PgtE protease enables virulent Salmonella to evade C3-mediated serum and neutrophil killing

Abstract

Non-typhoidal Salmonella serovars, such as Salmonella enterica serovar Typhimurium (STm), are a leading cause of inflammatory diarrhea in otherwise healthy individuals. Among children, the elderly, and immunocompromised individuals, STm can spread to systemic sites and cause potentially lethal bacteremia. Phagocytic cells and the immune complement system are pivotal to preventing the dissemination of STm. PgtE, an STm outer membrane protease, has been previously described to cleave over a dozen mammalian protein substrates in vitro, including complement protein C3. However, these activities have mostly been observed with mutant, avirulent strains with a truncated O-antigen that renders bacteria sensitive to complement killing. Here, we report that virulent STm utilizes PgtE to evade complement-mediated killing in vivo. The wild-type pathogen increases pgtE expression and PgtE proteolytic function within macrophages and in macrophage-like in vitro growth conditions, concomitant with physiologic O-antigen truncation in these environments. Furthermore, we found that wild-type STm's resistance to complement-mediated serum and neutrophil killing is PgtE-dependent. We propose that PgtE promotes the systemic spread of STm by acting as a second line of defense against complement when STm escapes from a macrophage.IMPORTANCENon-typhoidal Salmonella serovars primarily cause gastrointestinal infections but can also lead to bacteremia through mechanisms that are not completely elucidated. Here we show that the outer membrane protease PgtE enables virulent Salmonella to evade complement-mediated killing in vivo, thereby promoting bacteremia. We also demonstrate that pgtE expression and PgtE proteolytic function are increased within macrophages and under macrophage-like growth conditions. Upon escaping from macrophages, PgtE protects against complement-mediated killing by serum and neutrophils, promoting the systemic spread of the pathogen. These findings highlight PgtE as a potential therapeutic target for preventing Salmonella bacteremia.

Keywords: Salmonella; bacteremia; complement; macrophages; neutrophils; proteases.

Fig 1

PgtE promotes smooth STm survival in vivo by evading complement C3. (A–E) 6- to 10-week-old C3^+/+ and C3^−/− littermates were injected intraperitoneally (IP) with 10⁴ CFU wild-type (WT) or isogenic PgtE-deficient (ΔpgtE) Salmonella strain IR715. Mice were euthanized 24 hours after infection, and bacterial burden in the (B) blood, (C) liver, and (D) spleen was quantified. (E) Weight loss = (weight at 24 hours/weight at time of infection) × 100%. (F–J) 6- to 8-week-old C57B6/J mice were IP-injected with PBS (Control) or cobra venom factor (CVF). 24 hours after treatment, mice were infected IP with 10⁴ CFU of either IR715 WT or IR715 ΔpgtE. Mice were euthanized 24 hours after infection, and bacterial burden was assessed in the (G) blood, (H) liver, and (I) spleen. (J) The concentration of complement C3 in plasma measured by ELISA: dotted line represents the average from three uninfected control mice. (B, G) The dotted line represents the limit of detection of STm CFU in blood. (B–E) N = 16–17 per group pooled from six independent experiments. (G–I) N = 15 per group pooled from three independent experiments. (J) ELISA from one representative experiment. (B–E, G–I) Outliers found by ROUT outlier analysis Q = 1% are removed. Data were analyzed by the Kruskal-Wallis test (non-parametric, non-paired) followed by Dunn’s multiple comparison test. Adjusted P values from Dunn’s multiple comparison test: *P < 0.05. **P < 0.01. ***P < 0.001. ****P < 0.0001. ns = not significant. (B–E) No significant difference was found between STm WT burden in infected C3^+/+ and C3^−/− mice (P value not shown). (G–I) No significant difference was found between STm WT burden in infected PBS-treated and CVF-treated mice (P value not shown). Symbols represent data from individual mice. Bars represent the (B–D, G–I) geometric means or (E, J) mean. (Panel A and F schemes created in BioRender [M. Lee, 2025, https://BioRender.com/d08j342].)

Fig 2

Wild-type STm cleaves complement C3 in a PgtE-dependent manner when grown in conditions that mimic the phagosome or when grown in macrophages. (A, B) Temporal and spatial distribution of PgtE-positive STm inside BMDMs. (A) BMDMs were infected with mCherry-STm carrying a plasmid encoding for a PpgtE::gfp transcriptional reporter fusion. Representative confocal microscopy images from 1 hour and 8 hours post-infection are displayed. GFP-positive bacteria (green), Salmonella (red), and the cell nuclei (DAPI; blue) are shown. White lines represent outlines of the approximate cell perimeters. Inset panels show 2× enlarged regions; scale bars are 10 µm. (B) Kinetics of intracellular pgtE expression in BMDMs. The number of GFP-positive bacteria at each timepoint was scored by fluorescence microscopy and reported as a percentage of total (red) bacteria (n = 3 experiments). (C, E) Smooth STm IR715 wild-type (WT), an isogenic ΔpgtE mutant, the ΔpgtE complemented in trans (ΔpgtE pPgtE), a rough E. coli transformed with a pWSK29 plasmid containing a functional pgtE gene (pPgtE) or a pgtE gene with a single point mutation inactivating PgtE (pPgtE D206A) were cultured overnight in (left) LB or (right) InSPI2 LowMg²⁺ minimal media. (D) Alternatively, STm was isolated from BMDMs 8 hours after infection. STm and E. coli were then incubated with normal human serum for (C) 8 hours or (D) 13 hours. PgtE-dependent complement cleavage in supernatants was assessed by western blot analysis with an anti-complement C3/C3b/iC3b/C3d antibody. (E) Alternatively, after overnight culture, STm and E. coli were lysed, run on a 4%–12% Tris-glycine gel, and stained with Pro-Q Emerald 300 Lipopolysaccharide Gel Stain Kit to assess the O-antigen chain length. (F) Expression of pgtE mRNA was assessed by qPCR in STm strains (WT, ΔpgtE, and ΔpgtE pPgtE) cultured in InSPI2 LowMg²⁺ minimal media; n.d., not detected. (G) Western blot analysis of STm WT or STm pgtE-FLAG cultured overnight in LB or InSPI2 LowMg²⁺ minimal media. The bottom half of the membrane was stained with anti-FLAG tag antibody. The top half of the membrane was stained with anti-DnaK as a loading control.

Fig 3

PgtE does not increase STm survival in macrophages under tested conditions. (A–F) BMDMs were infected at an MOI = 1 with IR715 WT, ΔpgtE, and ΔpgtE pPgtE that were either (A–C) not opsonized or (D–F) opsonized with normal mouse serum. (A, D) Thirty minutes after infection, prior to gentamicin treatment, BMDMs were lysed with 1% Triton-X 100 and STm CFU were enumerated. Alternatively, BMDMs were incubated with 100 µg/mL gentamicin for 30 minutes, followed by (B, E) 7 hours or (C, F) 23 hours with 20 µg/mL gentamicin and then lysed with 1% Triton-X 100. (A–F) N = 7 from three independent experiments. Symbols represent data from BMDMs from individual mice, bars represent the geometric means. Data were analyzed by the Kruskal-Wallis test (non-parametric, unpaired) followed by Dunn’s multiple comparison test. Adjusted P values from Dunn’s multiple comparison test: *P < 0.05.

Fig 4

PgtE promotes survival of smooth, virulent STm in serum. (A–C) Serum-killing assays were performed with smooth STm IR715 wild-type, isogenic PgtE-deficient (ΔpgtE), and ΔpgtE complemented in trans (ΔpgtE pPgtE). Strains were cultured overnight (A) in LB or (B, C) in InSPI2 LowMg²⁺ minimal media. STm at 10⁶ CFU/mL was then incubated with (A, B) 20% normal human serum (NHS) or (C) 20% C3-depleted human serum at 37°C shaking at 300 rpm. CFU were enumerated at 0 minutes, 45 minutes, and 90 minutes. Percent survival = (CFU at 45 minutes or 90 minutes/CFU at 0 minutes) × 100%. (A, C) n = 2, (B) n = 6 from 2 to 3 independent experiments. Bar and error represent geometric mean and standard deviation. Data were analyzed by two-way ANOVA followed by Sidak multiple comparison test. Adjusted P values from Sidak multiple comparison test: *P < 0.05.

Fig 5

PgtE promotes the survival of iNTS strain D23580 in serum when cultured in media mimicking the SCV luminal environment. (A–C) Serum-killing assays were performed with smooth STm D23580 wild type and an isogenic PgtE-deficient mutant (ΔpgtE). Strains were cultured overnight (A) in LB or (B and C) in InSPI2 LowMg²⁺ minimal media. STm at 10⁶ CFU/mL was then incubated with (A, B) 20% normal human serum (NHS) or (C) 20% C3-depleted human serum at 37°C shaking at 300 rpm. (A–C) CFU were enumerated at 0 minutes, 45 minutes, and 90 minutes. Percent survival = (CFU at 45 minutes or 90 minutes/CFU at 0 minutes) × 100%. (A, C) n = 2–3, (B) n = 6. (D, E) D23580 WT and ΔpgtE were cultured overnight in (left) LB or (right) InSPI2 LowMg²⁺ minimal media. (D) After overnight culture, STm was lysed, supernatants were run on a 4%–12% Tris-glycine gel, and the gel was stained with Pro-Q Emerald 300 Lipopolysaccharide Gel Stain Kit to assess O-antigen chain length. (E) Alternatively, STm was then incubated with NHS for 8 hours. PgtE-dependent complement cleavage in supernatants was assessed by western blot analysis with anti-complement C3/C3b/iC3b/C3d antibody. (F, G) IR715 WT and ΔpgtE and D23580 WT and ΔpgtE were cultured overnight in InSPI2 LowMg²⁺ minimal media. (F) The strains were then incubated with NHS for 8 hours, and complement cleavage was assessed by western blot as described in panel E. (G) Serum-killing assay was performed as in panesl A and B, and CFU were enumerated at 90 minutes (n = 4). The dotted line represents an average of inoculum concentrations. (A–C, G) Bar and error represent geometric mean and standard deviation. Data were analyzed by two-way ANOVA followed by Sidak multiple comparison test. Adjusted P values from Sidak multiple comparison test: *P < 0.05. ***P < 0.001.

Fig 6

PgtE enhances STm survival in neutrophil-killing assays and reduces complement-mediated neutrophil ROS response. Neutrophils were isolated (Stem Cell EasySep kit) from the bone marrow of (A–E) C57BL/6 mice and (E) Cybb-deficient mice. For neutrophil-killing assays, smooth STm IR715 wild-type and an isogenic PgtE-deficient (ΔpgtE) strain were cultured overnight in (A) LB or (B and C, E) InSPI2 LowMg²⁺ minimal media. STm was then (A and B [left]) not opsonized or (A and B [right], E) opsonized with normal mouse serum (NMS). (C) Alternatively, STm was opsonized with serum from C3^+/+ and C3^−/− littermates. (A–C, E) Neutrophils were then infected at an MOI = 10. STm CFU was enumerated 2.5 hours post-infection. Percent survival in neutrophils = (CFU in wells with neutrophils at 2.5 hours/CFU in control wells at 2.5 hours) × 100%. Each dotted line represents a paired experiment with neutrophils isolated from one mouse. (D) To determine neutrophil reactive oxygen species production, luminol assays were performed with STm cultured overnight in (left) LB or (right) InSPI2 LowMg²⁺ minimal media then opsonized with serum from (top) C3^+/+ and (bottom) C3^−/− littermates. Neutrophils were infected at an MOI = 10. Relative light unit reads were performed every 2 minutes with a BioTek Synergy HTX. Error bars represent mean + SD from three biological replicates from 1 of 3 representative experiments. (F–I) Eight-week-old Cybb^X−/X− females or Cybb^X−/Y hemizygous males were infected IP with 10⁴ CFU WT and ΔpgtE STm. Mice were euthanized 24 hours after infection, and bacterial burden in the (G) blood, (H) liver, and (I) spleen was assessed. (A–C, E) N = 5–10 from 3 to 4 independent experiments. Symbols represent data with neutrophils from individual mice, bars represent the means. (A–C, E) Data were analyzed by one-way ANOVA Kruskal-Wallis test followed by Dunn’s comparison test. Adjusted P values from Dunn’s multiple comparison test: *P < 0.05, **P < 0.01. (D) Data were analyzed by two-way ANOVA. Time × column factor: ****P < 0.0001. (D) bar and error represent mean + SD. (G–I) Symbols represent data from individual mice, bars represent the geometric means. (G) The dotted line represents the limit of detection. (G–I) N = 7–8 from two independent experiments. (Panel F scheme created in BioRender [M. Lee, 2025, https://BioRender.com/d08j342].)