Overexpression of T3SS translocation signals in Salmonella causes delayed attenuation

ABSTRACT

Engineering pathogens is a useful method for discovering new details of microbial pathogenesis and host defense. However, engineering can result in off-target effects. We previously engineered Salmonella enterica serovar Typhimurium to overexpress the secretion signal of the type 3 secretion system effector SspH1 fused with domains of other proteins as cargo. Such engineering had no virulence cost to the bacteria for the first 48 hours post infection in mice. Here, we show that after 48 hours, the engineered bacteria manifest an attenuation that correlates with the quantity of the SspH1 translocation signal expressed. In IFN-γ-deficient mice, this attenuation was weakened. Conversely, the attenuation was accelerated in the context of a pre-existing infection. We speculate that inflammatory signals change aspects of the target cell’s physiology, which makes host cells less permissive to S. Typhimurium infection. This increased degree of difficulty requires the bacteria to utilize its T3SS at peak efficiency, which can be disrupted by engineered effectors.

INTRODUCTION

The ability to engineer bacterial pathogens is a useful tool for both uncovering details about microbial pathogenesis and for learning about host defenses. The secretion signals of type 3 secretion system (T3SS) effectors can be used to engineer bacteria that translocate novel proteins into the host cell cytosol (1–6). However, engineering a pathogen can lead to off-target effects. Overexpression of engineered T3SS proteins can block the formation of the needle complex, clog the secretion apparatus, or cause reduction in expression of other T3SS effectors (1–3, 7–9).

To determine how engineering a pathogen affects its virulence ability in vivo, the engineered pathogen needs to be compared to a non-engineered control. A competitive index infection model allows for direct comparison of two different strains of bacteria within a single mouse (10). This strategy allows for normalization of data within each mouse, which both reduces the experimental variability and reduces the total number of mice needed to achieve adequate power. Competitive index experiments have been frequently used to study Salmonella enterica serovar Typhimurium pathogenesis, where, as originally described by Beuzón and Holden, a mutant strain containing an ampicillin-resistant vector is competed against a wild-type strain containing a kanamycin-resistant vector, although alternative antibiotic selection can also be used (10–15). We have also used this competitive index approach to compare engineered S. Typhimurium strains that include transgenes on their ampicillin-resistant plasmid to wild-type kanamycin-resistant controls (16–19).

We previously engineered S. Typhimurium to induce regulated cell death in the infected host cell. First, we created a pyroptosis-inducing S. Typhimurium that constitutively expresses the NLRC4 agonist FliC (FliC^ON: P_sseJ fliC fliS), which revealed the in vivo utility of pyroptosis in clearing intracellular infection (19, 20). In this construct, the flagellar chaperone FliS is included to promote secretion of FliC. Engineering another NLRC4 agonist, PrgJ, corroborated the result, but no chaperones were included (20). We recently designed an apoptosis-inducing S. Typhimurium that expresses the pro-apoptotic BH3 domain of murine BID fused to the T3SS secretion signal of SspH1 via an HA tag (BID^ON: P_sseJ sspH1^SS-HA-Bid^BH3) (16). When the BH3 domain of BID is exposed, for example, by cleavage by granzyme B, this BH3 domain interacts with BAK and BAX to cause mitochondrial permeability, resulting in activation of caspase-9-driven apoptosis. Thus, the BID^ON strain attempts to trigger cell intrinsic apoptosis. We did not include a chaperone in this case as SspH1 is not known to require a chaperone for translocation (21).

These constructs successfully induce pyroptosis or apoptosis in bone marrow-derived macrophages in vitro (16–19). We used the competitive index model to show that pyroptosis-inducing S. Typhimurium is cleared compared to a vector control within the spleen during the first 48 hours post-infection (hpi) (16). However, apoptosis-inducing S. Typhimurium is not cleared compared to the vector control within the spleen during the first 48 hpi (16). Here, we report that although the engineering itself is neutral for 48 hpi, it causes attenuation that only manifests after 48 hpi.

RESULTS

Engineered S. Typhimurium are attenuated only after 48 hpi

As previously reported, apoptosis-inducing S. Typhimurium with the BID^ON plasmid (P_sseJ sspH1^SS-HA-Bid^BH3) compete equally with the non-engineered vector control (pWSK129) S. Typhimurium for the first 48 hours post-infection (hpi) within the spleen of wild-type (WT) mice [Fig. 1A and B, (16)]. Here, we now show that after 48 hpi, BID^ON no longer competes equally with the vector control. Instead, the vector control continues to replicate at a steady logarithmic rate, whereas after 48 hpi, BID^ON burdens stabilize (Fig. 1A and B), which could be explained by a lack of replication paired with no clearance of the existing bacteria. We previously showed that our engineered S. Typhimurium strains grow normally in vitro, suggesting this is a uniquely in vivo phenotype (16). S. Typhimurium strains infect both the spleen and liver; however, competitive indices in the liver are more variable, and thus we focused our studies on the spleen, which is more reproducible (Fig. S1A and B). This attenuation holds true during both low (10³ CFU)-dose and high (10⁵ CFU)-dose infection in the context of competitive index infection (Fig. 1A and B; Fig. S2A and B), as well as in mice infected with individual strains (Fig. S2C), and after either intraperitoneal or intravenous injection (Fig. S2D and E). Furthermore, this effect was not influenced by whether the BID^ON construct was in an ampicillin- or kanamycin-resistant plasmid (Fig. S2F and G).

Fig 1.

Engineered S. Typhimurium shows normal virulence before 48 hpi, but become attenuated after 48 hpi. (A and B) Time course of intraperitoneal competitive index infection in WT mice with BID^ON versus a vector control. Data combined from two independent experiments, with the line representing mean ± SD, n = 8 mice per timepoint. Data represented as competitive index ratio (A) or individual burdens of the vector and BID^ON (B). (C and D) Intraperitoneal competitive index infection with BID^ON or P_sseJ sspH1^SS-HA versus a vector control at 72 hpi. Data representative of three independent experiments. The line representing mean ± SD, n = 5 mice per condition. Data represented as competitive index ratio (C) or individual burdens of the vector and engineered strain (D). (E and F) Intraperitoneal competitive index infection with FliC^ON versus a vector control at 72 hpi. Data combined from two independent experiments, with the line representing mean ± SD, n = 7–9 mice per condition. Data represented as competitive index ratio (E) or individual burdens of the vector and FliC^ON (F). Data were analyzed using an unpaired two-tailed t test (C), a One-way ANOVA (A and E), or a two-way repeated measure ANOVA (B, D, F); n.s. P > 0.05, **P < 0.01; ****P < 0.0001.

We compared BID^ON S. Typhimurium, as a control, to an engineered control containing the same promoter, secretion signal, and HA tag, but lacking the pro-apoptotic BID^BH3 domain (P_sseJ sspH1^SS-HA) (16). Surprisingly, we found this SspH1^SS-HA control also had ~10-fold lower burdens than the vector at 72 hpi, showing that this difference in burdens was not due to BID^ON-induced apoptosis, but rather the combination of the P_sseJ promoter and/or the SspH1 translocation signal (Fig. 1C and D).

Our lab has also used S. Typhimurium strains engineered to express flagellin; FliC^ON uses the same P_sseJ promoter to express flagellin (FliC) and its chaperone (FliS) (P_sseJ fliC fliS) (17–19). FliC^ON S. Typhimurium strains are detected by NLRC4 both in vitro and in vivo, resulting in pyroptosis of the infected cell. In vivo, FliC^ON S. Typhimurium is cleared by pyroptosis at approximately a 10-fold clearance rate per day. When downstream pyroptotic signaling components are blocked, backup apoptotic pathways may instead be initiated. We previously showed that while this backup apoptosis is initiated in vitro, this did not translate to an ability of backup apoptosis to lead to clearance of FliC^ON S. Typhimurium in the spleen within the first 48 hpi (16).

We therefore wanted to investigate if FliC^ON S. Typhimurium also showed attenuation in mice deficient for pyroptosis at 72 hpi. In a side-by-side comparison, BID^ON in WT mice and FliC^ON in Gsdmd^–/– mice show the same kinetics of attenuation: both compete equally with the vector for the first 48 hpi, but show –10-fold attenuation in the spleen at 72 hpi (Fig. S3A and B). We further show that FliC^ON S. Typhimurium manifest this –10-fold attenuation even in Nlrc4^–/– mice that are unable to detect flagellin, as well as in Asc/Casp1/11^–/– mice that are unable to induce apoptotic backup pathways (Fig. 1E and F; Fig. S3C and D). This suggests that this attenuation is due to the engineered system and not a result of regulated cell death.

Tetracycline-inducible promoter does not result in attenuation

We have previously published a tetracycline-inducible version of FliC^ON, termed FliC^IND (tetR tetA fliC fliS) (19). We, therefore, wanted to investigate whether this tetracycline-inducible construct had the same attenuation as the constitutive construct. Mice were infected with a 1:1 ratio of engineered FliC^IND S. Typhimurium and a vector control S. Typhimurium and then treated with anhydrotetracycline for 3 days before harvest (Fig. 2A). In contrast to FliC^ON (Fig. 1E and F), 3 days of induction of the FliC^IND system did not result in attenuation in Gsdmd^–/– mice (Fig. 2B and C). We further determined that inducible versions of BID^ON (tetR tetA sspH1^SS-HA-Bid^BH3) and the SspH1^SS-HA (tetR tetA sspH1^SS-HA) control are also not attenuated at 72 hpi (Fig. 2D and E). As the inducible tetA promoter is weaker than the constitutive sseJ promoter (Fig. 2F and G), this suggests that the quantity of protein produced causes the attenuation.

Fig 2.

Engineered S. Typhimurium with tetracycline-inducible promoter are not attenuated after 48 hpi. (A) Schematic of the tetracycline-inducible infection model. (B–E) Mice were intraperitoneally infected with a 1:1 ratio of tetracycline-inducible engineered flgB S. Typhimurium (“Strain^IND”) and a vector control flgB S. Typhimurium. Bacterial burdens in the spleen were determined at 96 hpi. (B and C) Competitive index infection with FliC^IND versus a vector control. Data combined from two independent experiments, with the line representing mean ± SD, n = 8–9 mice per genotype. Data represented as competitive index ratio (B) or individual burdens of the vector and FliC^IND (C). (D and E) Competitive index infection with FliC^IND, BID^IND, or tetR tetA sspH1^SS-HA versus a vector control. Data combined from two independent experiments, with the line representing mean ± SD, n = 10 mice per condition. Data represented as the competitive index ratio (D) or individual burdens of the vector and engineered strain (E). (F and G) Western blot analysis of SPI2-induced S. Typhimurium treated with or without anhydrotetracycline (atc) to activate the tetracycline-inducible promoter. Data representative of two independent experiments. Data are shown as representative blot (F) or quantification of combined experiments (G). Data were analyzed using an unpaired two-tailed t-test (B), a one-way ANOVA (D), a two-way repeated measure ANOVA (C and E), or a one-sample t-test (G; hypothetical mean = 1); n.s. P > 0.05; *P < 0.05; **P < 0.01; ****P < 0.0001

Overexpression of secretion signal results in attenuation

We then wanted to isolate which component of the constitutive (P_sseJ) engineered system was responsible for this attenuation. We therefore pared down the P_sseJ sspH1^SS-HA construct into P_sseJ sspH1^SS- and P_sseJ-only constructs. We found that it is the presence of the SspH1 secretion signal that is responsible for the attenuation at 72 hpi (Fig. 3A and B). We wanted to see if this attenuation was unique to the SspH1 secretion signal, so we created new constructs using the SspH2 and SseJ secretion signals. When we infected mice with these constructs, we found that P_sseJ sspH2^SS had only minor attenuation at 72 hpi and P_sseJ sseJ^SS had no attenuation at 72 hpi (Fig. S4A and B). Since these constructs all used the same promoter and had the same optimized ribosomal binding site, we expected them all to produce similar amounts of protein. To verify this, we added an HA epitope for Western blot detection. Surprisingly, we detected vastly different amounts of protein by Western blot. While P_sseJ sspH1^SS-HA produced robust amounts of protein, P_sseJ sspH2^SS-HA produced very little detectable protein, and P_sseJ sseJ^SS-HA produced even less protein (Fig. S4C). We have not determined why these constructs have different protein expression levels, but this could be caused by variance in the mRNA half-life or protein stability. Therefore, the attenuation effect tracks with the quantity of the protein being produced.

Fig 3.

Overexpression of the SspH1 secretion signal is responsible for attenuation of engineered system. (A and B) Intraperitoneal competitive index infection with P_sseJ, P_sseJ sspH1^SS, or P_sseJ sspH1^SS-HA versus a vector control at 72 hpi. Data combined from two independent experiments, with the line representing mean ± SD, n = 8–10 mice per condition. Data represented as the competitive index ratio (A) or individual burdens of the vector and engineered strain (B). Data were analyzed using a one-way ANOVA (A) or a two-way repeated measure ANOVA (B); n.s. P > 0.05; ****P < 0.0001.

Promoter strength affects the magnitude of attenuation

Another way to change protein quantity is to alter the promoter strength. We, therefore, swapped the P_sseJ promoter driving sspH1^SS-HA for two other SPI2-driven promoters (P_sifB or P_sseI) (22). We first validated that changing the promoter altered the total protein produced via Western blot; P_sseJ produced the greatest amount of SspH1^SS-HA protein, P_sifB produced slightly less protein, and P_sseI produced barely detectable amounts in vitro (Fig. 4A and B). P_sseJ sspH1^SS-HA had the strongest attenuation at ~10-fold at 72 hpi, with P_sifB sspH1^SS-HA having weaker attenuation and P_sseI sspH1^SS-HA having no attenuation (Fig. 4C and D). The timing of these attenuation effects (Fig. 4E and F) was roughly 12 hours faster than the kinetics of SspH1^SS-HA-BID^BH3 bacteria (Fig. 1; Fig. S2A), suggesting that different cargo fused to the secretion signal can have minor effects on the rate of onset of attenuation. Overall, the strength of the attenuation correlated with the protein expression levels.

Fig 4.

The amount of engineered protein produced affects the magnitude of attenuation of engineered S. Typhimurium. (A and B) Western blot analysis of SPI2-induced S. Typhimurium. Data representative of four independent experiments. Data shown as representative blot (A) or quantification of combined experiments (B). (C and D) Intraperitoneal competitive index infection with P_sseJ sspH1^SS-HA, P_sseI sspH1^SS-HA, or P_sifB sspH1^SS-HA versus a vector control at 72 hpi. Data combined from two independent experiments, with the line representing mean ± SD, n = 9 mice per condition. Data represented as the competitive index ratio (C) or individual burdens of the vector and engineered strain (D). (E and F) Competitive index infection with Amp^R Vector (pWSK29), P_sseJ sspH1^SS-HA, P_sseI sspH1^SS-HA, or P_sifB sspH1^SS-HA versus a vector control. Mice were intraperitoneally infected with 5 × 10³ CFU each of engineered and vector control strains. Data combined from two independent experiments, with the line representing mean ± SD, n = 7–8 mice per condition. Data represented as the competitive index ratio (E) or individual burdens of the vector and engineered strain (F). Data were analyzed using a one-sample t-test (B; hypothetical mean = 1), a one-way ANOVA (C), a two-way ANOVA (E), or a two-way repeated measure ANOVA (D and F); n.s. P > 0.05; *P < 0.05; **P < 0.01; ****P < 0.0001.

We also created P_sifB and P_sseI versions of FliC^ON as a control and assessed clearance via pyroptosis in WT mice. We found that the original P_sseJ drove the strongest clearance, with P_sifB trending toward weaker clearance, and P_sseI showing the weakest clearance (Fig. 5A and B). As previously shown (19), clearance of FliC^ON occurs rapidly within the first 24 hours after infection, which was seen with P_sseJ and P_sifB, but clearance driven by the weaker P_sseI promoter manifested only at 48 hours (Fig. 5C and D). Therefore, even when small amounts of FliC are produced, this results in clearance, indicating that all three promoters are functional.

Fig 5.

Engineered S. Typhimurium is cleared in correlation to FliC production. (A and B) Intraperitoneal competitive index infection with P_sseJ fliC fliS, P_sseI fliC fliS, or P_sifB fliC fliS versus a vector control at 72 hpi. Data from one independent experiment, representative of two experiments (P_sseJ and P_sseI), with the line representing mean ± SD, n = 3–5 mice per condition. Data represented as the competitive index ratio (A) or individual burdens of the vector and engineered strain (B). (C and D) Competitive index infection with P_sseJ fliC fliS, P_sseI fliC fliS, or P_sifB fliC fliS versus a vector control. Mice were intraperitoneally infected with 5 × 10³ CFU each of engineered and vector control strains. Data from one independent experiment, representative of two experiments (P_sseJ and P_sseI), with the line representing mean ± SD, n = 3–4 mice per condition. Data represented as the competitive index ratio (C) or individual burdens of the vector and engineered strain (D). Data were analyzed using a one-way ANOVA (A), a two-way ANOVA (C), or a two-way repeated measure ANOVA (B and D); n.s. P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001

Attenuation is worsened by an inflammatory environment

Finally, we wanted to determine if this attenuation was caused by specific immune defenses. The attenuation of P_sseJ sspH1^SS-HA or P_sseJ sspH1^SS-HA-Bid^BH3 was not abrogated in knockout mice lacking several pyroptotic genes (Fig. 6A and B; Fig. S5A and B). Attenuation was also not lost in Bid^–/– mice that are resistant to certain forms of apoptosis (Fig. S5C and D). We did find a partial reduction in attenuation in Ifng^–/– mice, but not a complete return to the vector control (Fig. S5E and F). This suggests that the inflammatory state modifies how the attenuation is manifested. Remarkably, this effect does not manifest attenuation for roughly the first 48 hours of infection.

Fig 6.

Attenuation of engineered S. Typhimurium is worsened by a pre-existing inflammatory environment. (A and B) Intraperitoneal competitive index infection with P_sseJ sspH1^SS-HA versus a vector control at 72 hpi. Data combined from two independent experiments, with the line representing mean ± SD, n = 8–10 mice per condition. Data represented as the competitive index ratio (A) or individual burdens of the vector and P_sseJ sspH1^SS-HA (B). (C) Schematic of the ΔaroA S. Typhimurium pre-infection model. (D and E) Competitive index infection of P_sseJ sspH1^SS-HA or P_sseI sspH1^SS-HA versus a vector control with ΔaroA S. Typhimurium intraperitoneal pre-infection. Mice were intravenously infected with 5 × 10⁴ CFU each of engineered and vector control strains. Bacterial burdens in the spleen were determined at 48 hours post competitive index infection. Data combined from two independent experiments, with the line representing mean ± SD, n = 8–10 mice per condition. Data represented as the competitive index ratio (D) or individual burdens of the vector and engineered strain (E). Data were analyzed using a one-way ANOVA (A), a two-way ANOVA (D), or a two-way repeated measure ANOVA (B and E); n.s. P > 0.05; ***P < 0.001; ****P < 0.0001

We hypothesized that if this later-stage inflammatory state was pre-existing, then P_sseJ sspH1^SS-HA would be attenuated faster. To test this, we pre-infected mice with ΔaroA S. Typhimurium (carrying no plasmids); despite its auxotrophy, ΔaroA does provoke a pro-inflammatory response in the spleen [Fig. S5G, (23)]. After 3 days of ΔaroA S. Typhimurium infection, we infected mice with P_sseJ sspH1^SS-HA or P_sseI sspH1^SS-HA versus a vector control. Two days later, we harvested the spleen to quantify engineered S. Typhimurium (ampicillin-resistant) versus the vector S. Typhimurium (kanamycin-resistant); note that the ΔaroA strain lacks antibiotic resistance and was not quantitated (Fig. 6C). We found that P_sseJ sspH1^SS-HA had a worse attenuation in the ΔaroA S. Typhimurium pre-infection group compared to a PBS control (Fig. 6D and E). Pre-infection did not create an attenuated effect of the weak promoter construct P_sseI sspH1^SS-HA (Fig. 6D and E). These results suggest that the attenuation effect of SspH1 translocation signal overexpression depends upon the inflammatory state of the tissue in conjunction with the total amount of protein expressed.

DISCUSSION

Engineered pathogens can be a useful tool in uncovering details of microbial pathogenesis or host immune defenses. However, here, we show that over-expression of the T3SS secretion signal of SspH1 results in a delayed-onset attenuation of engineered S. Typhimurium. We were surprised to observe this attenuation manifesting only at timepoints after 48 hpi, but not earlier. The lack of attenuation at 48 hpi in vivo explains why this effect was not noticed in our earlier publications (16–19). Such attenuation is not seen in broth culture, and in mice, it is not affected by initial CFU dose, route of infection, or use of different antibiotic selection markers. We observed a similar effect with the FliC FliS-expressing plasmids in Nlrc4^–/– mice; however, it should be noted that this effect is much weaker than pyroptotic clearance in WT mice. Overall, this engineering-induced attenuation seems to be due to a combination of the total expression of the T3SS secretion signal combined with the host inflammatory state.

Why does this attenuation only manifest after 48 hpi? There are several possible explanations that could occur either within the bacterium or within the host response to infection. In considering bacterial effects, we do not know the relative quantity of the engineered proteins compared to endogenous T3SS effectors; it may be that our engineered protein represents a significant quantity of the total protein being translocated. Alternately, the engineering could indirectly affect the expression of other T3SS effectors, as has been seen during YopH overexpression in Yersinia pseudotuberculosis (3). Conversely, the engineered substrates could clog the SPI-2 T3SS, resulting in suboptimal effector translocation, analogous to how SptP^SS-GFP fusions can clog a T3SS apparatus (2, 9). Finally, overexpression of T3SS effectors can result in failure to assemble T3SS structures (2). In all these possible mechanisms, the engineered strain could be under-translocating necessary endogenous effectors.

It is surprising that the attenuation only manifests after 48 hpi. This suggests that the proposed under-translocation of endogenous effectors is at first inconsequential to the bacterium. We speculate that an aspect of the inflammatory environment then changes at later times. This change could either be from new elements being introduced to the inflammatory environment or simply that a slow increase in the overall amount of total inflammation across the course of infection reaches a threshold around 48 hpi at which the engineered bacteria are now disadvantaged. Our results with pre-infection and IFN-γ-deficient mice support this notion of a change in the in vivo environment causing the attenuation. This could increase the degree of difficulty for the bacterium to reprogram host cells. Perhaps in this new more challenging inflamed environment, S. Typhimurium requires nearly 100% SPI-2 T3SS efficiency in order to continue logarithmic replication.

MATERIALS AND METHODS

Plasmid construction

P_sseI was designed by targeting the 303 bp upstream of S. Typhimurium T3SS effector protein SseI, identified by the region between ORFs. Similarly, P_sifB was designed by targeting the 402 bp upstream of S. Typhimurium T3SS effector protein SifB. The secretion signals for S. Typhimurium T3SS effector proteins SspH2 (AA1-140) and SseJ (AA1-142) were previously determined by our lab (22). HA tags were added to sseJ^SS and sspH2^SS by PCR-amplifying these regions with 3’ primers that contained the HA tag. To create the various constructs listed below, the existing FliC^ON and pTA021 (P_sseJ sspH1^SS-HA) plasmids were digested to swap out the promoter, or secretion signal, as appropriate. For creating the tetracycline inducible constructs, the sspH1^SS-HA-Bid^BH3 or sspH1^SS-HA regions from BID^ON or pTA021 (respectively) were PCR-amplified and then inserted into a pWSK29 tetR tetA backbone derived from FliC^IND. To create pTA030 and pTA031, either the P_sseJ or P_sseJ sspH1^SS regions of P_sseJ sspH1^SS-HA were PCR-amplified from pTA021 and then inserted into pWSK29.

Bacterial strains and culture conditions

All Salmonella enterica serovar Typhimurium strains were derived from ATCC 14028s. ΔaroA 14028s S. Typhimurium was a gift from Sam Miller’s lab (unpublished). In some experiments, flgB 14028s S. Typhimurium was used to eliminate flagellin expression from chromosomal loci (Fig. 1A and B, one replicate; Fig. S1B, S2A and B, S2D and E, S5A and B, two replicates, Fig. S5C and D, one replicate, and Fig. S5E and F, two replicates). However, we found that experimental results were identical using wild-type S. Typhimurium; therefore, wild type 14028s S. Typhimurium were used for all other experiments, with the exception of the tetracycline-inducible infections (Fig. 2B through G), where the flgB mutation was used to provide the tetracycline-resistant background for S. Typhimurium. Plasmids are listed in Table 1. All strains were grown in 2 mL Miller’s LB Broth (Apex Bioresearch Products, Cat. 11–120) with appropriate antibiotics overnight at 37°C with aeration. For protein expression experiments, S. Typhimurium strains were induced to express SPI2 by back-diluting an overnight culture to OD600 = 0.026 in 3 mL SPI2-inducing media and then growing 16–20 hours in a 37°C shaker (24). SPI2 medium is prepared by combining 100 mL 5 x modified N-Minimal Salts (5 mM KCl, 7.5 mM (NH₄)₂SO₄, 0.5 mM K₂SO₄, 1 mM KH₂PO₄, 100 mM Tris, 100 mM BisTris, 200 mM MgCl₂, 100 mM Hepes in 500 mL dH₂O, pH 6.5), 2.5 mL 20% Casamino Acids (final 0.1% wt/vol), and 3.5 mL 40% glycerol (final 38 mM) with dH₂O to 500 mL final volume and was filter-sterilized using a GenClone vacuum filter system with a 0.22-µm filter (Genesee, Cat. #25–227) (24). SPI2-induced bacteria were washed once with PBS prior to calculating OD.

TABLE 1.

Plasmids

Plasmid	Alias	Abx	Notes	Reference
pWSK29	AmpR vector	Amp	Low copy vector	(25)
pWSK129	Vector or KanR vector	Kan	Low copy vector	(25)
pDM1	FliCON or AmpR FliCON	Amp	pWSK29 expressing fliC fliS from sseJ promoter	(19)
pTA007	BIDON, AmpR BIDON, or PsseJ sspH1SS-HA-BidBH3	Amp	pWSK29 expressing sspH1SS-HA-mBidBH3 from sseJ promoter	(16)
pTA021	sspH1SS-HA or PsseJ sspH1SS-HA	Amp	pWSK29 expressing sspH1SS-HA from sseJ promoter	(16)
pTA015	KanR FliCON	Kan	pWSK129 expressing fliC fliS from sseJ promoter	(16)
pTA016	KanR BIDON	Kan	pWSK129 expressing sspH1SS-HA-mBidBH3 from sseJ promoter	(16)
pTA024	PsseI sspH1SS-HA	Amp	pWSK29 expressing sspH1SS-HA from sseI promoter	This work
pTA025	PsifB sspH1SS-HA	Amp	pWSK29 expressing sspH1SS-HA from sifB promoter	This work
pTA026	PsseI fliC fliS	Amp	pWSK29 expressing fliC fliS from sseI promoter	This work
pTA027	PsifB fliC fliS	Amp	pWSK29 expressing fliC fliS from sifB promoter	This work
pTA028	tetR tetA sspH1SS-HA	Amp	pWSK29 expressing sspH1SS-HA from tetracycline-inducible promoter	This work
pEM087	tetR tetA fliC fliS	Amp	pWSK29 expressing fliC fliS from tetracycline-inducible promoter	This work
pTA003	tetR tetA sspH1SS-HA-BidBH3	Amp	pWSK29 expressing sspH1SS-HA-mBidBH3 from tetracycline-inducible promoter	This work
pTA030	PsseJ	Amp	pWSK29 expressing sseJSS promoter	This work
pTA031	PsseJ sspH1SS	Amp	pWSK29 expressing sspH1SS from sseJ promoter	This work
pTA032	PsseJ sspH2SS	Amp	pWSK29 expressing sspH2SS from sseJ promoter	This work
pTA033	PsseJ sseJSS	Amp	pWSK29 expressing sseJSS from sseJ promoter	This work
pTA034	PsseJ sspH2SS-HA	Amp	pWSK29 expressing sspH2SS-HA from sseJ promoter	This work
pTA035	PsseJ sseJSS-HA	Amp	pWSK29 expressing sseJSS-HA from sseJ promoter	This work

Mice and mouse infections

All mouse strains were bred and housed at Duke University in a pathogen-specific free facility. For infection, mice were transferred to a BSL2 infection facility within Duke University, and mice were allowed to acclimate for at least 2 days prior to infection. Wild-type C57BL/6J (Jackson Laboratory #000664), Nlrc4^–/– (26), Gsdmd^–/– (27), Casp1^–/– (27), Casp1^–/–Casp11^129mt/129mt (referred to as Casp1/11^–/–) (28), Asc^–/–/Casp1/11^–/– (referred to as Asc/Casp1/11^–/–) (29), Gsdme^–/– (30), Bid^–/– (31), Ifng^–/– (32), and GBP^Chr3–/– (33). Animal protocols were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of North Carolina at Chapel Hill or by the IACUC at Duke University and met guidelines of the US National Institutes of Health for the humane care of animals.

For competitive index infections, the inoculum was composed of a 1:1 ratio of the vector control (pWSK129) and experimental strain (“Strain^ON”) S. Typhimurium. In experiments using Kan^R BID^ON or Kan^R FliC^ON, pWSK29 was instead used as the vector control. Mice were infected intraperitoneally with 5 × 10² CFU per strain, unless otherwise specified. Wild-type C57BL/6 mice were used in all experiments unless otherwise indicated. Bacterial burdens in the spleen were determined at 72 hpi unless otherwise indicated. Ratio of the vector to Strain^ON is graphed as “competitive index.” For CFU data, paired vector and Strain^ON data from each mouse are connected by a single line. For the tetracycline-inducible model, mice were infected with 5 × 10² CFU per strain S. Typhimurium and then treated intraperitoneally with 0.1 mg/mouse anhydrotetracycline (Sigma Aldrich Cat. 37919–100MG-R) in PBS daily, as described in Fig. 2A. For the ΔaroA pre-infection model, mice were infected intraperitoneally with 10⁶ CFU of ΔaroA S. Typhimurium or treated with 200 µL PBS. About 72 hours later, mice were then infected intravenously with 5 × 10⁴ CFU per strain of indicated competitive index strains, as described in Fig. 6C. Spleens were harvested at indicated timepoints and homogenized in a 2-mL homogenizer tube (Fisher Brand Cat. 14–666-315) containing 1 mL sterile PBS and one 5 mm stainless steel bead (QIAGEN Cat. 69989). Spleens were homogenized using a Retsch MM400 homogenizer for 5 min at 30 Hz. After homogenization, lysates were serially diluted in a ratio of 1:5 in sterile PBS and plated on LB plates containing appropriate antibiotics. Plates were incubated overnight at 37°C, and the colony forming units were counted. Competitive index results are presented as vector CFU/experimental CFU, normalized to the ratio of the plated inoculum.

Western blot analysis and quantification

Bacteria were SPI2-induced, as described above. For tetracycline-inducible constructs, bacteria were induced with 200 ng/mL anhydrotetracycline during SPI2 induction. Bacteria were then spun down and concentrated into 100 µL PBS. 25 µL 4 x Laemmli Sample Buffer was mixed with 75 µL of the culture to lyse the bacteria. Samples were boiled for 5 min at 95°C and frozen at −80°C until analyzed. About 12 µL of the sample was loaded into a 4%–12% polyacrylamide TGX Stain-Free gel (Bio-Rad Cat. 4568086) and run for 1 hour 15 min at 15 mA per gel. Gel was UV-activated in order to visualize total protein. Protein was then transferred onto a 0.45-µm PVDF membrane (Millipore, Cat. IPFL85R), blocked with 5% non-fat dried milk in TBS plus 0.01% Tween (TBST) for 1 hour at room temperature, and incubated overnight at 4°C with mild agitation in 5% milk in TBST plus α-HA (1:2000, mouse, Biolegend, Cat. MMS-101R). Membranes were washed with TBST and then incubated for 1 hour at room temperature with goat anti-mouse (1:10000, Jackson ImmunoResearch, Cat. 115–035-062). ThermoFisher Scientific SuperSignal West Pico PLUS (Cat. 34580) ECL was used with a 150 second exposure. Images were taken using an Azure 500 Infrared Imaging System.

Protein bands were quantified using ImageJ (win-64 version 1.53) as described in (34, 35). Briefly, both the total protein and the HA antibody images were converted into 8-bit images. Then, the background for each image was saved as a new image using the “Subtract Background” function. The background image was then subtracted from the original 8-bit image using the “Image Calculator” function to create a uniform background. In this calculated image, the “Rectangular Selections” tool was used to draw uniform selections around each lane. Using the “Analyze→ Gels” tool, a histogram was created for each lane representing the total pixels detected in the lane, and the “Label Peaks” function provided quantification of the area under the histogram peak. For each sample, the area of the histogram for the HA band was divided by the area of the histogram for the total protein lane. This was to normalize each HA band to the total protein loaded for that sample. From there, the normalized HA band size of each sample was divided by the normalized HA band size of the pTA021 (P_sseJ sspH1^SS-HA) sample to determine the relative ratio of the SspH1-HA protein quantities across different promoters or secretion signals within each experiment.

Statistics

All statistical analyses were performed with GraphPad Prism 9. Discrete data were first assessed for normal distribution using a Shapiro–Wilk normality test. Data with normal distribution were analyzed with either unpaired two-tailed t test (two groups) or a one-way ANOVA (three or more groups). Discrete data that did not have a normal distribution were analyzed with a Mann–Whitney (two groups) or Kruskal–Wallis (three or more groups). Experiments with two factors were analyzed with a two-way ANOVA. Multiple comparisons were computed using Tukey’s multiple comparison test except for Fig. S1F, in which Šídák’s multiple comparison test was used to compare pre-selected groups. Western blot quantification was analyzed using a one-sample t-test with a hypothetical mean of 1.

DATA AVAILABILITY

All study data are included in the article and have been uploaded to LabArchives. Plasmid maps of novel constructs will be made available upon request and have been saved in Benchling.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/iai.00329-23.

Figure S1. iai.00329-23-s0001.tif.

Competitive indices in the spleen are more reproducible than the liver.

Figure S2. iai.00329-23-s0002.tif.

Attenuation of engineered S. Typhimurium is not affected by infection method.

Figure S3. iai.00329-23-s0003.tif.

FliC^ON S. Typhimurium are attenuated at 72 hpi in mice deficient for cell death signaling pathways.

Figure S4. iai.00329-23-s0004.tif.

Reduced expression of engineered secretion signal reduces attenuation.

Figure S5. iai.00329-23-s0005.tif.

Attenuation of BID^ON S. Typhimurium after 48 hpi partially driven by IFN-γ.

Supplemental legends. iai.00329-23-s0006.docx.

Legends for Fig. S1 to S5.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.