Effect of Deletion of Genes Involved in Lipopolysaccharide Core and O-Antigen Synthesis on Virulence and Immunogenicity of Salmonella enterica Serovar Typhimurium

Abstract

Lipopolysaccharide (LPS) is a major virulence factor of Salmonella enterica serovar Typhimurium and is composed of lipid A, core oligosaccharide (C-OS), and O-antigen polysaccharide (O-PS). While the functions of the gene products involved in synthesis of core and O-antigen have been elucidated, the effect of removing O-antigen and core sugars on the virulence and immunogenicity of Salmonella enterica serovar Typhimurium has not been systematically studied. We introduced nonpolar, defined deletion mutations in waaG (rfaG), waaI (rfaI), rfaH, waaJ (rfaJ), wbaP (rfbP), waaL (rfaL), or wzy (rfc) into wild-type S. Typhimurium. The LPS structure was confirmed, and a number of in vitro and in vivo properties of each mutant were analyzed. All mutants were significantly attenuated compared to the wild-type parent when administered orally to BALB/c mice and were less invasive in host tissues. Strains with ΔwaaG and ΔwaaI mutations, in particular, were deficient in colonization of Peyer's patches and liver. This deficiency could be partially overcome in the ΔwaaI mutant when it was administered intranasally. In the context of an attenuated vaccine strain delivering the pneumococcal antigen PspA, all of the mutations tested resulted in reduced immune responses against PspA and Salmonella antigens. Our results indicate that nonreversible truncation of the outer core is not a viable option for developing a live oral Salmonella vaccine, while a wzy mutant that retains one O-antigen unit is adequate for stimulating the optimal protective immunity to homologous or heterologous antigens by oral, intranasal, or intraperitoneal routes of administration.

INTRODUCTION

Live attenuated Salmonella vaccines can colonize the gut-associated lymphoid tissue (e.g., Peyer's patches [PP]) and the secondary lymphatic tissues to elicit immune responses against both Salmonella and the heterologous antigens delivered by the vector during colonization of the host (9, 10, 12). Attenuation of Salmonella virulence can be achieved either by deleting the genes involved in central metabolic pathways such as aroA, aroC, purA, pabA, and pabB or by deletion or regulated expression of global regulators such as phoPQ, fur, and crp and other genes important for establishing an infection (reviewed in references 8, 10, 12, and 14).

Lipopolysaccharide (LPS) of Salmonella is a recognized virulence determinant and is essential for functions, including swarming motility (60), intestinal colonization (47), serum resistance (57), invasion/intracellular replication (43), and resistance to killing by macrophages, all of which are critical for successful infection by the pathogen. LPS consists of three different regions: a conserved lipid A, a short core oligosaccharide (C-OS), and the O-antigen polysaccharide (O-PS) assembled in a variable number of oligosaccharide repeating units (51). Lipid A is a ligand that stimulates the TLR4-MD2-CD14 pathway to activate the antimicrobial host defense during bacterial infection (41, 50). The O-PS is Salmonella's primary defense against serum complement activation (42). Truncating the LPS O-antigen and ceasing its synthesis in vivo are strategies for attenuating Salmonella for use as live vaccines.

Galactose is a component of both core and O-antigen, and mannose is a component of O-antigen. The galE and pmi genes encode enzymes that are critical for UDP-galactose and GDP-mannose synthesis, respectively. Thus, galE (19, 29) and pmi (13, 37, 58) mutants of Salmonella enterica serovar Typhimurium are conditionally rough (lacking O-antigen), depending on the availability of galactose or mannose, respectively, resulting in strains that are attenuated and immunogenic. For use as vaccines, these strains are grown with their respective sugars prior to immunization and produce complete O-antigen. RfaH, a transcriptional antiterminator, is required for the synthesis of secreted and surface-associated cell components of S. Typhimurium, including O-PS and core sugar components of LPS (2). An rfaH deletion mutant is described as “gently rough,” and exhibits deep-rough characteristics and stimulates protective immune response against Salmonella challenge in BALB/c mice (13, 28). The wzy (rfc) gene encodes the O-antigen ligase, required for polymerization of multiunit O-antigen molecules. A wzy (rfc) mutant produces semirough LPS, consisting of a complete core and a single O-antigen subunit. We have previously shown that a ΔpabA ΔpabB vaccine strain carrying a Δwzy mutation is immunogenic and can effectively deliver the heterologous Streptococcus pneumoniae antigen PspA, resulting in protective immunity against Streptococcus pneumoniae challenge (34). The immunogenicity of rfaH and wzy mutant strains can be further enhanced by programmed downregulation of rfaH or wzy expression in a recombinant attenuated Salmonella vaccine (RASV) without compromising their ability to colonize lymphoid tissues (34, 35). These previous studies led us to consider manipulation of other genes involved in the synthesis of LPS, including the genes for core oligosaccharide (C-OS) that have not been explored previously in the context of vaccine construction.

C-OS is assembled by the sequential transfer of sugars from nucleotide mono-sugar or diphospho-sugar donors to a lipid A acceptor by specific glycosyltransferases in the inner membrane (17, 63). The genes encoding these proteins are organized in the waa operon (rfa operon) (53). The potential of Salmonella mutants with truncated core and O-antigen as attenuated vaccines has not been systematically analyzed. In this work, we constructed strains with an ordered series of S. Typhimurium nonpolar deletion mutations causing stepwise truncations of core sugars and evaluated these mutant strains for virulence and ability to colonize BALB/c mice. The effect of each mutation in a well-characterized attenuated vaccine strain on its ability to deliver the Streptococcus pneumoniae antigen PspA and to affect an antigen-specific protective immune response was also determined.

MATERIALS AND METHODS

Bacterial strains, plasmids, media, and growth conditions.

The strains and plasmids used in this study are listed in Table 1. S. Typhimurium cultures were grown at 37°C in LB broth (4) or nutrient broth (Difco) or on LB agar with or without 0.1% arabinose. Selenite broth, with or without supplements, was used for enrichment of Salmonella from mouse tissues. Diaminopimelic acid (DAP) (50 μg/ml) was added for the growth of Δasd strains (45). LB agar containing 5% sucrose was used for sacB gene-based counterselection in allelic exchange experiments. MOPS (morpholinepropanesulfonic acid) minimal medium (46) with/without 10 μg/ml p-aminobenzoic acid (pABA) was used to confirm the phenotype of ΔpabA ΔpabB mutations.

Table 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Descriptiona	Source
Suicide vectors
pRE112	sacB mobRP4 R6K ori Cm+	15
pYA4278	sacB mobRP4 R6K ori Cm+, pRE112-T-vector	Lab collection
pYA4896	ΔwaaG42	pRE112
pYA4897	ΔwaaI43	pRE112
pYA4718	ΔrfaH49	35
pYA4898	ΔwaaJ44	pRE112
pYA4899	ΔwbaP45	pRE112
pYA4900	ΔwaaL46	pRE112
pYA4717	Δwzy-48	34
Recombinant plasmids
pYA3493	Plasmid Asd+; pBR ori β-lactamase signal sequence-based periplasmic secretion plasmid (blaSS)	32
pYA4088	852-bp DNA encoding the α-helical region of PspA (aa 3-285) into pYA3493; blaSS-pspA	65
S. Typhimurium
χ3761	UK-1	24
χ11308	ΔwaaG42	In χ3761
χ11309	ΔwaaI43	In χ3761
χ9945	ΔrfaH49	35
χ11310	ΔwaaJ44	In χ3761
χ11311	ΔwbaP45	In χ3761
χ11312	ΔwaaL46	In χ3761
χ9944	Δwzy-48	In χ3761
χ9241	ΔpabA1516 ΔpabB232 ΔasdA16 ΔaraBAD23 ΔrelA198::araC PBADlacI TT	65
χ11313	ΔpabA1516 ΔpabB232 ΔasdA16 ΔaraBAD23 ΔrelA198::araC PBADlacI TT ΔwaaG42	Derivation of χ9241
χ11314	ΔpabA1516 ΔpabB232 ΔasdA16 ΔaraBAD23 ΔrelA198::araC PBADlacI TT ΔwaaI43	Derivative of χ9241
χ9884	ΔpabA1516 ΔpabB232 ΔasdA16 ΔaraBAD23 ΔrelA198::araC PBADlacI TT ΔrfaH49	35
χ11315	ΔpabA1516 ΔpabB232 ΔasdA16 ΔaraBAD23 ΔrelA198::araC PBADlacI TT ΔwaaJ44	Derivation of χ9241
χ11316	ΔpabA1516 ΔpabB232 ΔasdA16 ΔaraBAD23 ΔrelA198::araC PBADlacI TT ΔwbaP45	Derivation of χ9241
χ11317	ΔpabA1516 ΔpabB232 ΔasdA16 ΔaraBAD23 ΔrelA198::araC PBADlacI TT ΔwaaL46	Derivation of χ9241
χ9885	ΔpabA1516 ΔpabB232 ΔasdA16 ΔaraBAD23 ΔrelA198::araC PBADlacI TT Δwzy-48	Derivation of χ9241
E. coli K-12
χ7232	endA1 hsdR17 (rK− mK+) glnV44 thi-1 recA1 gyrA relA1 Δ(lacZYA-argF)U169 λpir deoR (φ80dlac Δ(lacZ)M15)	54
χ7213	thi-1 thr-1 leuB6 glnV44 fhuA21 lacY1 recA1 RP4-2-Tc::Mu λpir ΔasdA4 Δzhf-2::Tn10	54
χ7122	E. coli O78	Lab collection
χ3550	S. Enteritidis	Lab collection
S. pneumoniae WU2	Wild-type virulent, encapsulated type 3	7

^aThe mutations introduced into χ9241 background are indicated in bold.

Plasmids and mutant strain construction.

The primers used in this study are listed in Table 2. DNA manipulations were carried out as described elsewhere (55). Transformation of Escherichia coli and S. Typhimurium was done by electroporation. Transformants were selected on LB agar plates containing appropriate antibiotics. Selection for Asd⁺ plasmids was done on LB agar plates.

Table 2.

Primers used in this work

Primer name	Sequence 5′–3′
DrfaG-1F	GGATGCTATTGCACGCGGCTG
DrfaG-1R	TCACCTGCCTGCAGGCTCATAGCAGGCTGTCCAAA
DrfaG-2F	CTATGAGCCTGCAGGCAGGTGATTTAGATGGTTGAGC
DrfaG-2R	TGATGTTCTGGTTAGCGGGT
DrfaI-1F	GTATGGTTGGCAAAGCGCGC
DrfaI-1R	TATGGGTTCCTGCAGGAGTGATCACTTTTGTAATTTC
DrfaI-2F	ATCACTCCTGCAGGAACCCATAGGTGATGTAATG
DrfaI-2R	TTAGTTTGAGGCAATACCTC
DrfaJ-1F	AGTGAATAAAGCGCGTTTTG
DrfaJ-1R	GAGGGGAACCTGCAGGTACATCACCTATGGGTTTTAT
DrfaJ-2F	GATGTACCTGCAGGTTCCCCTCCCTAAATCTCCA
DrfaJ-2R	AGCGTAATGTTTTAATTTCC
DrfbP1F	CAACTGATAAAAGTCAATCC
DrfbP1R	GTAAGCTTACCTGCAGGTTAATCCTCACCCTCTGA
DrfbP2F	GATTAACCTGCAGGTAAGCTTACCGAGAAGTACTGA
DrfbP2R	ATACGACGAGGCGTTTCGAG
DrfaL-1F	GTGGGCAACACCGGATTACGG
DrfaL-1R	GCGTTTTTTTACTTTTCTCCACAATAGGTTTGG
DrfaL-2F	TGGAGAAAAGTAAAAAAACGCGCTGATACTTAT
DrfaL-2R	TTCTATCATATAGCCGATCTG

For the waaG42 deletion, two pairs of primers, DrfaG-1F/DrfaG-1R and DrfaG-2F/DrfaG-2R, were used to amplify approximately 350-bp fragments upstream and downstream of gene rfaG, respectively, from the χ3761 genome. The two fragments were then joined by PCR using primers RfaG-1F and RfaG-2R. The terminal A was added at both ends to the resulting PCR product by using the Gotaq enzyme (Promega). To facilitate convenient insertion of this PCR product, we constructed a T-cloning suicide vector, pYA4278, by inserting a blunt-end P_trc green fluorescent protein (GFP) cassette flanked by AhdI sites (66) into SmaI and XbaI sites of pRE112. The PCR product was ligated to pYA4278 to generate plasmid pYA4896, which carries a deletion of the entire waaG gene. A similar strategy was used to construct plasmids pYA4897 (ΔwaaI43), pYA4898 (ΔwaaJ44), pYA4899 (ΔwbaP45), and pYA4900 (ΔwaaL46) (Tables 1 and 2). The mutations were independently introduced into S. Typhimurium χ3761 by allelic exchange using suicide vectors pYA4896, pYA4897, pYA4898, pYA4899, and pYA4900 to generate χ11308, χ11309, χ11310, χ11311, and χ11312, respectively. The mutations were also introduced into S. Typhimurium vaccine strain χ9241 to yield strains χ11313, χ11314, χ11315, χ11316, and χ11317, respectively. χ9944 and χ9885 were generated by transforming pYA4717 (Δwzy-48) into χ3761 and χ9241, respectively. The presence of both pabA1516 and pabB232 mutations in Salmonella strain χ9241 and its derivatives was verified by the inability of the strains to grow in MOPS minimal medium without p-aminobenzoate. The genotype of the ΔasdA16 strain was confirmed by PCR, and the phenotype was confirmed by the inability of the strain to grow in media without DAP. The ΔaraBAD23 mutation was verified by PCR and by its white colony phenotype on MacConkey agar supplemented with 1% arabinose. LPS profiles of Salmonella strains were examined by the method of Hitchcock and Brown (25) using cultures standardized based on the optical density at 600 nm (OD₆₀₀).

MIC test.

MICs of different antimicrobial substances were determined (64). Twofold serial dilutions of the bile salt deoxycholate (0.3 to 40 mg/ml), polymyxin B (0.075 to 4.7 μg/ml, Sigma), SDS (0.35 to 20 mg/ml, Sigma), and ox bile (1 to 20 mg/ml, Sigma) were made across 96-well plates. Bacteria were grown to an OD₆₀₀ of 0.8 to 0.9 in nutrient broth with or without 0.1% arabinose, harvested, and washed in phosphate-buffered saline (PBS). Cells were diluted to 1 × 10⁵ to 5 × 10⁵ CFU in nutrient broth with or without arabinose, and 0.1 ml of the diluted cell suspension was added to each well containing the antimicrobial substance. The microtiter plates were incubated overnight at 37°C. The optical density of each culture was determined using a model SpectraMax M2e (Molecular Devices, CA) plate reader. The threshold of inhibition was an OD₆₀₀ of 0.1. Actual titers were determined by spreading culture dilutions onto LB plates followed by overnight incubation at 37°C. Assays were repeated three times.

Motility assay.

Swimming motility was assessed on LB plates solidified with 0.3% agar and supplemented with 0.5% glucose. The plates were allowed to dry at room temperature for 2 h. Freshly grown bacteria were collected from LB agar plates without arabinose, washed, and diluted in saline. Six microliters of bacterial suspension (approximately 1 × 10⁶ CFU) was spotted onto the middle of the plates and incubated at 37°C for 6 h. The diameter of the colonies was measured. Experiments were repeated three times.

Attachment and invasion assays.

Attachment and invasion tests were performed as described previously (18). The human intestinal cell line INT407 was cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal calf serum, 2 mM l-glutamine, and 100 μg/ml penicillin and streptomycin. Cells were seeded 16 h prior to infection at a density of 5 × 10⁵ cell/well (24-well plates). Bacteria were grown overnight statically in 3 ml LB. One ml of the overnight culture was added to 50 ml LB and grown at 37°C to an OD₆₀₀ of 0.95. The culture was diluted to 2 × 10⁷ CFU/ml in DMEM without antibiotics. The tissue culture medium was replaced by DMEM supplemented with 10% fetal calf serum and 2 mM l-glutamine immediately before 100 μl of the bacterial suspension was added to each well to achieve a multiplicity of infection (MOI) of 5:1. The plates were centrifuged at 200 × g for 5 min to synchronize the interaction between bacteria and monolayer cells and incubated at 37°C in a CO₂ (5%) incubator to allow attachment for 1 h. Wells were washed 3 times with PBS, followed by the addition of 1% Triton X-100 to lyse the INT407 cells. Aliquots of serial dilutions made from each well were spread onto LB plates and incubated overnight to determine the number of attached bacteria. Attachment was calculated as follows: percent attachment = 100 × (number of cell-associated bacteria/initial number of bacteria added). The same procedure was used to evaluate invasion with the following modification. Prior to the lysis step, 1 ml of DMEM containing 100 μg/ml gentamicin to kill extracellular bacteria was added and the monolayer was incubated for an additional 1 h at 37°C. Invasion was calculated as follows: percent invasion = 100 × (number of bacteria resistant to gentamicin/initial number of bacteria added). Both adherence and invasion assays were performed three times.

Serum bactericidal assay.

Pooled human serum (PHS) was purchased from Equitech-Bio and stored at −70°C until used. Heat-inactivated serum (HIS) was generated by incubating PHS for 2 h at 56°C. Bacteria grew statically in LB media overnight, and the next day they were grown to an OD₆₀₀ of 0.9 to 0.95 under 37°C at 180 rpm. Bacteria (1 ml) were resuspended in Dulbecco's phosphate-buffered saline (DPBS) to a final concentration of 2 × 10⁸ to 3 × 10⁸ CFU/ml; 50-μl portions of the bacterial suspensions were added to 50 μl PHS or HIS and incubated for 30 min at 37°C. After incubation, the samples were placed on ice to stop further bacteriolysis. Serial dilutions of the samples in PBS were plated on LB agar plates and incubated overnight at 37°C. Bacterial colonies were counted to determine survival.

Determinations of LD₅₀ and colonization in mice.

All animal procedures were approved by the Arizona State University Animal Care and Use Committee. Seven-week-old, female BALB/c mice were obtained from the Charles River Laboratories (Wilmington, MA). Mice were acclimated for 7 days after arrival before starting the experiments. Determination of the oral 50% lethal dose (LD₅₀) followed our standard procedures (35).

To evaluate the effect of the administration route on colonization, mice were inoculated orally with 20 μl of buffered saline with gelatin (BSG) containing 1 × 10⁹ CFU or intranasally (i.n.) with 10 μl of BSG containing 1 × 10⁹ CFU. Six days after oral inoculation, or 6 days for mutants and 3 days for wild-type after i.n. inoculation, three animals per group were euthanized, Peyer's patches (PP) were collected from all sites of the small intestinal surface, and bacterial colonization in PP represented the mean bacterial burden in pooled PP from each mouse. Spleen and liver samples were collected, and their individual weights were measured. Each sample was homogenized in a total volume of 1 ml BSG, and appropriate dilutions were plated onto MacConkey agar and/or LB agar to determine the number of viable bacteria. The remaining tissue homogenate was enriched for Salmonella by incubation in selenite cysteine broth overnight at 37°C. Samples that were negative by direct plating and positive by enrichment in selenite cysteine broth were recorded as 1 CFU/organ.

Construction of vaccine strain derivatives.

Each of the mutations under study was introduced into the previously described attenuated Salmonella strain χ9241 (35, 65). Expression plasmid pYA4088, which specifies a Bla-SS-PspA secreted fusion protein, was introduced into each of the χ9241 derivatives. The level of Bla-SS-PspA fusion protein produced by each strain was examined by using Western blots probed with anti-PspA antibody.

Immunogenicity of vaccine strains in mice.

Strains were grown statically overnight in LB broth with 0.1% arabinose at 37°C. Arabinose was added to reduce antigen expression during in vitro growth (61). The following day, 2 ml of the overnight culture was inoculated into 100 ml of LB broth with 0.1% arabinose and grown with aeration at 37°C to an OD₆₀₀ of 0.8 to 0.9. Cells were harvested by room temperature centrifugation at 4,000 rpm for 15 min, and the pellet was resuspended in 1 ml of BSG. Groups of mice (6 or 7 mice per group) were orally inoculated with 20 μl of BSG containing 1 × 10⁹ CFU of each strain, or mice were i.n. or intraperitoneally (i.p.) inoculated with 10 μl or 100 μl of BSG containing 1 × 10⁷ or 1 × 10⁶ of each strain, respectively, on day 0 and boosted on day 35 with the same dose of the same strain. Blood was obtained every 4 weeks by mandibular vein puncture. Blood was allowed to coagulate at 37°C for 2 h. Following centrifugation, the serum was removed and stored at −80°C.

Antigen preparation and ELISA.

Recombinant PspA (rPspA) was purified as described previously (32). The rPspA clone, which encodes the α-helical region of PspA (amino acids [aa] 1 to 302) in pET20b, was a kind gift from Susan Hollingshead at the University of Alabama at Birmingham. S. Typhimurium LPS was purchased from Sigma. Outer membrane proteins were prepared as described previously (32). An enzyme-linked immunosorbent assay (ELISA) was used to assay serum antibodies against S. Typhimurium LPS, rPspA, and bacterial outer membrane proteins, including those from Salmonella (SOMPs), as previously described (38). Color development (absorbance) was recorded at 405 nm using a SpectraMax M2e automated ELISA plate reader (Molecular Devices, Menlo Park, CA). A₄₀₅ readings 0.1 higher than PBS control values were considered positive (serum dilution start at 1:50).

Statistical analyses.

Statistical analyses were performed by using the GraphPad Prism 5 software package (Graph Software, San Diego, CA). Antibody titers were expressed as means ± standard deviations. The means were evaluated with one or two-way analysis of variance (ANOVA) and Bonferroni's multiple comparison test for comparisons among groups of antibody titers.

RESULTS

Mutant construction and LPS phenotypes.

We constructed an isogenic set of S. Typhimurium strains with nonpolar gene deletions that resulted in progressively shorter LPS, beginning with a Δwzy mutant whose LPS includes a single O-antigen unit, through a ΔwaaG mutant with a core truncated to the heptose residues (Fig. 1A). Each deletion mutation resulted in a strain with the expected length of LPS (Fig. 1B). The wild-type parent strain, UK-1, produced a complete or smooth LPS (lane UK-1 in Fig. 1B). The LPS produced by the ΔwaaG42 mutant included only lipid A and the inner core (Fig. 1A) and migrated the fastest in the gel as expected (Fig. 1B). The LPS produced by the other mutants, as ordered from left to right in Fig. 1B, included more core and/or O-antigen sugars, resulting in the expected reduced mobility in the gel. The ΔrfaH mutant produced a truncated core with a molecular mass similar to that of the ΔwaaI mutant, as previously reported (35, 39).

Fig. 1.

Schematic representation of S. Typhimurium lipopolysaccharide molecule (inner core, outer core, and O-antigen) and LPS phenotypes of mutant and parent strains. (A) Dotted lines indicate the level of LPS truncation resulting from each mutation (Kdo, 3-deoxy-d-manno-octulosonic acid; PPEtN, pyrophosphorylethanolamine; Hep, Heptose; GlcNAc, N-acetylglucosamine; Glc, glucose; Gal, Galactose; P, phosphate). (B) Cells from indicated strains were grown and processed for LPS analysis as described in Materials and Methods. LPS was visualized by silver staining polyacrylamide gel electrophoresis (PAGE) gels. The mutant strains (from left to right): χ11308 (ΔwaaG42), χ11309 (ΔwaaI43), χ9945 (ΔrfaH49), χ11310 (ΔwaaJ44), χ11312 (ΔwaaL46), χ11311 (ΔwbaP45), and χ9944 (Δwzy-48). The expected location of O-antigen regions and core is indicated on the right of the gel.

The ΔwaaL46 and ΔwbaP45 mutants synthesize complete core with no O-antigen attached. The ΔwaaL46 mutant produces an undecaprenyl-phosphate-O-unit (Und-PP-O-unit) through the action of WbaP (49), but it cannot be added to the lipid A core due to the absence of O-antigen ligase (WaaL) (5, 33). Alternatively, the ΔwbaP mutant synthesizes O-antigen ligase (WaaL) but lacks WbaP, the Und-P galactose phosphotransferase that produces an O-antigen ligase substrate by catalyzing the transfer of galactosyl-1-phosphate from UDP-galactose to Und-P. The Δwzy-48 mutant displayed the expected semirough phenotype with the addition of one O-antigen unit (Fig. 1B) (34).

To confirm the observed alterations in LPS profiles, we performed infection studies with the O-antigen-specific phage P22. Strains grown in LB broth were used as recipients for transduction assays. As expected, all mutants except the ΔrfaH49 and wild-type UK-1 strains were resistant to P22 infection (Table 3). The ΔrfaH49 strain was sensitive to P22 and was transducible; however, the number of transductants obtained was much lower than that obtained with the wild-type strain, as we observed previously (35), indicating that, although undetectable on our silver-stained gel (Fig. 1B), the ΔrfaH49 mutant still produced minute amounts of the full-length O-antigen.

Table 3.

MICs of antimicrobials, swimming motility, transduction efficiency, and virulence of S. Typhimurium strain χ3761 and its isogenic mutants

Strain	P22 phage sensitivitya	MIC (DOCb bile) (mg/ml)	MIC (polymyxin B) (μg/ml)	MIC (ox bile) (mg/ml)	Motility (mm) on soft agarc	LD50d
χ11308 (ΔwaaG42)	−	1.3	0.15	>20	3.5***	>109
χ11309 (ΔwaaI43)	−	2.5	0.08	>20	4***	>109
χ9945 (ΔrfaH49)	+	2.5	0.08	>20	17**	>109
χ11310 (ΔwaaJ44)	−	2.5	0.08	>20	15**	>109
χ11312 (ΔwaaL46)	−	2.5	0.3	>20	20**	>109
χ11311 (ΔwbaP45)	−	2.5	0.3	>20	7***	>109
χ9944 (Δwzy-48)	−	2.5	0.3	>20	45	>109e
χ3761	+++	10	0.6	>20	45	1 × 104

^a−, resistant; + and +++, degrees of sensitivity as determined by transduction efficiency.

^bDOC, deoxycholate.

^c**, P < 0.01 compared with wild-type UK-1; ***, P < 0.001 compared with wild-type UK-1.

^dLD₅₀, median lethal dose.

^eSome mice displayed symptoms (scruffy coat, lethargy) after inoculation with the Δwzy-48 strain but recovered after 14 days.

Phenotypic evaluation of mutant strains.

The mutants' growth rates were evaluated in LB broth. All strains were similar to the UK-1 parent except strains χ11308 (ΔwaaG42) and χ11309 (ΔwaaI43), which had a slightly lower rate of growth. Under similar conditions, it took these mutants 5 h to reach an OD₆₀₀ of 1.3, compared to 4.5 h for the parent strain UK-1 (data not shown). We also noted that these two mutants tended to settle out of suspension more quickly than the other strains when held statically for 30 min.

The effect of LPS alterations on the resistance of these mutants to detergents, such as deoxycholate (DOC) and ox bile, and to the antimicrobial peptide polymyxin B was determined by measuring the MIC in each case (Table 3). All mutants were more sensitive to DOC and polymyxin B than wild-type UK-1, and the sensitivity varied based on the degree of truncation of the outer core. The ΔwaaG42 mutant, lacking the entire outer core, was the most sensitive to DOC, with an MIC approximately 6-fold lower than that of the wild-type parent strain. These results are in agreement with our previous report that a ΔrfaH mutant is more sensitive to DOC than the parent strain (35). However, the strain with the ΔwaaG42 mutation exhibited at least 2-fold greater resistance to polymyxin B than the ΔwaaI43, ΔrfaH49 and ΔwaaJ44 mutants, each of which contains more of its outer core. Strains carrying the ΔwaaL46, ΔwbaP45, and Δwzy-48 mutations contain all the core sugars and displayed intermediate sensitivity between the smooth parent and the other mutants (Table 3). All strains were resistant to the highest concentrations of ox bile tested.

Rough mutants are unable to swarm because of insufficient moisture on the agar surface but can still swim in 0.3% soft agar (60). Therefore, we determined the relative ability of each mutant to swim on soft agar (Table 3). All mutants were able to swim to various degrees on soft agar. Three mutants, ΔwaaG42, ΔwbaP45, and ΔwaaI43, each had substantially reduced bacterial motility compared to other mutants (Table 3), whereas ΔrfaH49, ΔwaaJ44, and ΔwaaL46 strains showed 2-fold reductions in motility as determined by the motility diameter on soft agar. Deletion of wzy-48 had no effect on swimming motility, which was similar to that measured for wild-type UK-1 (34).

Attachment/invasion and serum sensitivity assays.

To examine the effect of these LPS gene mutations on attachment and invasion, we evaluated the ability of the mutants to attach to and invade INT407 human epithelial cells. All strains were similar to UK-1 in their ability to attach to INT407 cells, indicating that the outer core and O-antigen are not essential for this process (data not shown). In contrast, the ΔwaaI43 and ΔwaaG42 mutations resulted in a significant reduction in invasion compared to the other strains, indicating a requirement for the inner glucose and the galactose residues of core for this activity (Fig. 2A). Deletion of wzy resulted in a small (2-fold) but significant increase in invasion compared to that of wild-type UK-1 (P < 0.05), indicating that shortening the O-antigen contributes to bacterial invasion into epithelial cell. A similar phenomenon has been observed in Shigella, in which shortening the LPS by glucosylation promotes bacterial invasion and evasion of innate immunity by enhancing TTSS function without compromising the protective properties of LPS (62).

Fig. 2.

INT407 epithelial cell invasion capability and serum resistance of S. Typhimurium wild-type UK-1 and its isogenic mutants. (A) Relative invasion of INT407 cells by S. Typhimurium wild-type UK-1, χ11308 (ΔwaaG42), χ11309 (ΔwaaI43), χ9945 (ΔrfaH49), χ11310 (ΔwaaJ44), χ11312 (ΔwaaL46), χ11311 (ΔwbaP45), and χ9944 (Δwzy-48). Values represent invasion of each strain relative to that of the wild type. Assays were performed in triplicate on three independent occasions. Averages ± standard errors (error bars) are shown. Values that are significantly different from that of the wild type by the one-way ANOVA test and Dunnett post test are indicated by asterisks (*, P < 0.05; **, P < 0.01). (B) Approximately 1 × 10⁸ CFU/ml bacteria in DPBS (Dulbecco's phosphate-buffered saline) were incubated with either HIS (heat-inactivated serum) or PHS (pooled human serum) (final serum concentration, 50%) for 30 min at 37°C. Bacterial survival was enumerated by plating and counting colonies the following day. Data from three independent experiments are shown and expressed as means ± SD. Values that are significantly different from that of the wild type by the one-way ANOVA test and Dunnett post test are indicated by asterisks (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Complete LPS is an important factor in serum resistance (57). We examined the resistance of our mutants to the bactericidal effects of human serum (Fig. 2B). All mutants were significantly more sensitive to human serum than the smooth parent strain. In general, the trend of sensitivity to the pooled human serum (PHS) followed the length of the LPS, with strains producing longer LPS being more resistant. These data are consistent with a previous study showing that a wbaP mutant was more sensitive to serum than a wild-type strain (30). However, one notable exception was that the ΔwaaG42 mutant, producing the shortest core structure, was as resistant to serum killing as the ΔwaaL mutant, which carries a complete core, and more resistant than mutants with intermediate core structures. There were no differences in the sensitivity of any of the strains to heat-inactivated serum.

Virulence and colonization of mutant strains in mice.

We evaluated our mutants for virulence by determining the oral LD₅₀ of each strain in BALB/c mice. All mutants were highly attenuated, with no deaths at the highest dose tested, 1 × 10⁹ CFU (Table 3). To determine how well each strain could colonize mouse tissues, groups of mice were orally inoculated with 1 × 10⁹ CFU and the bacterial loads in Peyer's patches (PP), spleen, and liver were determined 6 days after inoculation. The mean CFU counts of the wild-type UK-1 strain recovered from PP (Fig. 3A), spleen (Fig. 3B), and liver (Fig. 3C) were approximately 10⁴, 10⁶, and 10⁶, respectively. Colonization levels of all mutants were significantly lower than the parent strain UK-1 (P < 0.05 to 0.001), and the relative differences observed among mutants were consistent for all organs examined. Notably, inactivation of waaG or waaI essentially eliminated the ability of S. Typhimurium to cause systemic infection. Both strains colonized PP poorly (<10² CFU/g), and their presence in spleen and liver was below the limit of detection (Fig. 3B and C). The ΔrfaH49 mutant colonized both spleen and liver better than either the ΔwaaG42 or ΔwaaI43 mutants. Although the ΔwaaJ44, ΔwaaL46, ΔwbaP45, and Δwzy-48 mutants were better able to colonize PP, spleen, and liver, the numbers of mutant bacteria isolated from each organ were significantly less than those of their UK-1 parent. Inflammation of the spleen, as determined by spleen weight, was also reduced significantly in mice inoculated with the mutants (P < 0.05 to 0.001) (Fig. 3D).

Fig. 3.

Colonization of mice by χ3761 and its isogenic mutants 6 days after oral inoculation. Groups of mice (3 mice/group) were orally inoculated with approximately 1 × 10⁹ CFU of the indicated strains. Log₁₀ CFU/g tissue determined on day 6 postinoculation in Peyer's patches (PP) (A), spleen (B), and liver (C) of inoculated BALB/c mice. (D) Day 6 spleen weights from inoculated mice. The horizontal lines represent the means, and the error bars represent the standard errors of the means. ***, P < 0.001 compared with UK-1, **, P < 0.01 compared with UK-1; *, P < 0.05 compared with UK-1. Samples that were negative by direct plating and positive by enrichment in selenite cysteine broth were recorded as 1 CFU/organ.

When Salmonella is administered orally, it must overcome the low-pH barrier of the stomach, followed by the harsh environment of the intestinal tract, including the presence of organic acids, bile salts such as DOC, and antimicrobial peptides similar to polymyxin B, before interacting with host cells. While we did not test the acid resistance of our mutants, previous reports have indicated that rough strains of Salmonella are more sensitive to the presence of organic acids than smooth strains (3). This, coupled with the observation that our mutants exhibit increased sensitivity to DOC and polymyxin B, led us to examine alternate routes of immunization to bypass the barriers imposed by the alimentary tract. Therefore, we evaluated the ability of the ΔwaaI43, ΔwaaL46, and Δwzy-48 mutants to colonize PP, spleen, and liver after i.n. inoculation. Mice receiving the mutants were evaluated after 6 days, and mice receiving the virulent parent were evaluated after 3 days (Fig. 4). All three mutants achieved levels approximately 10-fold greater for each tissue than the levels reached when administered orally. The ΔwaaL and Δwzy mutants were able to colonize all tissues at levels comparable to those for UK-1, indicating that O-antigen is not required for colonization when Salmonella is administered intranasally. There does appear to be some benefit of core, since the ΔwaaI mutant did not colonize as well as the other strains.

Fig. 4.

Colonization of mouse organs by χ3761 and its isogenic mutants after intranasal (i.n.) administration with 10⁹ CFU. Log₁₀ CFU/g tissue were determined on day 6 postinoculation for the mutants and on day 3 postinoculation for the wild-type strain in Peyer's patches (A), spleen (B), and liver (C) of BALB/c mice. The horizontal lines represent the means, and the error bars represent standard errors of the means.

Effects of the mutations on the immunogenicity of recombinant S. Typhimurium strains by oral administration.

We were also interested in determining the effect of core length on the ability of an attenuated vaccine to deliver a heterologous antigen. To address this, we introduced each of the mutations described above into attenuated Salmonella strain χ9241 (Table 1). Strain χ9241 is attenuated due to the presence of the ΔpabA ΔpabB mutations and produces smooth LPS. We introduced plasmid pYA4088 encoding pspA, a pneumococcal gene encoding the protective antigen PspA, into each strain. Immunoblot analyses of subcellular fractions from χ9241(pYA4088) have shown that 51.5% of the Bla-ss-PspA fusion protein is secreted into the periplasm, 5.4% is secreted out of the cell, and the remaining 43.1% is retained in the cytoplasm (65). Nine groups of mice were immunized on day 1 to 2 with 1 × 10⁹ of each χ9241 derivative and boosted with the same dose of the same strain 5 weeks later. Mice immunized with strain χ9241 carrying the empty vector pYA3493 served as negative controls. All mice survived immunization, and we did not observe any signs of disease in the immunized mice during the experimental period. We measured serum IgG antibody responses to rPspA (Fig. 5A) and to Salmonella LPS (Fig. 5B) in immunized mice. Parent strain χ9241(pYA4088) and its isogenic Δwzy-48 counterpart induced the highest levels of IgG against PspA and LPS among all the vaccine strains. At 8 weeks post-primary inoculation (3 weeks after the boost), the Δwzy-48 and ΔwbaP45 vaccine strains induced anti-PspA IgG titers similar to those of the parent strain χ9241(pYA4088). Anti-PspA IgG titers in sera from mice immunized with the other vaccine strains were significantly lower (P < 0.05 to 0.001). Serum IgG against Salmonella LPS induced by each of the χ9241 derivatives was also significantly lower (P < 0.01) than that of the parent strain χ9241(pYA4088).

Fig. 5.

Serum responses in orally immunized and control mice. Total serum IgG specific for rPspA (recombinant pneumococcal surface protein A) (A), S. Typhimurium LPS (B), IgG1 specific for rPspA, (C) and IgG2a specific for rPspA (D) were measured by ELISA. Each group has 7 mice. Data represent reciprocal anti-IgG antibody titers in pooled sera from mice orally immunized with attenuated Salmonella carrying either pYA4088 (pspA) or pYA3493 (control) at the indicated number of weeks postimmunization. The error bars represent variations between triplicate wells. Mice were boosted at week 5. ***, P < 0.001; **, P < 0.01; *, P < 0.05 compared to titers from mice immunized with χ9241(pYA4088). #, the ΔwbaP45 derivative of χ9241 was significantly (P < 0.05) different from the ΔwaaL46 derivative. No immune responses were detected to PspA in mice immunized with χ9241 containing the control plasmid (reciprocal titer, <1:50). Immune responses (IgG, IgG1, and IgG2a) against PspA were below the level of detection in mice immunized with waaG42 and rfaH49 at 4 weeks (reciprocal titers, <1:50). Immune responses against LPS were below the level of detection in mice immunized with the ΔwaaG42, ΔwaaI43, ΔrfaH49, ΔwaaJ44, and ΔwaaL46 strains at 4 weeks and with the ΔwaaG42 strain at 8 weeks (reciprocal titers, <1:50).

The immune responses to PspA were further evaluated by measuring the levels of anti-PspA IgG isotype subclasses IgG1 and IgG2a in each group of mice. At week 4, all strains elicited a balanced Th1/Th2 response, based on the fact that the mice produced similar titers of IgG1 and IgG2a (Fig. 5C and D), with the exception of the ΔwbaP strain, which elicited higher titers of IgG2a than IgG1 compared to the other mutants (P < 0.05), indicating a Th1-biased response. Interestingly, the wbaP strain elicited significantly higher titers of IgG1 and IgG2a than the waaL strain (Fig. 5C and D), though both mutations resulted in the same LPS structure (Fig. 1). By week 8, the titers of IgG2a in mice immunized with χ9241(pYA4088) and the ΔwaaJ, ΔwaaL, ΔwbaP, and Δwzy mutants were higher than the IgG1 titers, indicating a shift to a Th1-biased response. The other mutants were poorly immunogenic, and the IgG1/IgG2 ratio was the same at weeks 4 and 8.

Effects of the mutations on the immunogenicity of recombinant S. Typhimurium strains by i.n. or i.p. administration.

Based on our observation that i.n. inoculation provided better access of the ΔwaaI, ΔwaaL, and Δwzy mutants to mouse lymphoid tissues (Fig. 4), we wanted to assess the immunogenicity of χ9241 derivatives when administered i.n. or i.p. Both are routes that bypass the alimentary tract. LD₅₀ values of Salmonella vary based on the route of administration used; therefore, different doses were used to immunize the mice via different routes of administration. Four groups of mice were immunized i.n. with 1 × 10⁷ CFU or i.p. with 2 × 10⁶ CFU of strain χ9241 or its ΔwaaI43, ΔwaaL46, or Δwzy-48 derivatives, all carrying plasmid pYA4088, and boosted with the same dose of the same strain 5 weeks later. Serum IgG responses to PspA and to Salmonella LPS were measured by ELISA (Fig. 6A). The Δwzy-48 mutant induced the highest titers of anti-PspA IgG in the i.n. immunized groups, significantly higher than those for the parent strain χ9241 at week 8. Also at week 8, the anti-PspA titers in mice immunized with the parent strain χ9241(pYA4088) and the isogenic ΔwaaL46 mutant were identical.

Fig. 6.

Serum responses in mice immunized by intranasal (i.n.) or intraperitoneal (i.p.) administration. Reciprocal serum IgG titers in mice immunized i.n. or i.p. with the indicated Salmonella strains carrying pYA4088. Anti-PspA (pneumococcal surface protein A) responses in mice immunized i.n. (A) or i.p. (B) and anti-LPS responses in mice immunized i.n. (C) or i.p. (D) were measured by ELISA. There were 6 mice in each group. Mice were boosted at week 5. The error bars represent variations between triplicate wells. Mice were boosted at week 5. ***, P < 0.001; **, P < 0.01; *, P < 0.05 compared to titers from mice immunized with χ9241(pYA4088).

When administered i.p., the Δwzy mutant also induced higher anti-PspA serum IgG titers than the other mutants and titers were similar to those for the parent strain, χ9241(pYA4088), by week 8 (Fig. 6B). The anti-PspA titers elicited by the ΔwaaI43 and ΔwaaL46 mutants were significantly lower than the titers observed in mice immunized with the parent strain. The anti-LPS IgG titers induced by the three mutants were significantly lower than titers elicited by the parent strain by either immunization route (Fig. 6C and D).

Evaluation of immunogenicity against OMPs from different enteric bacteria.

Previous work has shown that a complete lack of or regulated expression of dominant surface antigens, such as O-antigen, increases the levels of serum antibodies that are cross-reactive to conserved surface epitopes of other enteric bacteria (13, 35, 44). Therefore, we determined whether sera raised against our mutants that have truncated core and O-antigen had a similar effect. We evaluated the reactivity of pooled immune sera (n = 7) from orally immunized mice described in the previous experiment (Fig. 5) against purified OMPs from homologous and heterologous wild-type strains by ELISA (Fig. 7). The results roughly correlated with the results we obtained against PspA and LPS (Fig. 5). Antibodies raised against the parent vaccine χ9241(pYA4088) reacted more strongly to OMPs from three enteric strains tested than those induced by any of its mutant derivatives. In all eight vaccine strains, ΔwaaG and ΔwaaL derivatives consistently induced the lowest reactivity against OMPs, while the Δwzy derivative induced the highest cross-reaction against the OMPs (Fig. 7). Similar results were also achieved from the immunized mice by i.n. or i.p. route administration (Fig. 8), although the Δwzy mutant delivered i.p. elicited the highest level of antibodies cross-reactive to Salmonella enterica serotype Enteritidis (Fig. 8B).

Fig. 7.

Cross-reactivity of serum IgG from orally immunized mice against OMPs (outer membrane proteins) from other enteric bacteria. IgG cross-reactivity of sera (1: 1,000 dilution) obtained from mice (n = 5) immunized with χ9241(pYA4088) or its mutant derivatives to homologous and heterologous OMPs 8 weeks after the primary immunization by oral administration. Sera from mice immunized with χ11313 (ΔwaaG42), χ11314 (ΔwaaI43), χ9884 (ΔrfaH49), χ11315 (ΔwaaJ44), χ11316 (ΔwbaP45), χ11317 (ΔwaaL46), χ9885 (Δwzy-48), and χ9241 (parent vaccine strain) were individually tested against purified outer membrane proteins from enteric bacteria S. Typhimurium χ3761 (A), an avian pathogenic E. coli O78 strain χ7122 (B), or S. Enteritidis χ3550 (C). The error bars represent variations from triplicate wells. ***, P < 0.001; **, P < 0.01; *, P < 0.05 compared to titers from mice immunized with χ9241(pYA4088).

Fig. 8.

Cross-reactivity of serum IgG from intranasal (i.n) and intraperitoneal (i.p.) immunized mice against OMPs (outer membrane proteins) from three enteric bacteria. IgG cross-reactivity of sera (1: 1,000 dilution) obtained from mice (n = 5) immunized with χ9241(pYA4088) or its mutant derivatives to homologous and heterologous OMPs 8 weeks after the primary immunization by i.n. route administration (A) and by i.p. route administration (B). Sera from mice immunized with χ11314 (ΔwaaI43), χ11317 (ΔwaaL46), χ9885 (Δwzy-48), and χ9241 (parent vaccine strain) were each individually tested against purified outer membrane proteins from S. Typhimurium strain χ3761 (Typhimurium), avian pathogenic E. coli O78 strain χ7122 (E. coli), and S. Enteritidis strain χ3550 (Enteritidis). The error bars represent variations from triplicate wells. ***, P < 0.001; **, P < 0.01; *, P < 0.05 compared to titers from mice immunized with χ9241(pYA4088).

DISCUSSION

Attenuated salmonellae have the ability to stimulate both cellular and humoral immunity against homologous or heterologous antigens after oral administration (9, 10, 12, 14). O-antigen and core sugar comprise key virulence determinants (47, 57, 60). An incomplete core or O-antigen of LPS may result either from deficiency in glycosyltransferase activity, from the inability to synthesize sugars that are part of the core or O-antigen chain, or from deletion of LPS processing genes such as wzy, rfaH, or wzx (16, 20, 34, 35). The first can be achieved by deleting either waa or wba genes (31), leading to nonreversible truncation of LPS, and the second can be obtained by deleting genes such as galE or pmi, responsible for dNDP-sugar synthesis, leading to reversible LPS truncation (11, 13, 29, 37, 58). In this study, we systematically investigated the relationship between S. Typhimurium LPS core/O-antigen length and virulence and immunogenicity.

Deletion of the waa or wba genes targeted in this study resulted in truncated core structures (Fig. 1A and B) (31). rfaH mutants are reported to produce a mixed core structure containing the major Rb3, minor Ra, Rb1, Rb2, and some Ra being capped by shorter or longer polymers of O-antigen units (39, 56). Our results with the ΔrfaH49 mutant are consistent with the expected phenotype, as we observed primarily an Rb3 truncated core (similar to results with the ΔwaaI43 mutant) (Fig. 1A and B) in addition to O-antigen detectable only by our P22 phage transduction assay (Table 3). This is likely to be due to the fact that RfaH is a transcriptional anti-terminator that reduces the polarity of long operons by binding to the ops sequence located in the 5′ regions of the waa, wba, and other operons (2). Therefore, the rfaH deletion does not fully abolish synthesis of full-length transcripts (1, 2). As expected, all mutants with truncated outer core and O-antigen were more sensitive to DOC, polymyxin B, and pooled human serum than wild-type UK-1, while all mutants were resistant to ox source bile (Table 3 and Fig. 2C). It is surprising that the ΔwaaG42 mutant that contains only the inner core was as resistant to PHS as the ΔwaaL46 mutant that produces a complete core and significantly (P < 0.01) more resistant to serum than the ΔwaaJ44, ΔrfaH49, and ΔwaaI43 mutants (Fig. 2), each of which contains more sugar moieties attached to the terminal inner core (Fig. 1A). In addition, the waaG46 mutant was 2-fold more resistant to polymyxin B than the ΔwaaJ44, ΔrfaH49, and ΔwaaI43 mutants (Table 3). One explanation for this result is that the loss of O-antigen in our mutants may expose and/or activate outer membrane proteases that partially restore serum or polymyxin B resistance. It has been demonstrated that outer membrane proteins such as PgtE, PagC, and Rck contribute to serum and/or antimicrobial peptide resistance (21, 23, 26, 48). It is known that PgtE, a Salmonella β-barrel surface protease, can function to activate plasminogen in rough but not smooth strains (21, 36) and also to inactivate complement components (52). In addition, pgtE is regulated posttranscriptionally by PhoP (21, 36), a regulator known to direct modifications to LPS structure in vivo (22). It has been postulated that pgtE is activated in vivo, where it is known that rfa gene expression is downregulated (40). Thus, it is possible that the activity of PgtE or some other relevant protease is increased in our rough mutants and that in the ΔwaaG42 mutant, these postulated enzymes are more stable or more active than they are in some of the other rough mutants. In this regard, results reported by Nishio et al., who examined the effect of pagC expression in wild-type Salmonella enterica serovar Choleraesuis and an isogenic waaG mutant, demonstrated that the waaG mutant was more sensitive to serum than the wild-type strain (48), presumably due to the loss of O-antigen. However, deletion of pagC in the waaG mutant resulted in a further increase in sensitivity to serum (48), indicating that PagC was active in that mutant.

All mutants with a single mutation were avirulent in mice by oral administration. In addition, the mutants had significantly reduced abilities to colonize lymphoid organs of mice compared to those of the wild-type parent strain χ3761 when administered orally. The ability to colonize was roughly in proportion to the length of the LPS; i.e., mutants with longer core structures were better able to colonize. The ΔwaaG and ΔwaaI mutants were unable to reach the spleen and liver when administered orally (Fig. 3). However, when administered i.n., the ΔwaaI mutant was able to colonize these organs (Fig. 4), suggesting that increased sensitivity of both ΔwaaG and ΔwaaI mutants to bile in the intestinal tract (Table 3) and perhaps low pH in the stomach led to reduced survival in the intestines. This, coupled with a decreased ability to invade epithelial cells (Fig. 2) (6), could account for their observed inability to cross the intestinal epithelial barrier and reach the internal lymphoid organs. In all, the virulence and colonization data indicate that the O-antigen and outer core are required for optimal invasion via oral administration by Salmonella and are essential for resistance to bile, resistance to peptides, and immune evasion during invasion to the intracellular niche (27). However, when bypassing the intestinal tract, these mutants retained some capacity to colonize host tissues (Fig. 4) and the retention of some, albeit suboptimal, immunogenicity (Fig. 6).

To evaluate the effect of each of the LPS alterations on the immunogenicity of an attenuated S. Typhimurium vaccine, each mutation was moved into the vaccine strain χ9241 that contains attenuating mutations ΔpabA and ΔpabB, which do not affect LPS structure. Mice immunized with χ9241 derivatives expressing the heterologous antigen gene pspA produced anti-PspA serum IgG, although, with the exception of that of the Δwzy mutant, the titers were significantly lower than those in mice immunized with parent strain χ9241 (Fig. 5A). With the exception of the Δwzy mutant, the anti-LPS titers were significantly reduced for all mutants compared to the parent strain (Fig. 5B). This is likely to be due to a combination of the fact that the rough mutants do not produce O-antigen (Fig. 1B) and the poor overall immunogenicity of the rough mutants due to their decreased colonizing ability. However, the ΔwbaP mutant elicited detectable anti-LPS titers (Fig. 5B). Although we did not analyze the antibodies beyond what is shown in Fig. 5, it is likely that the responses detected were directed toward the lipid A and core regions of LPS, which were present in the ELISA coating antigen. This kind of response could be desirable if one is interested in producing a vaccine against other bacteria that share the same or similar core epitopes, such as in other Salmonella and Shigella spp. and pathogenic Escherichia coli (17, 51, 63).

Removal of the immunodominant O-antigen from the surface of Salmonella may result in enhanced immunogenicity of outer membrane proteins and other surface antigens (13, 44). In a previous study, we described a strain χ9241 derivative, χ9852, with arabinose-regulated expression of rfaH (35). This strain displayed a reversibly rough phenotype, dependent on arabinose availability. Mice orally immunized with strain χ9852(pYA4088) developed serum IgG antibodies with enhanced cross-reactivity against OMPs from a variety of enteric bacteria compared with mice immunized with the smooth parent strain, χ9241(pYA4088). This result has potential implications for developing a multivalent vaccine to protect against multiple pathogens in a single vaccine platform. In the current study, we found that mutations resulting in permanently rough χ9241 derivatives did not enhance production of cross-reactive antibodies to OMPs from any of the enteric bacteria tested (Fig. 7A to C), with the exception of the Δwzy strain at week 8, although, while statistically significant, the enhancement was much less than we observed previously with the regulated rfaH strain χ9852 (35), or even with an arabinose-regulated wzy strain (34). This most likely is due to the poor colonization and rapid elimination of the rough vaccine strains by the host, leading to their overall poor immunogenicity when administered orally.

In summary, we systematically examined the relationship between S. Typhimurium LPS length and virulence, colonization potential, and immunogenicity. We conclude that mutants with truncated core and O-antigen colonize mice poorly and thus elicit a suboptimal humoral response against a vectored antigen. However, the wzy-48 mutant was roughly comparable to the smooth parent strain χ9241 in most studies and therefore may have potential to be used, along with other attenuating mutations, in developing RASVs capable of delivering heterologous or homologous antigens to induce protective immunity. However, these results, combined with our previous studies, lead us to recommend the use of reversibly rough strains as a preferred mode of driving the production of cross-reactive core and/or OMP antibodies, due to the inability of permanently rough strains to colonize and elicit high-titer antibody responses.