) in Salmonella strain encoding mCherry fluorescent protein on chromosome for easier identification of bacteria or used a Salmonella strain expressing GFP as a transcriptional fusion to ssaG (Fig 1A) to control for plasmid-specific artifacts.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Expression from SPI-2 T3SS is bimodal.

(A) Schematic of the fluorescent reporter system. The constitutive Ptrc regulated expression of mCherry from chromosome. GFP was under the control of the PssaG, controlling expression of the SPI-2 T3SS needle protein, and was encoded either on a plasmid or on the chromosome (ssaG transcriptional fusion). (B, C) PssaG activity in vitro. Representative fluorescence microscopy image of GFP expression in STm WT + pPssaG-GFP (B) and STm ssaG::GFP (C) in a minimal medium MgMES pH 5 for 5 h. Scale bar, 2 µm. The arrowheads indicate bacteria with undetectable GFP fluorescence. (D) PssaG activity in SPI-2 inducing and non-inducing conditions. GFP expression in STm WT + pPssaG-GFP was analysed by flow cytometry at late exponential growth phase (OD 1.8 in LB, EX), in the minimal medium MgMES pH 5.0 (16 h, MgMES) and at late stationary growth phase (16 h in LB, ST). Negative gate (OFF) was set according to STm WT not carrying any reporter system. The bar chart shows quantification of GFP⁺ (ON) bacteria from 3 independent experiments in technical triplicates and show means ± SD. *** p < 0.001 (One-way ANOVA with Tukey HSD test). (E) PssaG activity over time. (i) GFP expression in STm WT + pPssaG-GFP was analysed by flow cytometry at indicated time after subculture from late stationary phase (16 h in LB) into MgMES pH 5. (ii) Quantification of GFP⁺ (ON) and GFP^- (OFF) bacteria as represented in (i). Data are from 3 independent experiments in technical triplicates and show means ± SD. (F) PssaG activity in macrophages. Representative fluorescence microscopy image of GFP expression in STm WT + pPssaG-GFP in RAW264.7 16 h p.i. The arrowheads indicate bacteria with undetectable GFP fluorescence. Maximal z axis projection. Scale bar, 10 μm. (G) PssaG activity in epithelial cells. Representative fluorescence microscopy image of GFP expression in STm WT + pPssaG-GFP in Mel JuSo cells 24 h p.i. The arrowheads indicate bacteria with undetectable GFP fluorescence. Maximal z axis projection. Scale bar, 10 μm. (H) Bimodality of PssaG activity in macrophages. RAW264.7 macrophages were infected with STm WT, STm ΔssrAB + pPssaG-GFP, STm WT + pPssaG-GFP, or STm WT + pPssaG-GFP-ASV for 16 h before hypotonic lysis. PssaG activity was monitored as GFP fluorescence in bacteria from cell lysate by flow cytometry. The bar plot represents quantification of GFP⁺ (ON) bacteria. Data are from at least 3 independent experiments in technical triplicates and show means ± SD. ***p < 0.001 (Paired two-sample t-test). (I) Bimodality of PssaG activity in epithelial cells. Mel JuSo cells were infected with STm WT, STm ΔssrAB + pPssaG-GFP, STm WT + pPssaG-GFP, or STm WT + pPssaG-GFP-ASV for 24 h before hypotonic lysis. PssaG activity was monitored as GFP fluorescence in bacteria from cell lysate by flow cytometry. The bar plot represents quantification of GFP⁺ (ON) bacteria. Data are from at least 3 independent experiments in technical triplicates and show means ± SD. ***p < 0.001 (Paired two-sample t-test).

https://doi.org/10.1371/journal.ppat.1013728.g001

To test the functionality of these reporters, we cultivated Salmonella PssaG reporter strains in MgMES pH 5.0, a minimal medium mimicking the conditions in the SCV, for 5 h and monitored GFP fluorescence using fluorescence microscopy and flow cytometry. Approximately two thirds of bacteria from both reporter strains showed high GFP fluorescence suggesting that the GFP reporters monitoring PssaG activity were functional in vitro (Fig 1B and 1C). As expected from a previous report [19], a subpopulation of bacteria did not show PssaG activity during growth in MgMES pH 5.0 (Fig 1B and 1C).

To test the responsiveness of the PssaG reporter, we compared the GFP fluorescence of Salmonella WT + pPssaG-GFP under SPI-2-inducing and non-inducing conditions. No GFP fluorescence was detected in exponentially growing Salmonella in LB (Fig 1D), a condition inducing SPI-1 T3SS expression while repressing SPI-2 T3SS expression [20]. However, shifting bacteria to MgMES pH 5.0 minimal medium inducing SPI-2 expression resulted in strong GFP fluorescence in approximately 80% of Salmonella WT + pPssaG-GFP. Statistical analysis [21] demonstrated that the distribution of bacteria into ssaG-non-expressing (SPI-2^OFF) and ssaG-expressing (SPI-2^ON) subpopulations followed a bimodal distribution.

Interestingly, Salmonella WT + pPssaG-GFP grown to the late stationary phase in LB displayed a weak GFP fluorescent signal in approximately 25% of bacteria. However, these bacteria did not form a distinct population suggesting that they either express ssaG at low levels not comparable to the fully virulent state or that the detected GFP signal resulted from increased autofluorescence (Fig 1D).

To investigate the development of bimodal SPI-2 T3SS expression, we analysed the development of GFP fluorescence (ssaG expression) in MgMES pH 5.0 over time. The separation of GFP^- and GFP⁺ subpopulations was already apparent 0.5 h after the transfer of late stationary Salmonella WT + pPssaG-GFP into MgMES pH 5.0 (Fig 1E), supporting an earlier finding that bacteria in stationary phase are ‘SPI-2-primed’ [20,22]. Although the vast majority of Salmonella WT + pPssaG-GFP expressed GFP, a subpopulation remained GFP^- throughout the experiment (Fig 1E).

To examine the influence of Salmonella growth phase on bimodal SPI-2 T3SS expression, we also studied the development of ssaG expression in bacteria transferred to MgMES pH 5.0 from the late exponential growth phase (OD₆₀₀ = 1.8). As expected, the onset of GFP fluorescence in Salmonella WT + pPssaG-GFP from the exponential culture was delayed in comparison to bacteria from late stationary culture (S1A Fig). However, the proportion of GFP⁺ Salmonella was similar between bacteria transferred from the late exponential and stationary culture after 5 h (Figs 1E and S1A).

To analyse ssaG expression in cellulo, we infected RAW264.7 macrophages or Mel JuSo epithelial cells with Salmonella WT + pPssaG-GFP. Salmonella grown to the stationary phase is commonly used to infect macrophages, whereas Salmonella from the late exponential phase is used to infect epithelial cells as the bacteria express SPI-1, enabling host cell invasion by Salmonella. Both SPI-2^OFF and SPI-2^ON subpopulations were detected in macrophages (Fig 1F and 1H) as well as in epithelial cells (Fig 1G and 1I). Importantly, Salmonella lacking the ssrAB two-component system regulating SPI-2 expression did not show GFP fluorescence suggesting that the reporter expression is fully dependent on the described SPI-2 master regulator.

Since GFP molecules have stability longer than the duration of our experiments, we used a destabilised GFP version (GFP-ASV) to identify Salmonella that expressed the SPI-2 T3SS just before the end of the experiment [23]. Although the bimodal distribution was preserved, we observed fewer GFP⁺ bacteria suggesting a dynamic regulation of SPI-2 expression (Fig 1H and 1I).

Ratiometric analysis of GFP expression from PssaG and mCherry expression from a synthetic promoter suggests that the bimodal distribution of ssaG expression is not caused by factors influencing overall transcriptome (S1B and S1C Fig). Similarly, the presence of the PssaG reporter system does not represent a growth disadvantage that could impose a pressure for decreased GFP expression (S1D and S1E Fig). Collectively, these data demonstrate that expression of the SPI-2 T3SS in host cells follows a bimodal pattern.

Expression of SPI-2 T3SS effectors is also bimodal

To examine whether the expression of SPI-2 T3SS effectors follows a bimodal pattern similar to that of the SPI-2 T3SS, we constructed the reporter plasmid pPsifA-BFP_PsteD-GFP encoding mTagBFP2 (referred to as BFP hereinafter) under the control of the PsifA and GFP under the control of the PsteD (Fig 2A). While SifA is crucial for SCV stability and Salmonella growth in macrophages, SteD reduces the interaction between dendritic cells and T cells, inhibits antigen presentation [24–28] but is dispensable for Salmonella intracellular growth.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Expression of SPI-2 T3SS effectors is also bimodal.

(A) Schematic of the fluorescent reporter system. The constitutive Ptrc regulated expression of mCherry from chromosome. BFP and GFP were under the control of the sifA, and steD promoter on a plasmid, respectively. (B) Activity of PsifA and PsteD expression in vitro. Representative fluorescence microscopy image of BFP and GFP expression in STm WT + pPsifA-BFP_PsteD-GFP in MgMES pH 5 for 5 h. The magenta arrow heads indicate expressing neither BFP nor GFP, blue arrowheads point to bacteria expressing more BFP than GFP, and green arrowhead points to a bacterium expressing more GFP than BFP. Scale bar, 2 μm. (C) Activity of PsifA and PsteD in epithelial cells. Representative fluorescence microscopy image of BFP and GFP expression in STm WT + pPsifA-BFP_PsteD-GFP in Mel JuSo cells 24 h p.i. Magenta arrowhead points to a bacterium expressing neither BFP nor GFP, blue arrowhead points to a bacterium expressing more BFP than GFP, and green arrowhead points to a bacterium expressing more GFP than BFP. Maximal Z axis projection. Scale bar, 10 μm. (D) Fluorescence profile of mCherry (Salmonella), BFP (PsifA), and GFP (PsteD) signal on the dotted line shown in (C). (E) Bimodality of PsifA and PsteD activity in macrophages. RAW264.7 macrophages were infected with STm WT + pPsifA-BFP_PsteD-GFP for 16 h before hypotonic lysis. PsifA and PsteD activity was monitored as BFP and GFP fluorescence, respectively, in bacteria from cell lysate by flow cytometry. The bar plot represents quantification of bacteria in depicted gates. Data are from 22 independent experiments in technical triplicates and show means ± SD. (F) Bimodality of PsifA activity measured using destabilized reporter in macrophages. RAW264.7 macrophages were infected with STm WT + pPsifA-GFP, or STm WT + pPsifA-GFP-ASV for 16 h before hypotonic lysis. PsifA activity was monitored as GFP fluorescence in bacteria from cell lysate by flow cytometry. The bar plot represents quantification of bacteria in depicted gates. Data are from 7 independent experiments in technical triplicates and show means ± SD. (G) Bimodality of PsifA and PsteD activity in vivo. C57BL/6 mice were inoculated orally with STm WT + pPsifA-BFP_PsteD-GFP. 5 days post inoculation, individual splenocytes were isolated and lysed as in (D). PsifA and PsteD activity was monitored as BFP and GFP fluorescence, respectively, in bacteria from cell lysate by flow cytometry. The bar plot represents quantification of bacteria in individual quadrants. Data are from 4 independently injected mice and show means ± SD. (H) Bimodality of PssaG and PsifA activity in macrophages. RAW264.7 macrophages were infected with STm WT + pPssaG-BFP_PsifA-GFP for 16 h before hypotonic lysis. PssaG and PsifA activity was monitored as BFP and GFP fluorescence, respectively, in bacteria from cell lysate by flow cytometry. The bar plot represents quantification of bacteria in depicted gates. Data are from 3 independent experiments in technical triplicates and show means ± SD. (I) Bimodality of PssaG and PsteD activity in macrophages. RAW264.7 macrophages were infected with STm WT + pPssaG-BFP_PsteD-GFP for 16 h before hypotonic lysis. PssaG and PsteD activity was monitored as BFP and GFP fluorescence, respectively, in bacteria from cell lysate by flow cytometry. The bar plot represents quantification of bacteria in depicted gates. Data are from 3 independent experiments in technical triplicates and show means ± SD.

https://doi.org/10.1371/journal.ppat.1013728.g002

Simultaneous microscopic and flow cytometry analysis of PsifA and PsteD reporter activity in Salmonella WT + pPsifA-BFP_PsteD-GFP grown in MgMES pH 5.0 (Fig 2B) as well as in bacteria inside host cells (Fig 2C–2E) revealed heterogeneous expression of the fluorescent reporter proteins. We detected bacteria not expressing either of the studied effectors, bacteria expressing higher amounts of sifA than steD, and bacteria expressing similar or higher amounts of steD than sifA. Similarly to SPI-2 expression, the expression of individual effectors also showed dynamic pattern as the destabilized pPsifA-GFP-ASV reporter (Figs 2F and S2A) also showed bimodal GFP distribution but overall less GFP⁺ bacteria. These results suggest that not all bacteria express all effectors in identical pattern.

Examination of reporter fluorescence kinetics upon bacterial transfer to MgMES pH 5.0 from late exponential or stationary phase revealed a bimodal expressional pattern for both PsifA (S1G and S1I Fig) and PsteD (S1H and S1J Fig) that was similar to that of the PssaG reporter. Importantly, bimodal expression from PsifA and PsteD was also recapitulated in spleens of C57BL/6 mice 5 days after oral infection with Salmonella WT + pPsifA-BFP_PsteD-GFP (Fig 2G).

To compare the expression of sifA and steD with that of the SPI-2 T3SS, we constructed the reporter plasmids pPssaG-BFP_PsifA-GFP and pPssaG-BFP_PsteD-GFP. Infection of RAW264.7 macrophages showed that all SPI-2^ON bacteria also expressed sifA (Fig 2H). Interestingly, a subpopulation of the SPI-2^ON bacteria did not express steD (BFP⁺; Fig 2I).

To examine expression differences between effectors linked to Salmonella proliferation in the SCV and other SPI-2 T3SS effectors, we compared expression patterns of sifA and steD with sseJ, involved in SCV stability [29], and steC, influencing actin cytoskeleton remodelling and cell migration [30,31]. To limit the potential impact of bacterial growth phase on the reporter system, we infected epithelial Mel JuSo cells with bacteria in late exponential phase (OD₆₀₀ = 1.8). Similar to Salmonella WT + pPsifA-BFP_PsteD-GFP (S2B Fig), Salmonella WT + pPsseJ-BFP_PsteD-GFP formed a subpopulation expressing sseJ, but not steD (BFP⁺; S2C Fig). Infection of Mel JuSo cells with Salmonella WT + pPsifA-BFP_PsseJ-GFP led to formation of only BFP^-GFP^- and BFP⁺GFP⁺ subpopulations (S2D Fig).

Infection of Mel JuSo cells with Salmonella WT + pPsteC-BFP_PsifA-GFP showed that expression of steC followed a pattern similar to sifA and sseJ. Infection of Mel JuSo cells with Salmonella WT + pPsteC-BFP_PsifA-GFP resulted in formation of only BFP^-GFP^- and BFP⁺GFP⁺ subpopulations (S2E Fig). In contrast, infection of Mel JuSo cells with Salmonella WT + pPsteC-BFP_PsteD-GFP resulted in formation of small BFP⁺ bacterial subpopulation expressing steC but not steD in addition to bacteria expressing both or neither effector (S2F Fig). Furthermore, we examined expression of several other effectors, all showing bimodal expressional pattern (S2G Fig).

To further examine the subpopulation expressing only one effector, we compared the BFP⁺ subpopulations among individual reporter Salmonella strains. We detected significantly more bacteria expressing sifA and sseJ than steC, while steD was expressed by the smallest bacterial proportion (S2H and S2I Fig). Considering that SifA and SseJ are involved in SCV function, SteC influences functions of the infected cell, while SteD shows paracellular activity, we hypothesise that the percentage of bacteria expressing particular effectors might depend on the importance of the effector for intracellular bacterial survival (i.e., SCV function > cytosolic function > paracellular function; S2H and S2I Fig).

Collectively, these data show that the bimodal expression of individual effectors results in the formation of distinct Salmonella subpopulations. We have also observed that effectors more critical for Salmonella survival being expressed by a greater proportion of bacteria. Thus, careful examination of effector expressional pattern might help with uncovering its physiological function.

SPI-2^OFF phenotype is not heritable

To test if the formation of individual subpopulations was caused by a heritable mutation, we infected Mel JuSo cells with Salmonella WT + pPsifA-BFP_PsteD-GFP, sorted out the BFP^-GFP^-, BFP⁺, and BFP⁺GFP⁺ bacteria, and examined their capacity to reestablish these populations after reinfection of Mel JuSo cells (Fig 3A). As expected, the first infection resulted in formation of all three subpopulations (Fig 3B). After regrowing the sorted bacteria to identical physiological state in the late exponential phase (OD₆₀₀ = 1.8) and reinfecting Mel JuSo cells, the bacteria from all three subpopulations again formed all three subpopulations in unchanged proportions (Fig 3C). This shows that no heritable defect was responsible for bimodal effector expression.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. SPI-2^OFF phenotype is not heritable.

(A) Schematic showing experimental procedure to track heritability of bimodal effector expression. Mel JuSo cells were infected with STm WT + pPsifA-BFP_PsteD-GFP for 24 h before hypotonic lysis. Liberated bacteria were FACS sorted based on their fluorescence intensity of BFP and GFP into BFP^-GFP^-, BFP⁺, and BFP⁺GFP⁺ populations. Sorted bacteria were grown separately in LB medium and then used to reinfect freshly prepared Mel JuSo cells for 24 h before hypotonic lysis. PsifA and PsteD activity was monitored as BFP and GFP fluorescence, respectively, in bacteria from cell lysate by the instrument used for FACS. Created in BioRender. Pospíšilová, M. (2025) https://BioRender.com/yzlr0ea. (B) Data obtained during the original sorting. Data are from 3 independent experiments in technical triplicates and show means ± SD. (C) Data obtained from reinfection of Mel JuSo cells with bacteria from the BFP^-GFP^-, BFP⁺, and BFP⁺GFP⁺ bacterial populations. (i) Representative zebra plots showing bacterial populations in reinfection samples. (ii) Quantification. Data are from 3 independent experiments in technical triplicates and show means ± SD.

https://doi.org/10.1371/journal.ppat.1013728.g003

Bimodal expression of SPI-2 T3SS and its effectors depends on mode of SsrB activation

To identify the mechanism regulating the bimodality of expression of the SPI-2 T3SS and its effectors, we first examined viability and metabolic capacity of SPI-2^OFF Salmonella. We constructed the reporter plasmid pPssaG-BFP_PBAD-GFP encoding BFP under the control of the PssaG, allowing the detection of SPI-2^ON bacteria, and GFP under the control of the inducible arabinose promoter BAD. Incubation of RAW264.7 macrophages infected with Salmonella WT + pPssaG-BFP_ PBAD-GFP with arabinose for 4 h led to GFP expression in approximately 90% of BFP⁺ bacteria (S3A Fig), suggesting that not all living bacteria (actively expressing BFP) respond to arabinose. Among BFP^- Salmonella, about 60% responded to arabinose, confirming that while some of BFP^- bacteria might be dead, the majority of SPI-2^OFF Salmonella were alive and sharing some metabolic capabilities with SPI-2^ON Salmonella inside host cells (S3A Fig). Similarly, the SPI-2^OFF and SPI-2^ON subpopulations showed no differences in the mCherry expression from the constitutive Ptrc (S1F Fig), suggesting that the global expression pattern did not differ between these subpopulations in RAW264.7 macrophages.

Following the invasion of epithelial cells, a fraction of Salmonella escape from the SCV, where the SPI-2 expression could be induced [32,33]. To determine whether SPI-2^OFF Salmonella reside in the SCV or are present in the host cytosol, we ectopically expressed the SCV marker LAMP1-mTurquoise in Mel JuSo cells before infecting them with Salmonella WT + pPssaG-GFP. We detected no difference in the localisation of LAMP1-mTurquoise to the vicinity of GFP⁺ and GFP^- bacteria (S3B Fig), suggesting that SPI-2^OFF bacteria reside in the SCV.

To investigate the involvement of signalling events and transcriptional regulators responding to the SCV environment, we constructed Salmonella ΔssrAB + pPssaG-GFP lacking SsrB, the master transcriptional regulator of SPI-2 T3SS expression, and its activator SsrA. As expected, Salmonella ΔssrAB + pPssaG-GFP failed to express GFP in response to nutrient limitation and acidic conditions in MgMES pH 5.0 (Fig 4A). In contrast, constitutive overexpression of SsrB^D56E-2HA (S4A Fig) with a point phosphomimetic mutation that enhances SsrB activity by mimicking the SsrB activation by SsrA inside the SCV [34] led to constitutive GFP expression even in SPI-2-repressing conditions during exponential growth (Figs 4A and S4B). Importantly, the GFP^- subpopulation was absent in Salmonella ΔssrAB + pPtrc-ssrB^D56E + pPssaG-GFP under all tested conditions, suggesting that either the presence or activity of SsrB is involved in regulation of bimodality of SPI-2 T3SS expression.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Bimodal expression of SPI-2 T3SS and its effectors depends on activation of SsrB.

(A) PssaG activity in vitro. (i) PssaG activity was monitored as GFP fluorescence in STm WT + pPssaG-GFP, STm ΔssrAB + pPssaG-GFP, and STm ΔssrAB + Ptrc-ssrB^D56E + pPssaG-GFP by flow cytometry after 16 h of growth in MgMES pH 5, at the late exponential phase OD 1.8 (EX) in LB (when SPI-1 expression is induced), or at the late stationary phase (16 h) in LB (when SPI-2 expression is primed). PssaG activity was monitored as GFP fluorescence in bacteria by flow cytometry. (ii) Quantification of GFP MFI in bacteria as represented in (i). Data are from 3 independent experiments in technical triplicates and show means ± SD. ns, p > 0.05; ***p < 0.001 (One-way ANOVA with Tukey HSD test). (B) Influence of SsrB on modality of PssaG activity in macrophages. (i) RAW264.7 macrophages were infected with STm WT + pPssaG-GFP, STm ΔssrAB + pPssaG-GFP, ΔssrAB + Ptrc-ssrB^WT + pPssaG-GFP, or STm ΔssrAB + Ptrc-ssrB^D56E + pPssaG-GFP for 16 h prior hypotonic lysis. PssaG activity was monitored as GFP fluorescence in bacteria from cell lysate by flow cytometry. (ii) Quantification of GFP MFI in bacteria as represented in (i). Data are from 3 independent experiments in technical triplicates and show means ± SD. **p > 0.01; ***p < 0.001 (One-way ANOVA with Tukey HSD test). (C) Influence of SsrB on modality of PssaG activity in presence of SsrA. (i) RAW264.7 macrophages were infected with STm WT + pPssaG-GFP, STm ΔssrB + pPssaG-GFP, ΔssrB + Ptrc-ssrB^WT + pPssaG-GFP, or STm ΔssrB + Ptrc-ssrB^D56E + pPssaG-GFP for 16 h prior hypotonic lysis. PssaG activity was monitored as GFP fluorescence in bacteria from cell lysate by flow cytometry. (ii) Quantification of GFP MFI in bacteria as represented in (i). Data are from 3 independent experiments in technical triplicates and show means ± SD. ns, p > 0.05; ***p < 0.001 (One-way ANOVA with Tukey HSD test).

https://doi.org/10.1371/journal.ppat.1013728.g004

To determine whether the bimodality of SPI-2 T3SS expression is regulated by the presence or activity of SsrB, we infected RAW264.7 macrophages with Salmonella WT + pPssaG-GFP, Salmonella ΔssrAB + pPssaG-GFP, Salmonella ΔssrAB + pPtrc-ssrB^WT + pPssaG-GFP, and Salmonella ΔssrAB + pPtrc-ssrB^D56E + pPssaG-GFP. As expected, flow cytometry analysis of PssaG activity in intracellular bacteria showed that macrophages infected with Salmonella WT + pPssaG-GFP contained SPI-2^ON and SPI-2^OFF bacteria, whereas macrophages infected with Salmonella ΔssrAB + pPssaG-GFP contained only the SPI-2^OFF bacteria (Fig 4B). Importantly, constitutive overexpression of ssrB^WT in Salmonella ΔssrAB + pPtrc-ssrB^WT + pPssaG-GFP resulted in weak, unimodal PssaG activity in the absence of ssrA. In contrast, constitutive overexpression of SsrB^D56E in Salmonella ΔssrAB + pPtrc-ssrB^D56E + pPssaG-GFP significantly increased PssaG activity yet more, without affecting its unimodal distribution (Figs 4B and S4C).

To dissect the individual roles of SsrA and SsrB in regulation of the bimodality of the SPI-2 T3SS expression, we infected RAW264.7 macrophages with Salmonella WT + pPssaG-GFP, Salmonella ΔssrB + pPssaG-GFP with intact ssrA but deleted ssrB, and its complemented variants Salmonella ΔssrB + pPtrc-ssrB^WT + pPssaG-GFP and Salmonella ΔssrB + pPtrc-ssrB^D56E + pPssaG-GFP. As expected, ssrA alone was not sufficient for induction of detectable PssaG activity (Fig 4C). Constitutive overexpression of ssrB^WT in Salmonella ΔssrB + pPtrc-ssrB^WT + pPssaG-GFP restored bimodal PssaG activity. The PssaG activity in the SPI-2^ON subpopulation of Salmonella ΔssrB + pPtrc-ssrB^WT + pPssaG-GFP almost matched that of the SPI-2^ON subpopulation in Salmonella WT + pPssaG-GFP. In contrast, constitutive overexpression of SsrB^D56E in Salmonella ΔssrB + pPtrc-ssrB^D56E + pPssaG-GFP led to unimodal PssaG activity that was, however, lower than that in SPI-2^ON Salmonella ΔssrB + pPtrc-ssrB^WT + pPssaG-GFP (Fig 4C). Collectively, these results suggest that while SsrB is essential for expression of the SPI-2 T3SS, the bimodality is regulated at the level of SsrB activation by SsrA rather than at the level of SsrB expression.

SPI-2^OFF bacteria benefit from activity of effectors translocated by SPI-2^ON bacteria in macrophages

To analyse the benefits of bimodal expression of the SPI-2 T3SS and its effectors inside host cells, we examined the growth benefit of SPI-2^OFF Salmonella in the presence or absence of SPI-2^ON bacteria. To mimic the SPI-2^OFF population, we used the Salmonella ΔSPI-2 strain lacking the entire SPI-2 island including the ssrAB two-component system. Salmonella ΔSPI-2 strains are deficient in survival and proliferation in macrophages [35–37]. As a source of SPI-2^ON bacteria, we used Salmonella WT, since the phosphomimetic ssrB^D56E inducing the SPI-2^ON bacteria also activates the expression of genes beyond those encoding the SPI-2 T3SS and its effectors [38,39].

First, we injected C57BL/6 mice intraperitoneally with either a 1:1 mixture of Salmonella mCherry WT and Salmonella GFP ΔSPI-2 or with each strain individually. The presence of bacteria in splenic macrophages was determined by flow cytometry 24 h p.i. (S5A Fig). As expected, we recovered significantly fewer Salmonella GFP ΔSPI-2-infected macrophages than Salmonella mCherry WT-infected macrophages from the spleens of mice injected with individual bacterial strains (S5B Fig). However, Salmonella GFP ΔSPI-2 exhibited significant growth benefit in macrophages that also contained Salmonella mCherry WT compared to Salmonella GFP ΔSPI-2-infected macrophages from the same double-infected mouse. (Figs 5A, 5B, S5C and S5D). However, this experimental setup does not distinguish between bacteria phagocytosed and transported to the spleen at different times p.i. and thus cannot control for import of bacteria proliferating under different selection pressure in other tissues.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. SPI-2^OFF bacteria benefit from activity of effectors translocated by SPI-2^ON bacteria in macrophages.

(A) Schematic of mouse coinfection simulating SPI-2^ON and SPI-2^OFF populations. STm glmS::GFP ΔSPI-2 simulating the SPI-2^OFF population and STm glmS::mCherry WT providing the SPI-2^ON population were mixed 1:1 and injected intraperitoneally into C57BL/6. 24 h post injection, spleens were dissociated and STm WT and STm glmS ΔSPI-2 fluorescence was examined in splenic macrophages by flow cytometry. The ratio of MFI of a particular fluorescent protein in double- and single-infected splenic macrophages was calculated as the growth benefit. Created in BioRender. Pospíšilová, M. (2025) https://BioRender.com/zngh9p8. (B) Presence of STm WT supports growth of STm ΔSPI-2 in vivo. C57BL/6 mice were injected intraperitoneally as described in A). The growth benefit values for using STm glmS::mCherry WT and STm glmS::GFP ΔSPI-2 from the same mouse are linked. Data are from 3 independent experiments in technical triplicates and show medians, Q1 and Q3. *p < 0.05, **p < 0.01, ***p < 0.001 (One-sample t-test, where hypothesized population mean (μ) = 1; # p < 0.05 paired two-sample t-test). (C, D) STm ΔSPI-2 has a growth benefit in the presence of STm WT in macrophages. RAW264.7 macrophages were infected with 1:1 mixture of STm glmS::mCherry WT and STm glmS::GFP ΔSPI-2. The amount of bacteria per one macrophage was determined at 0.5 and 8 h p.i. using and flow cytometry (C) and fluorescence microscopy (D). (C) (i) Representative flow cytometry zebra plot. (ii) Median fluorescence intensities (MFI) of populations 0.5 h p.i. (white columns) and 8 h p.i. (coloured columns). (iii) The growth benefit is calculated as the ratio of MFI of double-infected to MFI of single-infected splenic macrophages at 8 h p.i. normalized to the corresponding ratio at 0.5 h p.i. (also described in methods). Data are from 3 independent experiments in technical triplicates and show individual data points and means ± SD. *p < 0.05, ***p < 0.001 (One-sample t-test, where hypothesized population mean (μ) = 1; ## p < 0.01 paired two-sample t-test). (D) (i) Representative fluorescence microscopy images. Scale bars, 10 µm. (ii) Quantification of intracellular bacteria 8 h p.i. from (i). Data are from 3 independent experiments in technical triplicates and show individual data points and means. ns, *p < 0.05, **p < 0.01, ***p < 0.001 (Two-way ANOVA with Tukey HSD test).

https://doi.org/10.1371/journal.ppat.1013728.g005

To conduct the experiment in a synchronised manner, we repeated this experiment in cellulo using RAW264.7 macrophages. As expected, Salmonella GFP ΔSPI-2 proliferated less than Salmonella mCherry WT in RAW264.7 macrophages infected with individual strains (S5F Fig). Flow cytometry analysis of bacterial numbers in RAW264.7 cells showed that Salmonella GFP ΔSPI-2 had a significant growth benefit in macrophages that also contained Salmonella mCherry WT when compared to Salmonella GFP ΔSPI-2 in macrophages from the same sample but infected with the mutant alone (Figs 5C and S5E). A small, but significant growth benefit was also observed for Salmonella mCherry WT in double-infected macrophages, although this growth benefit was significantly smaller than that of Salmonella GFP ΔSPI-2 in the same macrophages (Fig 5C). Fluorescence microscopy revealed that Salmonella GFP ΔSPI-2 proliferated to the same levels as Salmonella mCherry WT in co-infected macrophages, whereas it exhibited a growth defect in single-infected macrophages (Fig 5D). Although it has been shown before that SseF and SseG can be shared among Salmonella residing in the same cell [40], we show here that all SPI-2 T3SS effectors necessary for survival and proliferation of Salmonella ΔSPI-2 can be provided in trans at levels sufficient for complementation of the SPI-2 deletion. This suggests that the SPI-2^ON bacteria facilitate the survival and proliferation of SPI-2^OFF bacteria.

SPI-2^OFF Salmonella provide the growth pool for SPI-2^ON bacteria

Expression of the SPI-2 T3SS and its effectors likely impose a significant metabolic burden in the nutrient-limited SCV, a burden not experienced by SPI-2^OFF bacteria protected by SPI-2^ON bacteria in the same host cell. As a result, SPI-2^OFF Salmonella may proliferate faster than SPI-2^ON bacteria in the same host cell. To test this hypothesis, we examined individual Salmonella WT + pPsifA-BFP_PsteD-GFP for both proliferation of and reporter expression in MgMES pH 5.0 using automated fluorescence microscopy followed by a deep learning analysis algorithm (S6A–S6C Fig).

After an initial lag phase, Salmonella WT + pPsifA-BFP_PsteD-GFP proliferated exponentially until the colony size precluded analysis of individual bacteria (S6D Fig). As expected, the median fluorescence intensity of all bacteria (S6E Fig) increased over time in MgMES pH 5.0 during the experiment. However, only subset of Salmonella WT + pPsifA-BFP_PsteD-GFP exhibited significant increase in BFP and GFP fluorescence, while the majority showed only minor changes in fluorescence intensity. The percentage of fluorescent bacteria peaked around 2 h after transfer to MgMES pH 5.0, then declined throughout the rest of the experiment (S6F Fig). Interestingly, this decline correlated with the onset of bacterial proliferation after the lag phase, suggesting that non-fluorescent bacteria with inactive sifA and steD promoters had either shorter lag times or proliferated more rapidly.

To analyse the rate of Salmonella WT + pPsifA-BFP_PsteD-GFP proliferation, we analysed distribution of individual bacteria into separate populations based on BFP and GFP fluorescence using automated unbiased FlowSOM algorithm. This analysis identified 3 bacterial subpopulations (S6G Fig), with higher BFP and GFP fluorescence correlating with longer doubling times (S6H Fig).

During the experiment, some initially BFP^-GFP^- bacteria began to show BFP and GFP fluorescence (S6A Fig). This further supports our finding that SPI-2 T3SS expressional pattern is not heritable (Fig 3) and can change even during a single infection event. These results suggest that SPI-2^OFF Salmonella may eventually give rise to SPI-2^ON Salmonella, potentially replenishing the effector-expressing subpopulation.

To determine whether the lower reporter fluorescence in faster proliferating bacteria reflected lower effector expression or resulted from fluorescence dilution during consecutive cell divisions, we measured fluorescence dilution of GFP pre-synthesised from an inducible promoter using flow cytometry (S7A and S7B Fig). Growth of Salmonella WT + pPssaG-BFP_PBAD-GFP to stationary phase in the presence of arabinose, which induces PBAD, led to a uniform GFP expression (S7C Fig). As expected, subsequent subculture into arabinose-free MgMES pH 5.0 induced PssaG activity in part of the bacterial population, while the amount of pre-synthesised GFP decreased with bacterial proliferation (S7D and S7E Fig). Importantly, the SPI-2^ON bacteria showed higher GFP fluorescence than the SPI-2^OFF bacteria when the culture reached stationary phase in MgMES pH 5.0 at 8 h post subculturing (S7F Fig). In contrast, we detected no fluorescence dilution when GFP was expressed constitutively from chromosome (Salmonella glmS::Ptrc-GFP, S7D Fig) throughout the 8 h growth period in MgMES pH 5.0 mimicking the SPI-2 inducing environment of SCV, suggesting that expression from SPI-2 promoters and maturation time of GFP is short enough to limit the fluorescence dilution effect when its promoter is active in the nutrient-limited conditions.

To investigate proliferation rates during infection, we tracked dilution of GFP synthesised by bacteria prior to host cell infection using flow cytometry. We infected RAW264.7 and Mel JuSo cells with Salmonella WT + pPssaG-BFP_PBAD-GFP grown in the presence of arabinose, which induces PBAD, leading to homogenous GFP expression (Fig 6A). Infection of RAW264.7 macrophages for 6 h in the absence of arabinose allowing for dilution of pre-synthesized GFP between individual daughter cells, was sufficient for detection of an emerging SPI-2^ON Salmonella subpopulation (Fig 6B). Although GFP fluorescence of SPI-2^ON bacteria was decreased compared to pre-infection levels, it remained significantly higher than in GFP^-BFP^- bacteria (Fig 6B panel ii). The majority of SPI-2^ON bacteria lost their preexisting GFP fluorescence by 8 h p.i. (Fig 6C panel ii). However, the strength of SPI-2 expression inversely correlated with bacterial proliferation rates (Fig 6B–6D panels iii).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. SPI-2^ON Salmonella replicates slower in host cells.

(A) Schematic of GFP fluorescence dilution in STm + pPssaG-BFP_PBAD-GFP upon infection of host cells. STm + pPssaG-BFP_PBAD-GFP was cultured in presence of 0.5% arabinose to induce uniform GFP expression (inset) until the infection of RAW264.7 macrophages (B and C) or Mel JuSo epithelial cells (D) that was in the absence of arabinose. The infected host cells were lysed at indicated time p.i. and BFP and GFP fluorescence was examined in bacteria from cell lysate by flow cytometry. Loss of GFP fluorescence indicates high replicative capacity. Created in BioRender. Pospíšilová, M. (2025) https://BioRender.com/4ble4tj. (B and C) RAW264.7 macrophages were infected as described in A) for 6 h (B) or 8 h (C). (i) Representative zebra plot showing GFP dilution in the SPI-2^ON and SPI-2^OFF bacteria shows ‘SPI-2^ON’ (BFP⁺), BFP^-GFP^- and GFP⁺ populations. (ii) Median of GPF fluorescence of ‘SPI-2^ON’, BFP^-GFP^- and GFP⁺ populations. ns, p > 0.05; ***p < 0.001 (One-way ANOVA with Tukey HSD test). (iii) The ‘SPI-2^ON’ population was divided into ‘GFP^High’ and ‘GFP^Low’ subpopulations and medians of their BFP fluorescence intensity was compared. Data are from 3 independent experiments in technical triplicates and show means ± SD. ***p < 0.001 (Paired two-sample t-test). (D) Mel JuSo cells were infected as described in A) for 24 h. (i) Representative zebra plot showing GFP dilution in the SPI-2^ON and SPI-2^OFF bacteria shows ‘SPI-2^ON’ (BFP⁺), BFP^-GFP^- and GFP⁺ populations. (ii) Median of GPF fluorescence of ‘SPI-2^ON’, BFP^-GFP^- and GFP⁺ populations. ns, p > 0.05; ***p < 0.001 (One-way ANOVA with Tukey HSD test). (iii) The ‘SPI-2^ON’ population was divided into ‘GFP^High’ and ‘GFP^Low’ subpopulations and medians of their BFP fluorescence intensity was compared. Data are from 3 independent experiments in technical triplicates and show means ± SD. ***p < 0.001 (Paired two-sample t-test). (E) Increase of the SPI-2^OFF Salmonella population in time. Reanalysis of all experiments examining the percentage of the SPI-2^OFF population (BFP^-GFP^-) in STm WT + pPsifA-BFP_PsteD-GFP in 8 h, 10 h, and 16 h p.i. Each dot represents percentage of SPI-2^OFF (BFP^-GFP^-) bacteria from a single independent experiment. **p < 0.01 (Two-sample t-test with Welch’s correction).

https://doi.org/10.1371/journal.ppat.1013728.g006

Infection of Mel JuSo cells with Salmonella WT + pPssaG-BFP_PBAD-GFP for 24 h resulted in the formation of two distinct SPI-2^ON subpopulations (Fig 6D). While GFP^Low SPI-2^ON bacteria showed GFP dilution similar to that of GFP^- SPI-2^OFF bacteria, the GFP^High SPI-2^ON bacteria proliferated at a moderate rate (Fig 6D). Interestingly, a subset of SPI-2^OFF bacteria maintained high GFP fluorescence throughout the experiments (Fig 6B–6D).

To determine the origin of the non-proliferating GFP^High SPI-2^OFF population and test whether these bacteria could be intramacrophage-induced persisters [4], we examined SPI-2 T3SS effector expression in non-proliferating intracellular bacteria. Consistently with previous findings [41], we detected effector expression in persisters (S8A Fig). However, significantly fewer effector-expressing bacteria were found among persisters compared to actively growing intracellular bacteria (S8C Fig). The effector-expressing persisters appeared to split further into low-expressing and high-expressing subpopulations (S8A and S8B Fig). These results suggest that the observed GFP^High SPI-2^OFF subpopulation (Fig 6) consists of persisters and bacteria killed after their entry into the host cells.

If SPI-2^OFF Salmonella proliferate faster, their relative proportion among intracellular bacteria should increase in time. To test this hypothesis, we reanalysed all our experiments examining the proportion of non-fluorescent bacteria at 8, 10 and 16 h p.i. of RAW264.7 macrophages. Indeed, we found a significant increase in the proportion of SPI-2^OFF bacteria at later infection stages (Fig 6E). Collectively, our results suggest that the SPI-2^ON bacteria pay a metabolic cost related to effector expression, leading to slower proliferation compared to SPI-2^OFF bacteria.

SPI-2^OFF bacteria drives Salmonella escape from infected host cells

The intracellular proliferation of Salmonella in host cells eventually leads to host cell death and the escape of Salmonella into the extracellular environment [33,42]. To examine whether the faster-proliferating SPI-2^OFF Salmonella escapes more readily from infected cells, we infected RAW264.7 macrophages with Salmonella WT + pPssaG-GFP until the bacteria start to escape from host cells (Fig 7A) probably due to the onset of host cell death [43]. A comparison of bacteria escaped into the extracellular space and those remaining in still-living host cells revealed a significantly higher proportion of the SPI-2^OFF bacteria in the extracellular space than within living host cells (Fig 7B), confirming that the faster growing Salmonella population is the major cause of host cell death. Similarly, the majority of bacteria still residing in live RAW264.7 or Mel JuSo cells exhibited high activity of PsifA and PsteD, while only small fraction of extracellular bacteria was fluorescent (Fig 7C and 7D) suggesting that bacteria escaping into the extracellular space originated from the rapidly proliferating SPI-2^OFF population.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Salmonella escaping from infected cells is predominantly SPI-2^OFF.

(A) Schematic of SPI-2 activity measurement in Salmonella escaped from infected cells. Macrophages or epithelial cells were infected with Salmonella reporter strains. Extracellular bacteria were killed with gentamicin and washed away 2 h before the end of the experiment. The BFP and GFP fluorescence of bacteria escaped into the extracellular medium (DMEM with 20 μg/mL gentamicin) during the last 2 h of the experiment was measured by flow cytometry. Created in BioRender. Pospíšilová, M. (2025) https://BioRender.com/c97s213. (B) RAW264.7 macrophages were infected for 8 h with STm WT + pPssaG-GFP. PssaG activity in intracellular bacteria (Intra) from cell lysate and in extracellular bacteria (Extra) collected from supernatant of infected macrophages was monitored as GFP fluorescence by flow cytometry. Data are from 5 independent experiments in technical triplicates and show means ± SD. ***p < 0.001 (Paired two-sample t-test). (C) RAW264.7 macrophages were infected for 8 h with STm WT + pPsifA-BFP_PsteD-GFP. PsifA and PsteD activity in intracellular bacteria (Intra) from cell lysate and in extracellular bacteria (Extra) collected from supernatant of infected macrophages was monitored as BFP and GFP fluorescence by flow cytometry. Data are from 3 independent experiments in technical triplicates and show means ± SD. (D) Mel JuSo cells were infected for 24 h with STm WT + pPsifA-BFP_PsteD-GFP. PsifA and PsteD activity in intracellular bacteria (Intra) from cell lysate and in extracellular bacteria (Extra) collected from supernatant of infected Mel JuSo was monitored as BFP and GFP fluorescence by flow cytometry. Data are from 4 independent experiments in technical triplicates and show means ± SD.

https://doi.org/10.1371/journal.ppat.1013728.g007

Collectively, our results have revealed that the expression of the SPI-2 T3SS and its effectors is bimodal. This bimodality was regulated on the level of the SPI-2 master regulator two-component system SsrA-SsrB. The SPI-2^ON Salmonella created a permissive niche in the SCV, enabling SPI-2^OFF Salmonella inhabiting the same host cell to proliferate, escape from the host cell, and potentially disseminate into new host cells.

Ethics statement

Experiments involving infection of C57BL/6 mice were conducted in accordance with the Animal Welfare Committee of the Institute of Molecular Genetics of the Czech Academy of Sciences, v. v. i., in Prague, Czech Republic. Handling of animals was performed according to the Guidelines for the Care and Use of Laboratory Animals, the Act of the Czech National Assembly, Collection of Laws no. 246/1992. Permission no. 54–2022-P were issued by the Animal Welfare Committee of the Institute of Molecular Genetics of the Czech Academy of Sciences in Prague.

Bacterial strains

Bacteria were grown in Luria–Bertani (LB) medium supplemented with ampicillin (100 µg/mL), kanamycin (100 µg/mL) or chloramphenicol (30 µg/mL) as appropriate. See S1 table for all bacterial strains.

Plasmid construction

All primers used are listed in S2 Table. Promoter-reporter fusions were created using overlap-PCR [58] and inserted into the pFCcGi plasmid [59] using either BamHI-NotI (for mTagBFP2 reporter) or NsiI-SphI (for GFPmut3 reporter). Promoter regions PsifA [60], PsteD [61], PssaG [62], PsseJ [63], PsteC [64], PsifB [60] and PsseL [65] were adapted from previously conducted studies. The ssrB and ssrB^D56E genes were inserted into the pWSK29 plasmid [66] using HindIII-BamHI. The ssrB^D56E mutation was created using overlap-PCR [58]. To insert mCherry into the Tn7 loci in Salmonella genome, Ptrc-mCherry was inserted into pGRG25 as previously described using NotI-SphI and subsequently integrated downstream glmS on chromosomal DNA [67]. All plasmids were checked by Sanger sequencing.

Cell culture

For experiments throughout the article, RAW264.7 and Mel JuSo were selected as model host cells. While RAW264.7 macrophages are routinely used for Salmonella research, Mel JuSo represent epithelial cells that do not support Salmonella hyperproliferation in the cytosol. Mel JuSo have been used for Salmonella research by us and others previously [68]. RAW264.7 mouse macrophages and human Mel JuSo cells were maintained in Dulbecco’s modified eagle medium (DMEM; Sigma) supplemented with 10% heat-inactivated endotoxin-free fetal calf serum (FCS; Sigma).

Cell infection

For infection of RAW264.7 macrophages, overnight Luria broth (LB) cultures of S. Typhimurium strains (OD₆₀₀ ~ 3.5 to 4.0) were added to adherent macrophages at an MOI of 6:1, centrifuged at 110 g for 5 min and incubated at 37°C for 30 min. Cells were washed 3 times with PBS and incubated in fresh medium containing gentamicin (100 µg/mL) to kill extracellular bacteria. After 1 h, the antibiotic concentration was reduced to 20 µg/mL, and cells were processed at indicated times post infection (p.i.).

For infection of Mel JuSo cells, overnight LB cultures of S. Typhimurium strains were subcultured 1:100 into fresh LB and incubated with shaking at 37°C to late exponential phase (OD₆₀₀ = 1.8) before being added to Mel JuSo cells at an MOI of 100:1 for 30 min. Cells were washed 3 times with PBS and incubated in fresh medium containing gentamicin (100 µg/mL) to kill extracellular bacteria. After 1 h, the antibiotic concentration was reduced to 20 µg/mL, and cells were processed 24 h p.i.

At the end of the experiment, the host cells were washed with PBS and lysed by hypotonic lysis in deionised water. The bacteria were fixed with 4% PFA at 37°C for 10 min prior to flow cytometry measurement analysis of effector expression.

Mice

6 to 8 week old C57BL/6 mice (Charles River) were housed as 6 mice per individually ventilated cage under Specified Pathogen Free conditions. For infection experiments for detection of effector expression, mice were inoculated by oral gavage with 1x10¹⁰ colony forming units (CFU) of stationary phase S. Typhimurium 14028 + pPsifA-BFP_PsteD-GFP (OD₆₀₀ ~ 3.5 to 4.0) in 200 µL of PBS containing 3% NaHCO₃. For competitive infection experiments, mice were injected intraperitoneally with mixture of 5x10⁶ colony forming units (CFU) of stationary phase S. Typhimurium 14028 glmS::mCherry or S. Typhimurium 14028 glmS::GFP (either WT or ΔSPI-2::Cm; OD₆₀₀ ~ 3.5 to 4.0) in 200 µL of sterile PBS.

Spleens and mesenteric lymph nodes were isolated at indicated time after infection and enzymatically dissociated in the presence of collagenase IV (1 mg/mL) and DNAse I (0.1 mg/mL) in RPMI-1640 supplemented with 10 mM HEPES and 2% FCS for 30 min at 37°C. Enzymatic digestion was stopped with addition of EDTA (2 mM final concentration). Cells were subsequently collected through a 70 µm cell strainer on ice and washed twice with Ca²⁺, Mg²⁺ free HBSS supplemented with 10 mM HEPES and 2% FCS and used for flow cytometry analysis. For analysis of effector expression by flow cytometry, collected cells were pelleted by centrifugation at 500 g for 5 minutes and lysed by hypotonic lysis in deionised water. Cell debris was removed by filtration through 35 µm nylon mesh (BD) prior flow cytometry measurement. A fraction of liberated bacteria was plated on nonselective LB agar plates. The colonies were replica-transferred on a selective LB agar to monitor the presence of the reporter plasmid. No plasmid loss was observed throughout the experiments.

Fluorescence microscopy

To analyse the growth benefit of Salmonella using fluorescent microscopy, RAW264.7 macrophages grown on glass coverslips were infected as described above for 8 h. The cells were washed with PBS, fixed in 4% paraformaldehyde in PBS for 20 min at 37°C, labelled with 0.5 µg/mL DAPI for 10 minutes, washed 3 times in PBS and twice in dH₂O and then mounted onto glass slides using Aqua-Poly/Mount (Polysciences). Coverslips were imaged using an Olympus IX83 - inverted confocal wide field microscope (Olympus Life Science).

To monitor SPI-2 T3SS expression using the reporters in infected cells by confocal microscopy, Mel JuSo cells grown on glass coverslips were infected as described above for 24 h p.i. The cells were washed with PBS, fixed in 4% paraformaldehyde in PBS for 20 min at 37°C, washed 3 times in PBS and twice in dH₂O and then mounted onto glass slides using Aqua-Poly/Mount (Polysciences). Coverslips were imaged using a Stellaris 8 inverted confocal laser-scanning microscope (Leica Microsystems, Germany).

To assess Salmonella growth rate and SPI-2 T3SS expression using the reporters in vitro, bacteria from stationary culture were washed twice with MgMES pH 5.0 and diluted to 1 x 10⁸ CFU/mL prior to seeding on a glass coverslip in 2 µL of MgMES pH 5.0. The bacteria were covered with thin agarose pad (1% agarose in MgMES pH 5.0) and grown at 37°C. A microscopy image was taken every 10 minutes to monitor fluorescence of mCherry, GFP, and BFP using a Dragonfly 503 spinning disk confocal microscope (Andor, Oxford Instruments, UK).

Image analysis

Using Olympus IX83 with cellSens Dimension software images in Z-stack were deconvoluted with Constrained Iterative deconvolution package. Bacteria were detected and counted in infected cells using CellProfiler. Then the data was analysed using the R environment.

Images from Dragonfly spinning disk confocal microscope were analysed using Omnipose deep learning segmentation package (https://omnipose.readthedocs.io/), a project that specializes in bacterial species segmentation. Time-lapse images of mCherry fluorescence were used to identify the objects of interest. The images were further analysed using DeLTA 2.0 (https://delta.readthedocs.io/en/latest/), a deep learning-based image processing pipeline for segmenting and tracking single cells in time. A pretrained tracking model („mothemachine“) was used to explore bacterial division-related phenomena, the pre-processing and analysis was preformed using Python 3.11. In total, 50 colonies were analysed.

The obtained tracking results were further processed using custom-written scripts in MATLAB R2023b (The MathWorks, Natick, MA). Fluorescence of GFP and BFP was extracted and lineage trees showing changes in fluorescence intensities were constructed. Fluorescence of mTagBFP2 and GFP of individual events was analysed at the time of their appearance and disappearance (i.e., when a new daughter cell appeared and when it underwent a new division). These values were used by FlowSOM to identify events sharing fluorescent characteristics [69].

Flow cytometry

To analyse the growth benefit of Salmonella using flow cytometry, RAW264.7 macrophages were infected as described above for 8 h, detached from plastic using 5 mM EDTA to limit the mechanical stress, and fixed in 4% paraformaldehyde in PBS for 20 min at 37°C.

The growth benefit was calculated using MFI (median fluorescence intensities) from double-infected macrophages divided by MFI of single-infected macrophages and normalized to identically counted MFI ratio at 0.5 h p. i.

To analyse expression of fluorescent reporters in individual bacteria, the infected host cells were washed with PBS and then lysed by osmotic lysis using dH₂O. Liberated bacteria were fixed in 4% paraformaldehyde in PBS for 20 min at 37°C.

Data were acquired using BD LSR II (BD Biosciences) or ZE5 (Biorad) cytometer and analysed using FlowJo software.

Viability test

For the viability test of the SPI-2^OFF Salmonella, RAW264.7 and Mel JuSo cells were infected with Salmonella WT + pPssaG-BFP_PBAD-GFP as described above. 4 h before lysis (i.e., 12 h p.i. for RAW and 20 h p.i. for Mel JuSo), GFP expression was induced by adding 0.5% arabinose. Bacteria were then collected in a lysate and analysed by flow cytometry as described before.

Fluorescence-activated sorting of bacteria and reinfection

For fluorescence-activated sorting of bacteria from Mel JuSo lysate and subsequent reinfection, Mel JuSo cells were infected as described above. At 24 h p.i. the infected host cells were washed with PBS and then lysed by osmotic lysis using dH₂O. Lysate was collected, and bacteria were sorted by BD Influx based on their fluorescence. Sorted bacteria were grown O/N in LB containing ampicillin (100 µg/mL) and the next morning were subcultured 1:100 into fresh LB and incubated with shaking at 37°C to late exponential phase (OD₆₀₀ = 1.8) before being added to a new batch of Mel JuSo cells at an MOI of 100:1. Reinfection then proceeded the same way as the infection. Lysate was collected, and bacteria were measured by BD Influx.

Test of proliferation rate

To test the metabolic activity of Salmonella during infection, Salmonella WT + pPssaG-BFP_PBAD-GFP was grown over night in LB medium with ampicillin (100 µg/mL) and 0.5% arabinose to induce GFP expression. RAW264.7 and Mel JuSo cells were infected for 6, 8 (RAW264.7 macrophages) or 24 h (Mel JuSo) in absence of arabinose. Bacteria for flow cytometry analysis were collected as described above.

Quantification and statistical analysis

Statistical significance was calculated using two-way ANOVA, one-way ANOVA followed by Tukey HSD test or One- or two- sample t-test with Bonferroni correction and pairwise comparison if necessary as indicated in figure legends. All statistical analysis was carried out using the R environment version 4.4.0 except for growth benefit from Fig 5 and percentage of infected macrophages from S4 Fig that were carried out in the GraphPad Prism v9 software. Summarising data tables used for statistical analysis are available on Zenodo (https://doi.org/10.5281/zenodo.17199173).

Test of bimodality

Data were first downloaded in *.csv files from a flow cytometry analysis using the ViolinPlots plugin. Then, GFP fluorescence intensity from each bacteria was tested as a whole for uni- and bimodality in the R environment using the LaplacesDemon package [21].