All bacteria and plasmids used in this study are listed in Table 1. The O-serotype-converted Salmonella mutants were all derived from wild-type S. Typhimurium S100. Escherichia coli and S. enterica were grown at 37°C in Luria-Bertani (LB) broth or in LB agar. sacB gene-based counter selection in allelic exchange experiments was performed on LB agar containing 10% sucrose with no sodium chloride added and grown at 30°C (34). Media were supplemented with 25 µg/ml chloramphenicol for selection. Diaminopimelic acid (50 µg/ml) was added for the growth of the Δasd strain (35). Electrocompetent E. coli or S. enterica cells were prepared as described previously (36). In vitro growth rates of Salmonella strains were determined by optical density measurements.

DNA manipulations were performed using standard methods (37) and were approved by Division of Environmental Health and Safety of Sichuan Agricultural University. No restriction endonuclease sites were introduced when amplifying DNA fragments from chromosomes or plasmids. DNA concentrations and purity were measured using a Nanodrop ND-2000 spectrophotometer (Thermo Fisher Scientific). DNA fragments were assembled using Gibson Assembly Master Mix according to the manufacturer’s instructions (New England Bio Labs). PCR was performed in 20 µl reaction volumes in a Thermo Scientific Arktik Thermal Cycler. Reaction mixtures contained one volume of 1 µM (each) forward and reverse primers and DNA template (~50 ng plasmid DNA or ~100 ng chromosomal DNA) with an equal volume of 2× premix PrimeSTAR Max DNA polymerase (TaKaRa). Thermal cycler conditions were 98°C for 2 min; 30 cycles of 98°C for 1 min, 55°C for 30 s, and 72°C for 15–30 s/kb; and a final extension at 72°C for 10 min.

All primers used in this study are listed in Table S1 in Supplementary Material. We used a sacB gene-based suicide vector to counter-select deletion and/or insertion mutations. For deletion mutation suicide vector construction, two homologous DNA fragments, upstream and downstream of the gene or operon to be deleted, were amplified using primer pairs designated D-N-1F/D-N-1R and D-N-2F/D-N-2R (N represents the name of the gene or operon to be deleted). After DNA purification, these two fragments were fused by PCR using primer pairs designated D-N-1F/D-N-2R and linked into the suicide vector pYA4278 (38). For insertion mutation suicide vector construction, the allelic gene or operon to be inserted in a directed site was amplified using primer pairs designated (G)In-I-F/(G)In-I-R (I represents the name of the gene or operon to be inserted). The backbone of the suicide vector, which contained the homologous upstream and downstream sequence of the site to be inserted, was amplified using primer pairs designated (G)Vec-D-N-F/(G)Vec-D-N-R. All purified DNA fragments were assembled in order using Gibson Assembly (39), resulting in targeted gene or operon inserted into the directed site in a suicide plasmid. All allelic gene exchanges or whole-operon replacement in S. Typhimurium were achieved in two steps (Figure 1): deletion of the original gene or operon followed by insertion of the target gene or operon. Chromosomal modifications were introduced by suicide plasmids using standard methods (40, 41). All of the constructed mutants, either intermediates or final constructs, were routinely sequenced.

LPS were prepared and visualized by the method of Hitchcock and Brown (42). LPS samples were separated via 12.5% SDS-PAGE gels and transferred to nitrocellulose membranes using a Trans-Blot SD semidry transfer system (Bio-Rad, Hercules, CA, USA). Membranes were first incubated with O-antigen signal-factor rabbit antisera (BD Biosciences) or vaccinated murine pooled sera (1:100 dilution) followed by secondary anti-rabbit or anti-mouse horseradish peroxidase-conjugated antibody (Sigma) at a 1:1,000 dilution. Patterns were detected by chemiluminescence using western ECL blotting substrates (Bio-Rad).

P22HT int was propagated on S. Typhimurium S100 carrying the chromosomal integrated suicide vector pSS241 (43), which confers chloramphenicol resistance. Strains to be tested were grown overnight in LB broth at 37°C. Cultures were diluted 1:100 into fresh LB broth and grown at 37°C to an OD₆₀₀ of 0.6. Then, 10 µl of phage (1 × 10⁸ PFU) was mixed with 1 ml of bacteria (~5 × 10⁶ CFU) and incubated at 37°C for 30 min. After the incubation, the mixture was centrifuged and resuspended in 1 ml of PBS. A 100-µl aliquot was spread onto LB agar plate containing 25 µg/ml chloramphenicol. Colonies were counted after an overnight incubation at 37°C.

Motility assays were performed on LB plates containing 0.3% agar. The plates were allowed to dry at room temperature for approximately 2 h before the assays. Then, 6 µl of freshly grown bacteria (~5 × 10⁶ CFU) was spotted onto the middle of the plates and incubated at 37°C for 6 h. The diameters of the colonies (in millimeters) were measured.

The MICs of deoxycholate (DOC) and polymyxin B were determined using 96-well microtiter plates. Two-fold serial dilutions of DOC (0.39–59 mg/ml) and polymyxin B (0.078–10 µg/ml) were made along the plates. Bacteria were grown to an OD₆₀₀ of 0.6 and diluted to ~5.0 × 10⁴ CFU/ml in LB broth. Then, 100 µl of diluted bacteria suspension was added to each well and followed by an overnight incubation at 37°C. The optical density of each well was determined using an iMark™ Microplate Reader (Bio-Rad, Hercules, CA, USA). The threshold of inhibition was 0.1 at OD₆₀₀.

The human epithelial type 2 (Hep-2) cell line (ATCC strain CCL-6) was used to perform bacterial attachment and invasion assays as described previously (44). The bacteria were added to each well at a multiplicity of infection of 10:1. The percentage of attachment rate was calculated as follows: percentage of attachment = 100 × (number of cell-associated bacterial/initial number of bacterial added). The percentage of invasion rate was calculated as follows: percentage of invasion = 100 × (number of bacteria resistant to gentamicin/initial number of bacteria added).

All animal studies were conducted in compliance with the Animal Welfare Act and regulations stated in the Guide for the Care and Use of Laboratory Animals, which was approved by Sichuan Agricultural University Institutional Animal Care and Use Committee (Ya’an, China; Approval No. 2011028).

Six-week-old female BALB/c mice were purchased from Dashuo Biotechnology Co., Ltd. (Chengdu, China). To determine the 50% lethality dose (LD₅₀), bacteria were grown statically overnight at 37°C. Overnight cultures were diluted 1:100 into fresh LB media, grown at 37°C until reaching an OD₆₀₀ of 0.8–0.9. Cells were harvested by centrifugation at 3,452 × g at room temperature, washed once, and adjusted to the required inoculum density in buffered saline with gelatin (BSG). Groups of six mice each were infected orally with 20 µl of BSG containing various doses of S. Typhimurium S100 or its derivatives, ranging from 1 × 10⁴ to 1 × 10⁹ CFU. Animals were observed for 4 weeks after infection, and deaths were recorded daily. The LD₅₀ for each strain was calculated using the method of Reed and Muench (45). To evaluate colonization, groups of three mice were orally inoculated with 20 µl of BSG containing 1 × 10⁹ CFU bacteria. On days 4 and 8 post-inoculation, Peyer’s patches, spleen, and liver samples were collected. Samples were homogenized, and dilutions were plated onto MacConkey and LB agar to determine viable counts.

Groups of 12 mice each were inoculated orally with 20 µl of BSG containing approximately 1 × 10⁹ CFU vaccine strains on day 0 and boosted on day 14 with the same dose. Blood samples were collected after 28 days. Mice were challenged orally on day 56 with 5 × 10⁷ CFU of S. Typhimurium, S. Choleraesuis, or S. Enteritidis (~100 times LD₅₀).

S. Typhimurium and S. Enteritidis LPS were purchased from Sigma (St. Louis, MO, USA). S. Choleraesuis and S. Newport LPS were purified as described previously (46). A quantitative enzyme-linked immunosorbent assay (ELISA) was performed to determine serum antibody concentrations with the following modifications. Microtiter plates were coated with Salmonella LPS. The capture antibody, unlabeled goat antimouse IgG (H + L) (BD Pharmingen, San Diego, CA, USA) at 1 µg/ml in PBS, was added to extra uncoated wells to generate the standard curve. The plates were incubated overnight at 4°C, followed by blocking with PBS containing 5% BSA for 1 h at room temperature. For the LPS-coated wells, 100 µl of diluted serum was added to individual wells in triplicate. For the capture antibody-coated wells, the purified mouse IgG standard (for the standard curve quantification, BD Pharmingen, San Diego, CA, USA) was added, followed by two-fold serial dilutions starting at 0.5 µg/ml. The plates were incubated for 1 h at 37°C and then treated with biotinylated goat anti-mouse IgG (Southern Biotechnology Associates, Birmingham, AL, USA). The wells were developed with a streptavidin-alkaline phosphatase conjugate (Southern Biotechnology Associates, Birmingham, AL, USA), followed by a p-nitrophenylphosphate substrate (Sigma-Aldrich, St. Louis, MO, USA). Color development was recorded at 405 nm using an iMark™ Microplate Reader (Bio-Rad, Hercules, CA, USA). The ELISA standard curve was drawn using Curve Expert software (Hyams DG, Starkville, MS, USA). Serum antibody concentrations were calculated based on absorbance values and the standard curve.

Sera used for complement deposition assays were pooled sera taken from mice after the second immunization and were heated at 56°C for 30 min to inactivate endogenous complement. Bacteria were grown to an OD₆₀₀ of 0.8 and harvested by centrifugation at 6,000 rpm for 2 min. Bacterial pellets were washed, centrifuged, and resuspended to approximately 5 × 10⁸ CFU/ml in PBS. Then, 20 µl of bacterial sample was incubated with 80 µl of complement-inactivated sera at 37°C for 30 min. Bacteria were then washed once with PBS, resuspended and incubated with 100 µl of fresh naive BALB/c mouse sera at 37°C for 30 min. After another wash with PBS, the samples were incubated with 100 µl of FITC-conjugated goat anti-mouse complement C3c (Abcam) at a dilution of 1:100 on ice for 30 min in the dark. After incubation, the bacteria were washed with PBS, resuspended in 1% formaldehyde, and latter analyzed with a flow cytometer (BD FACSVerse™). The negative control was wild-type S. Typhimurium incubated with non-vaccinated complement-inactivated mice sera, and the positive control was wild-type S. Typhimurium incubated with complement-inactivated rabbit anti-O4 Salmonella sera (BD Biosciences). All other processes were the same as the test groups.

An in vitro assay was performed to analyze the differential uptake of S. Typhimurium, S. Enteritidis, S. Choleraesuis, and S. Newport by the RAW264.7 macrophage cell line. Briefly, 1 × 10⁵ RAW264.7 cells in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FBS (Newborn calf serum) and Pen/Strep were allowed to adhere to a 24-well plate for 24 h. Each well contained approximately 5 × 10⁵ RAW264.7 cells. Approximately 30 min prior to infection, the old media were replaced with fresh DMEM containing only 10% FBS. In each well, 5 × 10⁶ Salmonella opsonized with relative vaccinated mice pooled sera or naive serum (1 µl serum for a 100-µl volume of Salmonella in PBS) were added. Gentamicin was added to each well at a final concentration of 100 µg/ml at different time intervals of 0, 20, 40, and 60 min, and the plates were incubated for 1 h to kill non-phagocytized bacterial cells. After three washes with PBS, the macrophages were lysed with 1% Triton X-100 and appropriate dilutions were plated on LB agar plates. Colonies were enumerated the next day.

Data were analyzed using the GraphPad Prism 5 software (Graph Software, San Diego, CA, USA) by one-way or two-way ANOVA followed by Tukey’s multiple-comparison post-test. Kaplan–Meier survival curve comparisons were calculated by comparing two groups at each time point through the log-rank (Mantel–Cox) test. The data were expressed as the mean ± SEM. P < 0.05 was considered statistically significant.

To develop effective live attenuated vaccines against invasive NTS infections, we targeted the structurally hyper-variable O-antigens. The distinctive O-antigen gene clusters of groups B1, D1, C1, and C2 were compared (Figure S1 in Supplementary Material), together with their chemical structures, to identify the relevant sugar components and glycosidic linkages (Figure S2 in Supplementary Material). The genetic modifications we used to achieve the desired S. Typhimurium O-serotype conversions are shown in Figure 1. Specifically, (1) the abe gene in S. Typhimurium was replaced with prt-tyv_D1 from S. Enteritidis to convert the O4 serotype to O9, resulting in S1031 (Δabe:prt-tyv_D1) (O9) (Figure 1A); (2) the genes wzx_B1-wbaN were replaced with wzx_C2-wbaZ from S. Newport to convert O4 into O8, resulting in S1131 [Δ(wzx_B1-wabN):(wzx_C2-wbaZ)] (O8) (Figure 1B); and (3) the entire O-antigen gene cluster of group B1 was replaced with C1 from S. Choleraesuis to convert O4 into O7, resulting in S1124 [Δ(rmlB-wbaP):(wzy_C1-wzx_C1)] (O7) (Figure 1C). The LPS profiles of all these O-serotype-converted mutant strains were examined by silver staining and confirmed by western blotting (Figure 2). Notably, the LPS profiles of S1031 and S1131 differed from their parent S. Typhimurium, but exhibited similar patterns to S. Enteritidis (Figure 2A) and S. Newport (Figure 2C), respectively. However, the LPS profile of S1124 matched neither that of S. Typhimurium nor that of S. Choleraesuis (Figure 2B), but western blotting showed that S1124 generated LPS reactive with anti-O7 factor serum, indicating that O-antigen polysaccharide of S. Choleraesuis was successfully produced and ligated to S. Typhimurium core moiety in S1124 (Figure 2B). These O-serotype conversion mutations were later introduced into a live attenuated S. Typhimurium vaccine strain, S738 (Δcrp Δcya) (O4), resulting in S1075 (Δabe:prt-tyv_D1 Δcrp Δcya) (O9), S1157 [Δ(rmlB-wbaP):(wzy_C1-wzx_C1) Δcrp Δcya] (O7), and S1116 [Δ(wzx_B1-wabN):(wzx_C2-wbaZ) Δcrp Δcya] (O8), respectively, for further evaluation of immunogenicity and protective efficacy.

In vitro assays were done in triplicate. Phage P22 infections were performed to further examine the O-antigen structure of the mutants. The mutants were grown in LB broth and used as recipients for transduction assays. The number of transductions obtained from S1031 (O9) was similar to that from wild-type S100. However, we did not obtain any transductions from S1124 (O7) or S1131(O8) (Table 2).

We next evaluated the impact of the O-antigen modifications on virulence and survival attributes. The mutants were evaluated for their sensitivity to the bile salt DOC and the cationic antimicrobial peptide polymyxin B. The DOC MICs did not differ among these strains, whereas the polymyxin B MICs for wild-type S100 were twofold higher than those for S1124 (O7), S1031 (O9), and S1131 (O8) (Table 2). We observed slightly slower growth rates for S1031 (O9), S1124 (O7), and S1131 (O8) compared to wild-type S100, but the differences were not significant (Figure S3 in Supplementary Material). All of the mutants retained wild-type or near wild-type motility (Table 2). To be effective, a live attenuated Salmonella vaccine needs interact with host epithelial cells. Thus, we examined the ability of our Δcya Δcrp derivatives to attach to and invade Hep-2 cells. No significant differences among strains were observed (Figure S4 in Supplementary Material).

Wild-type S. Typhimurium S100, S. Enteritidis S246, and S. Choleraesuis S340 displayed high virulence, with LD₅₀ values of approximately 10⁵ CFU (47), whereas the LD₅₀ of wild-type S. Newport S264 was greater than 10⁹ CFU, indicating a non-virulent phenotype of S264 in the murine model. The LD₅₀ values of S1031 (O9), S1124 (O7), and S1131 (O8) were of a similar order of magnitude, approximately 10⁷ (Table 2). The colonization of each Δcya Δcrp vaccine candidate in murine Peyer’s patches, spleens, and livers was determined on days 4 and 8 after oral inoculation. All of the candidates displayed good colonization in Peyer’s patches, livers, and spleens, and no significant differences were observed among these groups. No deaths occurred during this period (Figure S5 in Supplementary Material).

To assess the immunogenicity of these vaccine candidates, mice were inoculated orally with approximately 10⁹ CFU of each strain on day 0 and boosted on day 14 with the same doses. Anti-S. Typhimurium, anti-S. Enteritidis, anti-S. Choleraesuis, and anti-S. Newport LPS serum antibodies were measured on day 28. The results are depicted in Figure 3. Mice vaccinated with S1075 (O9) mounted a significantly higher anti-S. Enteritidis LPS immune response than those vaccinated with S738 (O4). A similar result was observed in mice vaccinated with S1157 (O7) or S1116 (O8), which mounted significantly higher anti-S. Choleraesuis or anti-S. Newport LPS immune responses, respectively, than those vaccinated with S738 (O4). All vaccines induced a significantly higher IgG2a response than IgG1. The low level of IgG1/IgG2a ratio indicated that the cellular immunity was biased to Th1-type immune response (Figure S6 in Supplementary Material), consistent with our and other previous observations that Salmonella induced a predominant Th1-type response to either heterologous antigens or Salmonella own antigens (48, 49). Negative control groups (BSG) did not mount a detectable immune response. Apart from ELISA, we also performed western blotting to evaluate the sensitivity of polyclonal antibodies using pooled sera from vaccinated mice (Figure S7 in Supplementary Material). The sera from S738 (O4)- and S1075 (O9)-vaccinated mice were cross-reactive to LPS from S. Typhimurium and S. Enteritidis, while those from S1157 (O7)- and S1116 (O8)-vaccinated mice were specific to LPS from S. Choleraesuis and S. Newport, respectively. No positive bands were detected using pooled sera from the BSG control group.

C3 complement deposition is the key process for antibodies targeting surface antigens, leading to complement activation and subsequent serum bactericidal activity. Therefore, we determined the ability of serum antibodies from vaccinated mice to direct complement deposition on the surface of different wild-type Salmonella. Sera used in this assay were boosted pooled sera from mice vaccinated with S738 (O4), S1075 (O9), S1157 (O7), S1116 (O8), and equal-volume-mixed S1075 (O9) and S1157 (O7). The percentage of bacteria coated with C3 was determined by flow cytometry (Figure 4). Compared to negative controls, a high percentage of bacteria deposited with C3 complement on the surfaces of wild-type S. Typhimurium (Figure 4A), S. Enteritidis (Figures 4B,E), S. Choleraesuis (Figures 4C,F), and S. Newport (Figure 4D) were detected when incubated with mice sera induced by S738 (O4), S1075 (O9), S1157 (O7), and S1116 (O8), respectively. These results indicated that antibodies in mice sera induced by live vaccine candidates were able to trigger the classical pathway of complement activation. Furthermore, an in vitro assay was performed to analyze the differential uptake of S. Typhimurium, S. Enteritidis, S. Choleraesuis, and S. Newport by RAW264.7 macrophages. The aim of this assay was to evaluate the role of vaccine-induced antibody opsonization in the early stages of opsonophagocytosis. Inoculation with sera primed with a specific O-serotype-converted vaccine resulted in significantly increased uptake of the same O-serotype wild-type Salmonella by macrophages (Figure 5). The uptake of S. Enteritidis inoculated with sera from mice primed with S1075 (O9) was significantly higher than naive sera or other non-specific sera. Similar results were observed with S. Choleraesuis and S. Newport when opsonized with sera from mice primed with S1157 (O7) and S1116 (O8), respectively. In particular, the uptake of both S. Enteritidis and S. Choleraesuis by macrophages was significantly increased when opsonized with sera from mice co-vaccinated by S1075 (O9) and S1157 (O7).

Vaccinated mice were challenged orally on day 56 with a dose 100 times the LD₅₀ of S. Typhimurium, S. Choleraesuis, and S. Enteritidis to evaluate protective efficacy. When challenged with S. Enteritidis and S. Choleraesuis, 100% protection was observed in mice vaccinated with S1075 (O9) (Figure 6B) and S1157 (O7) (Figure 6C), respectively. Complete protection was also observed in all vaccinated mice when challenged with S. Typhimurium (Figure 6A). As wild-type S. Newport was non-lethal in mice at an oral challenge dose of 10⁹ CFU, the protective efficacy of vaccine S1116 (O8) or its combination with other vaccine candidates was therefore not evaluated. Most interestingly, mice vaccinated with mixed equal volumes of S1075 (O9) and S1157 (O7) were able to withstand challenges of S. Typhimurium, S. Enteritidis, and S. Choleraesuis, indicating that a S1075 (O9) and S1157 (O7) co-vaccination strategy may effectively prevent Salmonella serotype O4, O9, and O7 infections.