Phage-resistance alters Lipid A reactogenicity: a new strategy for LPS-based conjugate vaccines against Salmonella Rissen

Abstract

Salmonella enterica serovar Rissen (S. Rissen) is an emerging causative agent of foodborne diseases. The current emergence of antibiotic resistance makes necessary alternative therapeutic strategies. In this study, we investigated the potential of a phage-resistant strain of S. Rissen (R^R) as a tool for developing an effective lipopolysaccharide (LPS)-based vaccine. The LPS O-antigen is known to play critical roles in protective immunity against Salmonella. However, the high toxicity of the LPS lipid A moiety limits its use in vaccines. Here, we demonstrated that the acquisition of bacteriophage resistance by S. Rissen leads to structural modifications in the LPS structure. Using NMR and mass spectrometry, we characterized the LPS from phage-resistant strains as a smooth variant bearing under-acylated Lipid A portions (penta- and tetra-acylated forms). We then combined RT-qPCR and NMR-based metabolomics to explore the effects of phage resistance and LPS modification on bacterial fitness and virulence. Finally, we conducted in vivo studies to determine whether lysogeny-induced remodeling of LPS affects the host immune response. Results revealed that the under-acylated variant of LPS from R^R attenuates the inflammatory response in BALB/c mice, while eliciting a specific antibody response that protects against S. Rissen (R^W) infection. In conclusion, our findings suggest that phage resistance, through lipid A modification, may offer a novel strategy for reducing LPS toxicity, highlighting its potential as a promising biological approach for developing LPS-based vaccines against Salmonella infections.

Keywords: bacteriophages; host immune response; lipid A; salmonella infection; vaccine.

Figure 1

The R^R strain expresses the prophage SieB gene. Analysis of mRNA expression levels displayed increased levels of SieB gene in the R^R strain - following AGS cell line infection - compared to R^W strain. Results are reported as graph bars and represent mean ± SD of three biological replicates from two independent experiments. Two-way ANOVA followed by Turkey post-hoc correction was used to compare the SieB gene expression between R^R and/or R^W groups, according to different time points. ****p < 0.0001.

Figure 2

Reflectron negative MALDI-TOF MS spectra of S. Rissen R^R and R^W Lipid A.

Figure 3

Orthogonal Projection to Latent Structure Discriminant Analysis (OPLS-DA) of bacteriophage-sensitive or resistant S. Rissen medium samples, before or after phage excision (R^W; R^R; R^SФ1-; R^WФ1-, respectively). (A) Scores plot showing a distinct separation of R^R group (blue squares) from R^W, R^WФ1- and R^SФ1- groups (green, red and purple squares, respectively) and of R^WФ1- and R^SФ1- groups (red and purple squares, respectively) from the R^W group (green squares). (B) Loading plot indicating NMR variables (chemical shift) of metabolites responsible for between-classes separation, characterized by |p(corr)| value > 0.7.

Figure 4

Orthogonal Projection to Latent Structure Discriminant Analysis (OPLS-DA) of bacteriophage-sensitive or resistant S. Rissen medium samples, before or after phage excision (R^W; R^R; R^SФ1-; R^WФ1-, respectively). (A) Scores plot showing a distinct separation of R^SФ1- group (purple squares) from R^WФ1- group (red squares). (B) Loading plot indicating NMR variables (chemical shift) of metabolites responsible for between-classes separation, characterized by |p(corr)| value> 0.7.

Figure 5

Determination of 50% Lethal Dose (LD₅₀) of R^R and R^W in BALB/c mice. Survival curves of mice infected with different concentrations of R^W strain (A) or R^R strain (B). Uninfected control group included mice injected with PBS vehicle (n = 6/group). ** p < 0.01 (Long-rank Mentel-Cox test). Data are representative of two independent experiments. LD₅₀ was calculated according to the Reed and Muench method. LD₅₀ = Log₁₀ end-point bacterial dose = Log₁₀ (bacterial dose showing mortality next below 50%) + {[(50% - mortality at bacterial dose next below 50%)/(mortality at bacterial dose next above 50% - mortality at bacterial dose next below 50%)] x Log₁₀ (differences between bacterial doses used in the assay)]}.

Figure 6

R^R-LPS does not induce endotoxemia. (A) Experimental design for the intraperitoneal (i.p.) injection of R^R or R^W-LPS. Ten-week-old, female BALB/c mice (n = 6 per group) were injected intraperitoneally with 20 mg/Kg of R^R or R^W-derived LPS (R^R-LPS and R^W-LPS, respectively) dissolved in PBS solution and then body weight and temperature were evaluated. Mice (n = 6) in the control group (Control) were injected intraperitoneally with the same volume of PBS solution. (B) Body weight was measured 0, 8 and 24 hours after LPS stimulation. Graph represents mean ± SD of values from individual mice (dots). (C) Bar graph represents changes (%) of body weight 24 hours after LPS injection. Data are representative of a single experiment. Two-way ANOVA followed by Turkey post-hoc correction was used to compare body weight among three groups (Control, R^R-LPS and R^W-LPS) at different times. One-way ANOVA followed by Turkey post-hoc correction was used to compare body weight changes among three groups. **p < 0.01; ***p < 0.001; **** p < 0.0001. (D) Body temperature was measured 0, 2, 8 and 24 hours following LPS injection. (E) Bar graph represents change of temperature 24 hours after LPS injection. Data are representative of a single experiment. Two-way ANOVA followed by Turkey post-hoc correction was used to compare body temperature between R^R-LPS and R^W-LPS groups at different times following LPS injection. One-way ANOVA followed by Turkey post-hoc correction was used to compare the mean temperature among three groups. *p < 0.05; **p < 0.01; ****p < 0.0001.

Figure 7

Cytokine levels in serum of BALB/c mice stimulated with R^R or R^W-LPS. Mice (n = 6 per group) were injected intraperitoneally with 20 mg/Kg of R^R or R^W-LPS dissolved in PBS solution. Mice in the control group (Control) were injected intraperiteonally with the same volume of PBS solution. Serum was collected every 2 hours for 8 hours and 24 hours following injection. The cytokines: (A) TNF-alpha; (B) G-CSF; (C) IL-6; (D) IFN-gamma; (E) IL-10; (F) IFN-beta and (G) RANTES/CCL5 were analyzed by ELISA. Data are representative of two independent experiments. Two-way ANOVA followed by Turkey post-hoc correction was used to compare cytokine levels among the three groups (control, R^R- and R^W-LPS groups). ****p < 0.0001.

Figure 8

R^R-LPS favors antibody response. (A, C) Total IgM, IgG and IgA antibody response elicited after immunization with R^R-LPS or R^W-LPS. Results are represented as mean titers (μg/mL) in serum or feces ± SD from female BALB/c mice (n = 6 per group) immunized intraperitoneally three times with: 1) R^R-LPS (10 μg/100 μL PBS, teal green); 2) R^W-LPS (10 μg/100 μL PBS, amaranth red) or 3) vehicle (100 μL PBS, grey). Graph represents mean ± SD of values from individual mice (dots). Data are representative of a single experiment. Two-way ANOVA followed by Turkey post-hoc correction was used to compare total antibody levels among three groups. * p <0.05; **p <0.01; ****p <0.0001. (B, D) IgM, IgG and IgA end-point anti-R^R-LPS or R^W-LPS antibody titers. Results represent the antibody response elicited on day 7 after last immunization and are represented as mean titers (μg/mL) in serum or feces ± SD. Data are representative of a single experiment. Two-way ANOVA followed by Turkey post-hoc correction was used to compare total antibody levels. ** p <0.01; ****p <0.0001. ns, not statistically significant.

Figure 9

Antibodies from mice immunized with R^R-LPS display bactericidal activity against WT S. Rissen. The bacteriophage-sensitive strain of S. Rissen (R^W, 1 x 10⁶ CFU/mL) was incubated at 37°C for 3 hours alone, with purified antibodies derived from mice immunized with R^R- or R^W-LPS, with complement or with both purified antibodies and complement. Data are represented as Log₁₀ of CFU/mL (OD₆₀₀ = 1 = 5 x 10⁸ CFU/mL) and represent means ± SD of three independent experiments, each performed in triplicate. One-way ANOVA followed by Turkey post-hoc correction was used to compare Log₁₀ CFU/mL among different experimental conditions. **p < 0.01; ***p < 0.001; **** p < 0.0001.