Bi-valent polysaccharides of Vi capsular and O9 O-antigen in attenuated Salmonella Typhimurium induce strong immune responses against these two antigens

Salmonella Typhi is the causative agent of typhoid fever in humans, responsible for approximately 21 million infections and 222,000 deaths globally each year. The current licensed vaccines provide moderate protection to recipients aged >2 years. Prior work on typhoid vaccines has focused on injectable Vi capsular polysaccharide or Vi–protein conjugates and live, oral attenuated S. Typhi vaccines to induce humoral anti-Vi antibodies, while the value and importance of anti-O9 antibodies is less well established. In this study, we constructed a S. Typhimurium strain that synthesizes Vi capsular antigen in vivo and produces the immunodominant O9-antigen polysaccharide instead of its native O4-antigen. The live recombinant attenuated S. Typhimurium mutants were effective in stimulating anti-Vi and anti-O9 antibodies in a mouse model, and the surface Vi capsular expression did not affect the immune responses against the O9 O-antigen polysaccharide. Moreover, the resulting anti-Vi and anti-O9 antibodies were effective at killing S. Typhi and other Salmonella spp. expressing Vi or O9 antigen polysaccharides and provided efficient protection against lethal challenge by S. Typhimurium and S. Enteritidis. Our work highlights the strategy of developing live attenuated S. Typhimurium vaccines to prevent typhoid fever by targeting the both Vi capsular and O9 O-polysaccharide antigens simultaneously.


INTRODUCTION
Typhoid fever is a systemic infection in humans caused by Salmonella enterica subsp. enterica serovar Typhi (S. Typhi), with symptoms of fever, chills, respiratory distress, and abdominal pain, which are often non-specific and clinically indistinguishable from other febrile illnesses. 1 Although the first inactivated S. Typhi vaccine was licensed >100 years ago, typhoid fever remains a major public health concern with high mortality and morbidity worldwide. The global burden of typhoid fever estimated was approximately 21 million infections and 222,000 deaths annually. 2 Typhoid fever outbreaks are frequently reported in Sub-Saharan Africa, Southeast Asia, and other developing countries, with infants, children, and adolescents being the most commonly infected. In developed countries, most typhoid fever cases occur among travelers returning from typhoid-endemic areas. 3 Typhoid fever is predominantly associated with poor sanitation and asymptomatic carriers are often a source of contamination in food and water. The emergence of antibiotic resistance in clinical isolates of S. Typhi has resulted in typhoid fever being increasingly difficult to treat. 4 In the short term, vaccination is the most effective and economic method to prevent this disease.
The World Health Organization (WHO) currently recommends two vaccines for controlling typhoid fever outbreaks, an injectable purified Vi polysaccharide vaccine and a live attenuated oral Ty21a vaccine in capsule formulation. However, both vaccines are only moderately protective (50-70%), and their efficacies in preschoolaged children (<5 years of age) are either not acceptable or unknown. 5 Unfortunately, there is a high incidence of typhoid fever in children aged <2 years in developing countries. 6 Therefore, new vaccines with higher protective efficacy and immune responses in younger age groups are urgently needed.
Current typhoid vaccine development focuses primarily on injectable Vi-protein conjugates and oral live attenuated S. Typhi. Compared to Vi-conjugate vaccines, live attenuated vaccines offer a needle-free alternative, inducing strong humoral and mucosal immune responses, T-cell responses, and long-term T-cell memory. The idea of modifying the expression of one or more of the tviABCDE genes required for Vi synthesis has been explored. In strain CVD909, expression of tviABCDE was placed under transcriptional control of the P tac promoter, resulting in constitutive synthesis of Vi capsular polysaccharide. 7,8 However, none of these live vaccine candidates were effective at stimulating strong humoral anti-Vi serum responses. 8,9 Another approach to modifying tviABCDE expression has been to replace the native P tviA promoter with the in vivo-inducible P ssaG promoter from the Salmonella pathogenicity island 2 (SPI-2) gene ssaG of S. Typhimurium. A virulent S. Typhimurium strain carrying this construct elicited high levels of anti-Vi serum IgG after a single oral administration in mice. 10 Antibodies against Vi and O9 O-antigen polysaccharide are each capable of directing complement-mediated killing of S. Typhi, though killing by anti-O9 antibodies is dependent on the level of Vi antigen expression. 11 The protection conferred by the Food and Drug Administration-licensed typhoid vaccine Ty21a (Vi − ) is believed to be due in part to increased serum and mucosal antibodies against S. Typhi lipopolysaccharides (LPS), 12 while high anti-LPS titers were noted in individuals vaccinated with the experimental strain Ty800. 13 In light of the importance of both antigens, we constructed an attenuated S. Typhimurium-vectored typhoid vaccine engineered to produce both Vi and O9 polysaccharides.
The Vi capsular polysaccharide of S. Typhi is a linear homopolymer of α-1,4-linked N-acetylgalactosaminuronate (Gal-NAcA) with 60-70% O-acetylation at the monomeric C-3 position. 14 The lack of free hydroxyl groups in the Vi capsular inhibits C3 fixation and prevents complement deposition on the S. Typhi surface. 15 Vi capsular synthesis is regulated by several regulatory systems, including ompR-envZ, 16 rscB-rscC, 17 and RpoS, an alternative sigma factor, 18 and the former two regulators will upregulate Vi synthesis upon encountering a condition from high osmolarity to low osmolarity, while RpoS is involved in fine-tuning the synthesis of Vi capsular polysaccharide during this stage. The genes required for the biosynthesis of the Vi capsular is exclusively found in Vi-expressing strains. The viaB locus, residing on a 134-kb DNA island termed Salmonella pathogenicity island 7 (SPI-7), is composed of 10 genes involved in regulation (tviA), biosynthesis (tviBCDE), and export (vexABCDE) of the Vi capsular polysaccharide. 19 Tight regulation of Vi capsular production is coordinated primarily via the TviA protein, a positive regulator of Vi capsular biosynthesis and a negative regulator of flagellar gene expression and secretion of SPI-1 effectors. 20,21 The tviA promoter (P tviA ) is repressed under the high-osmolarity conditions in the intestinal lumen but is rapidly induced in the low-osmolarity environment present in tissues. 22,23 The VexE protein is required for Vi capsular anchoring in the outer membrane, and deletion of vexE leads to the extracellular release of the Vi capsular. 24 Vi capsular biosynthesis initiates from the inner plasma membrane, and mutations in genes encoding the export machinery (vexABCD) results in intracellular Vi capsular accumulation. 19 The O-antigens of S. Typhimurium (B1 group, immunodominant O4 serotype) and S. Typhi (D1 group, immunodominant O9 serotype) share a common trisaccharide backbone of 2)-α-Man(1→4)-α-Rha-(1→3)-α-Gal-(1→. 25 The unique dideoxyhexose sugars abequose and tyvelose contribute to O4 or O9 serogroup specificity, respectively. The gene cluster accounting for O-antigen polysaccharide synthesis of Salmonella O4 and O9 is located between galF and gnd in the chromosome. 26 The main differences between the gene clusters are the genes responsible for synthesis of these two unique dideoxyhexose sugars. 27 We have shown previously that S. Typhimurium with abe gene replacement with prt-tyv D1 from S. Enteritidis could convert the O4 serotype to O9. 28 In this study, we replaced the native tviA promoter (P tviA ) in the viaB locus with ssaG gene promoter (P ssaG ) 10 and then introduced this in vivo-regulated viaB locus into an O9 serotype-converted live attenuated S. Typhimurium vaccine to stimulate production of both anti-Vi capsular and anti-O9 O-antigen polysaccharide antibodies. Our results showed that the live recombinant attenuated S. Typhimurium mutants were effective in stimulating anti-Vi and anti-O9 antibodies in a mouse model and that the resulting anti-Vi and anti-O9 antibodies were effective at killing S. Typhi and other Salmonella spp. expressing Vi or O9 antigen polysaccharides.
Sustainable Vi capsular expression in O9 serotype-converted S. Typhimurium O9 O-antigen polysaccharide and Vi capsular production was visualized by silver staining and confirmed by western blotting (Fig. 2). The wild-type S. Typhimurium and S. Enteritidis strains display long O-antigen polysaccharides containing 20-100 repeating units (Fig. 2a), while most of the S. Typhi O-antigen polysaccharides are short. 32 Introduction of the Δ(wza-wcaM)::viaB and Δabe::prt-tyv D1 mutations did not influence LPS length, with all S. Typhimurium strains exhibiting the expected long-chain Oantigen polysaccharide (Fig. 2a). Western blotting results showed that the O-antigen produced by S. Typhimurium strain S1130 reacted with anti-O9 antisera and no longer reacted with anti-O4 antisera, as expected (Fig. 2b, c).
Production of Vi capsular by strains S1123 (O4, Vi + ) and S1130 (O9, Vi + ) mutants was confirmed by western blotting using anti-Vi antisera (Fig. 2d). When S. Typhi invades the mucosal epithelium or phagocytic cells, Vi capsule is produced to evade innate immune responses. 33,34 Despite these findings, there is evidence that the production of Vi antigen is downregulated in the phagosomal compartments in vivo, 35 with little or no Vi production in the spleen and liver. 10 However, placing viaB under transcriptional control of the SPI-2 P ssaG promoter allows sustained Vi production within the Salmonella-containing vacuole in macrophages. 10,29 We infected RAW264.7 cells with S1151 (O9, Vi + , ΔP tviA ::P ssaG ) to confirm that Vi capsule is produced intracellularly. As depicted in Fig. 3, replacement of the natural P tviA promoter with the intra-macrophage-inducible P ssaG promoter resulted in Vi polysaccharide production within macrophages. 10 This is consistent with previous results showing that the ssaG promoter is induced 400-fold in macrophages. 36 Characterization of the S. Typhimurium mutants in vitro All mutant strains were transducible with phage P22 (Supplementary Table S2), indicating that the converted O-polysaccharide was recognized by P22 and that in vitro grown cells did not produce enough Vi to block P22 infection. Susceptibility to deoxycholate did not differ among strains, and differences in sensitivity to polymyxin B among the recombinant strains were increased by two-fold (Supplementary Table S2). All strains had growth rates similar to wild-type strain S100, except strain S1160 (O9, Vi + , ΔP tviA ::P ssaG ΔvexA-E), which showed a moderate decrease in growth rate ( Supplementary Fig. S2). Swimming motility for each mutant was similar for all strains except strain S1160 (O9, Vi + , ΔP tviA ::P ssaG ΔvexA-E), which exhibited a significant decrease in swimming motility compared to wild-type strain S100, though it can still be considered to be motile (Supplementary Table S2).
As successful live attenuated Salmonella vaccines must attach and invade host gastrointestinal epithelial cells, we evaluated the ability of our Δcya Δcrp vaccine strains to interact with Hep-2 cells.

Western blots probed with anti-Vi antisera
Compared to the parental strain S1114 (O4, Δcrp Δcya), all strains attached equally well, except strain S1167, which produces Vi intracellularly. This strain exhibited a significantly lower rate of attachment ( Supplementary Fig. S3). In addition, we observed a significantly lower rate of invasion for all strains carrying viaB.
We next evaluated colonization of host tissues by the Δcya Δcrp derivatives of each strain. Mice were orally inoculated with approximately 1 × 10 9 CFU of each strain. Peyer's patches, spleens, and livers were harvested 4 and 8 days later. No significant differences were detected in colonization of the Peyer's patches among the mutants. While all strains colonized spleen and liver equally well on day 4, by day 8 we recovered significantly fewer CFUs of strains S1148, S1163, S1167, and S1168 as compared to the parental S1114. No deaths occurred during this period ( Supplementary Fig. S4).
recognition by the immune system. There were no significant differences in the IgG responses against the S. Typhimurium OMPs (Fig. 4f). The negative control groups (BSG) in each test mounted no detectable immune responses.
Protective efficacy of live attenuated vaccines against challenge with wild-type S. Typhimurium and S. Enteritidis We investigated whether our live vaccines would stimulate protective immunity against wild-type S. Typhimurium and S. Enteritidis challenge. 37 Since S. Typhi does not effectively colonize nor cause disease in mice, we decided to challenge with a mousevirulent strain of S. Enteritidis. Note that the vaccines converted to serotype O9 induced high levels of anti-O9 antibody immune responses with effective SBA activity against both S. Enteritidis and S. Typhi, which share the same O9 O-antigen polysaccharide. We also challenged a separate group of mice with S. Typhimurium to evaluate the ability of our strains to cross-protect. All the mice survived the 100× LD 50 challenge with S. Typhimurium, indicating that all of our strains provide a high level of homologous protection, even when the strains were converted to express the O9 epitope ( Supplementary Fig. S6A). Mice vaccinated with S1143 (O9, CA − , Δcrp Δcya) and S1163 (O9, Vi + , ΔP tviA ::P ssaG Δcya Δcrp) were completely protected against an oral challenge of 100× LD 50 of wild-type S. Enteritidis. Eighty percent of the mice in groups vaccinated with strains S1163 (O9, Vi + , ΔP tviA ::P ssaG Δcya Δcrp), S1167 (O9, Vi + , ΔP tviA ::P ssaG ΔvexE Δcrp Δcya), and S1168 (O9, Vi + , ΔP tviA ::P ssaG ΔvexA-E Δcrp Δcya) survived challenge, indicating a high level of protection. Mice vaccinated with S1114 (O4, CA − , Δcrp Δcya) and S1148 (O4, Vi + , ΔP tviA ::P ssaG Δcya Δcrp) were not protective, although there was a slight delay in time to death (Supplementary Fig. S6B).

DISCUSSION
The first licensed live vaccine for preventing S. Typhi infection was S. Typhi Ty21a, which was developed to mimic a natural mucosal infection. While this vaccine is well tolerated, it has several drawbacks, including an unclear understanding of the mutations responsible for its attenuation and the requirement of multiple doses for optimal immunogenicity. With the goal of developing live vaccines requiring only a single oral dose, multiple research groups have engineered new live attenuated vaccines to induce serum anti-Vi antibody responses. A series of potential live vaccine candidates (CVD908, CVD908-htrA, CVD909, Ty800, and ZH9) were designed and developed. 7,8,38 However, none of them stimulate high levels of anti-Vi serum antibodies, 7,8,38 although CVD909 has shown promise when used as the priming dose prior to boosting with purified Vi. 7 The production of Vi capsule is highly regulated in vivo. In the human intestinal lumen, a high-osmolarity environment, S. Typhi displays a non-capsulated, flagellated, and invasive phenotype associated with a functional type III secretion system 1 (T3SS-1). Transition through the intestinal epithelium rapidly induces the tviA gene, which upregulates Vi capsule production to protect against complement deposition and phagocytosis while simultaneously masking LPS, a TLR4 agonist, and repressing the biosynthesis of flagella and T3SS-1 effectors. 20,21 Shortly after invasion of the gut epithelium, S. Typhi encounters macrophages and other phagocytic cells in gut-associated lymphoid tissues. The interaction between S. Typhi and macrophages results in downregulation of Vi capsular expression, 35 which is believed to enable evasion of anti-Vi immune responses, and upregulation of type III secretion system 2 biosynthesis, which is essential for S. Typhi survival in macrophages. 39 Because anti-group D LPS antibodies can enhance the protection provided by anti-Vi serum antibodies, 11 we designed a S. Typhimurium vaccine to synthesize both Vi capsule and O9 O-antigen to elicit robust antibody responses against both polysaccharides ( Fig. 1 and Supplementary  Fig. S1). When S. Typhi is transiting through the intestinal epithelium, the exposed O-antigen polysaccharides on the surface are vulnerable to be targeted by anti-O-antigen antibodies. Once S. Typhi invades in intestinal epithelium, the upregulated expression of Vi capsular would be attacked by anti-Vi antibodies.
Janis et al. were the first to generate a live attenuated S. Typhimurium vaccine with in vivo-inducible Vi expression that stimulated high anti-Vi antibodies in the serum after single oral administration. 10 In their report, Hfr conjugation between S. Typhi Ty2 and S. Typhimurium C5 was used to generate a Vi-positive S. Typhimurium (C5.507 Vi + ). In this study, we replaced the entire CA operon with a functional viaB locus and with no extraneous S. Typhi sequences, resulting in S. Typhimurium mutant S1123 (O4, Vi + ,) (Fig. 1a). CA is a surface exopolysaccharide produced by many enteric bacteria, and deletion of the entire CA operon has no influence on virulence but increases the synthesis and exposure of heterologous antigens in Salmonella. 40 Additionally, the original O4 serotype in S. Typhimurium was converted to O9 by replacing the allelic abe gene with prt-tyv D1 from S. Enteritidis (Fig. 1e). While S. Typhi and S. Enteritidis share the same O9 O-antigen epitope, their O-antigens differ in polymer length (Fig. 2a). S. Enteritidis possess a long O-antigen polysaccharide while a short O9 Oantigen polysaccharide has evolved in S. Typhi due to inactivation of the fepE gene, a regulator of long O-antigen polysaccharide synthesis. 32,41 The Vi capsular instead plays this protective role on the surface by covering the short O9 O-antigen polysaccharide in S. Typhi. 32 In vitro analyses indicated that all Vi + strains remained P22 transducible, indicating that Vi production did not block P22 binding under the conditions of the assay (Supplementary Table  S2). This is consistent with our expectation that the P ssaG promoter is not active during growth in Luria-Bertani (LB). Swimming motility was unaffected in all strains except for S1160 (O9, Vi + , ΔP tviA ::P ssaG ΔvexA-E), which exhibited a slight defect, indicating that intracellular accumulation of Vi perturbed flagellar gene regulation, flagellar protein transport and assembly, or both.
Vi-producing strains S1137, S1151, S1159, and S1160 were attenuated approximately 1000-fold (Supplementary Table S2). This decrease in virulence is likely to be related to in vivo synthesis of the heterologous Vi antigen. The attenuated virulence did not impact the immune responses induced by the S1148 (O4, Vi + , ΔP tviA ::P ssaG Δcya Δcrp) and S1163 (O9, Vi + , ΔP tviA ::P ssaG Δcya Δcrp) vaccine strains, which produce surface-exposed Vi. However, immunization with strains S1167 (O9, Vi + , ΔP tviA ::P ssaG ΔvexE Δcrp Δcya) and S1168 (O9, Vi + , ΔP tviA ::P ssaG ΔvexA-E Δcrp Δcya) did not result in anti-Vi IgG in the serum (Fig. 4f), indicating that only surface exposed, lipid-anchored 24 Vi is immunogenic when delivered by live Salmonella. The fact that S1163 (O9, Vi + , ΔP tviA :: P ssaG Δcya Δcrp) could induce strong anti-O9 serum IgG responses was encouraging (Fig. 4d), demonstrating that our vaccine design was adequate to allow presentation of both antigens. One explanation for inducing antibodies against both saccharides is that the O-antigens produced by our construct are comparable to the length of Vi capsular and are longer than native S. Typhi Oantigen.
The primary functions of anti-Vi and anti-O-antigen antibodies are to direct complement deposition on the bacterial cell surface and to promote antibody-mediated phagocytosis by macrophages. 11 While opsonization with anti-Vi antibodies is effective, they would not be expected to impact S. Typhi cells in an environment where Vi production is downregulated. Conversely, anti-O9 antibodies are not effective in the presence of Vi 11 (Fig. 5). Thus we postulated that the synergistic actions of both anti-Vi and anti-O9 antibodies may overcome this innate immune evasion mechanism. We used the SBA assay to investigate this idea (Fig. 5). Indeed, when grown at high osmolarity (Vi-suppressing conditions), S. Typhi was sensitive to mouse sera containing anti-O9 antibodies and no anti-Vi antibodies (O9 + , Vi − ) (Fig. 5a, S1143 group), but was not sensitive to Vi + , O9 − mouse sera (group S1148). Conversely, S. Typhi cells grown at low osmolarity (Viinducing conditions) was sensitive to Vi + , O9 − mouse sera (Fig. 5b, S1148 group) but was not sensitive to Vi − , O9 + sera (S1143 group). The functionality of the anti-O9 and anti Vi-antibodies was confirmed by their ability to direct complement-mediated killing of S. Enteritidis (O9, Vi − ) and S. Paratyphi C (O7, Vi + ), respectively (Fig. 5c, d). Importantly, the sera from mice immunized with strain S1163 (O9 + , Vi + ) contained both anti-O9 and anti-Vi antibodies (Fig. 4b, e) and directed efficient complement-mediated killing of S. Typhi cells grown at either low-or high-osmolarity (Fig. 5a, b). This indicates that the strategy used to construct strain S1163 resulted in antibodies capable of attacking S. Typhi regardless of its Vi status. The protective efficacy of the anti-O9 response was further supported by our challenge results showing that immunization with S1163 resulted in complete protection against S. Enteritidis (Supplementary Fig. S4B).
Most live typhoid vaccine strains (e.g., S. Ty21a, 42 Ty800, 13 M01ZH09, 43 CVD908 44 ) and Salmonella strains designed to deliver heterologous antigens to combat human disease, such as χ9633 and χ9640, 45 were derived from S. Typhi, which is human hostrestricted and able to colonize humans systemically, including tissues such as liver, spleen, and bone marrow. Since wild-type S. Typhimurium does not penetrate to the deeper tissues, typically invading only local intestinal epithelial tissue, attenuated S. Typhimurium strains are often not considered for human use. However, there have been two clinical trials in which they were evaluated. Human subjects orally immunized with attenuated S. Typhimurium strains LH1160 46 and WT05 47 developed strong mucosal or serological responses. Most S. Typhi vaccines carry deletion mutations associated with systemic dissemination, such as phoPQ, htrA, and ssaV, with no bacteremia observed during vaccination, indicating that the protection would theoretically depend largely on local immune induction sites. 9 Therefore, we propose that a properly attenuated S. Typhimurium strain could Anti-Vi and O9 antibodies synergistically prevent typhoid P Li et al.
also be used as a live typhoid vaccine and as a vaccine vector for human use.
We envision that the immune responses elicited by orally administered S. Typhimurium vaccine S1163 (O9, Vi + , ΔP tviA ::P ssaG Δcya Δcrp) will provide protection as follows. When S. Typhi arrives at intestinal mucosal surfaces, it displays a non-capsulated, flagellated, and invasive phenotype susceptible to secreted anti-O9 IgA and IgM (Fig. 6, Phase I). S. Typhi cells that survive this first line of defense transit through the intestinal barrier. Vi capsule polysaccharides are rapidly produced and the flagellar synthesis is repressed. Vi-encapsulated S. Typhi will then encounter macrophages, dendritic cells, neutrophils, and complement systems. At this stage, vaccine-induced anti-Vi antibody (IgG) will enhance the bactericidal activities of these host systems (Phase II). Some S. Typhi cells will be captured by antigen-presenting cells (macrophages or dendritic cells) to interact with memory T cells. Vi capsule is shed within macrophages, leaving S. Typhi cells uncapsulated, allowing efficient processing to facilitate activation of B cells and T cells. More and efficient anti-Vi and anti-O9 antibodies (IgA, IgG) produced by activated B cell and activated T cell (cytotoxic T cells or natural killing cells) will aid in eliminating the infected cells and clearing systemic infection (Phase III).
Our work highlights the possibility of developing live attenuated S. Typhimurium vaccines to prevent typhoid fever by targeting the protective Vi capsular and O9 O-polysaccharide antigens simultaneously. This strategy was also effective at providing strong protection against S. Typhimurium challenge ( Supplementary Fig. S4). This work provides a mechanism for developing multi-valent live attenuated vaccine against group B and group D Salmonella infections.

MATERIALS AND METHODS
Additional materials and methods could be found in Supplementary Text S1. Bacteria, plasmids, and culture conditions Bacteria and plasmids used in this study are listed in Table 1. Escherichia coli and Salmonella strains were aerobically grown at 37°C in LB broth or on LB agar. To induce maximum Vi capsular production, wild-type S. Typhi was grown aerobically in low-osmolarity broth (10 g/l tryptone, 5 g/l yeast extract, with no added NaCl) to mid-log phase at 37°C. 17 To inhibit Vi capsular synthesis, wild-type S. Typhi were grown in high-osmolarity LB broth (approximately 500 mM NaCl). When necessary, chloramphenicol was added at 25 µg/ml for selection of transconjugants.

Molecular and genetic procedures
Suicide vectors and primers used in this study are listed in Table 1 and  Supplemental Table S1, respectively. DNA fragments were assembled using Gibson Assembly Master Mix according to the manufacturer's instructions (New England BioLabs). sacB gene-based sucrose counterselectable suicide vectors were used to construct unmarked deletion and/ or insertion mutations in S. Typhimurium. 48 Specifically, for deletion mutations, two homologous DNA fragments, the upstream and downstream regions of the gene or operon being deleted, were amplified. After purification, these two fragments were fused by PCR and cloned into the pYA4278 suicide vector. The conjugational transfer of recombinant suicide vectors to S. Typhimurium was performed using the suicide vector donor strain χ7213. Transconjugants were selected on chloramphenicol agar without supplemental DAP. The second homologous recombination event, resulting in excision of the suicide vector from a S. Typhimurium chromosome, was selected on 10% sucrose LB plates without sodium chloride and grown at 30°C. Successful gene deletion mutations were confirmed by PCR screening and DNA sequencing. For the insertion mutations, the genes or operon being inserted and the suicide vector backbone containing the directed insertion site were amplified. After purification, these two linear DNA fragments were assembled sequentially, resulting in a new circular suicide vector with new genes or operons replacing those previously deleted. The subsequent insertion mutation processes were the same as those described above for the deletion mutations. The deletion and insertion mutations constructed for this study are illustrated in Fig. 1. LPS silver staining and western blotting LPS silver staining were prepared, separated, and visualized using the method provided by Hitchcock and Brown. 49 For western blotting, anti-Oantigen single-factor rabbit antisera (BD Biosciences) or anti-Vi polymer rabbit antisera (BD Biosciences) were used to probe the blots in polyvinylidene difluoride membranes. Twenty-four 6-week-old female BALB/c mice per group were vaccinated orally on day 0 with 20 μl BSG containing approximately 1 × 10 9 CFU vaccine strains and boosted on day 28 with the same dose. Blood samples and vaginal secretions were collected from randomly selected 12 mice in each group on day 56 after the booster immunization, and the mice in each group were randomly challenged orally on day 63 with 5 × 10 7 CFU of S. Typhimurium or S. Enteritidis (~100 times the LD 50 ). 37 S. Typhimurium and S. Enteritidis LPS (Sigma, St. Louis, MO, USA) were used to measure immune responses. IgG and IgA antibodies specific to S. Typhimurium and S. Enteritidis LPS in the serum or vaginal secretions were measured using the quantitative enzyme-linked immunosorbent assay (ELISA) as described previously. 50 Antibody concentrations were calculated based on absorbance values and the standard curve. Quantitative measurement of the mouse IgG antibody against the S. Typhi Vi polysaccharide in the serum was achieved using the Typhoid Vi IgG ELISA Kit (cat. #990-520-MTG, Alpha Diagnostic Intl. Inc. San Antonio, TX, USA) according to the manufacturer's instructions. IgG concentrations (U/ml) specific to Vi were calculated based on calibrator value graphs.
Serum bactericidal activity assay SBA was performed as previously described with a few modifications, 51 S. Typhi, S. Paratyphi C, and S. Enteritidis were grown to an OD 600 of 0.6 in either low-(0 ml NaCl), normal-(170 mM NaCl) or high-osmolarity (500 mM NaCl) LB broth as indicated. After centrifugation and resuspension, the logphase cultures were diluted in PBS to a concentration of approximately 1 × 10 4 CFU/ml. Pooled vaccinated and non-vaccinated mouse sera were heatinactivated at 56°C for 20 min, and two-fold serial dilutions in PBS starting from 1:200 to 1:25,600 were made in 96-well plates. Optimal SBA results were achieved by combining 25 µl of active baby rabbit complement (25% final concentration) with 15 µl of PBS, 50 µl of diluted mice pooled sera, and 10 µl of diluted bacteria (~350 CFU). In total, 10 µl of the mixture from each well was spread on LB agar plates after 60 min to assess the bactericidal activity. The spread LB agar plates were incubated overnight at 37°C, and the viable CFUs were counted the next day. The negative control contained only bacteria and complement, and the bactericidal activity was determined as the percentage of CFU counts in each pooled sera dilution compared to the CFU counts of the negative control. The SBA graphs depict the percentage of bacterial growth as a function of the sera antibodies in each diluted pool.

Statistical analysis
Data were analyzed using the GraphPad Prism 5 software (Graph Software, San Diego, CA) by one-way or two-way analysis of variance 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 means ± SEM. P < 0.05 was considered statistically significant.