Analysis of Spleen-Induced Fimbria Production in Recombinant Attenuated Salmonella enterica Serovar Typhimurium Vaccine Strains

ABSTRACT Salmonella enterica serovar Typhimurium genome encodes 13 fimbrial operons. Most of the fimbriae encoded by these operons are not produced under laboratory conditions but are likely to be synthesized in vivo. We used an in vivo expression technology (IVET) strategy to identify four fimbrial operons, agf, saf, sti, and stc that are expressed in the spleen. When any three of these operons were deleted, the strain retained wild-type virulence. However, when all four operons were deleted, the resulting strain was completely attenuated, indicating that these four fimbriae play functionally redundant roles critical for virulence. In mice, oral doses of as low as 1 × 105 CFU of the strain with four fimbrial operons deleted provided 100% protection against challenge with 1 × 109 CFU of wild-type S. Typhimurium. We also examined the possible effect of these fimbriae on the ability of a Salmonella vaccine strain to deliver a guest antigen. We modified one of our established attenuated vaccine strains, χ9088, to delete three fimbrial operons while the fourth operon was constitutively expressed. Each derivative was modified to express the Streptococcus pneumoniae antigen PspA. Strains that constitutively expressed saf or stc elicited a strong Th1 response with significantly greater levels of anti-PspA serum IgG and greater protective efficacy than strains carrying saf or stc deletions. The isogenic strain in which all four operons were deleted generated the lowest anti-PspA levels and did not protect against challenge with virulent S. pneumoniae. Our results indicate that these fimbriae play important roles, as yet not understood, in Salmonella virulence and immunogenicity.

B acterial pathogens produce adhesins, often associated with fimbrial structures on the cell surface, to facilitate their initial interactions with host tissues (1). The chromosome of Salmonella enterica serovar Typhimurium contains 13 fimbrial operons, agf (csg), bcf, fim, lpf, pef, saf, stb, stc, std, stf, sth, sti, and stj (2)(3)(4). While the functions of a few of these fimbriae, including type 1 fimbriae (Fim), have been characterized (1,5), the functions of most fimbriae are unknown. This is due, in part, to the fact that only type 1 and Agf fimbriae are produced under laboratory growth conditions (6). Type 1 fimbriae are produced when cells are grown at 37°C, and Agf fimbriae are produced when cells are grown at 26°C (7). While it is possible that some of these other fimbriae may be required for life outside a host (8), it is likely that many play an as yet undiscovered role in host interactions.
The agf operon encodes thin aggregative fimbriae (9) in Salmonella, and these fimbriae were later found to be similar to the fibronectin-binding surface structure known as curli (10) originally described in Escherichia coli (11). Thin aggregative fimbriae (hereafter Agf fimbriae) and curli are not produced in vitro at 37°C (11). Production of Agf fimbriae is typically induced in laboratory settings by growing cells at 26°C. Pef fimbriae mediate adherence to the murine small intestine and are required for fluid accumulation in infant mice. Expression of pef genes is regulated by DNA methylation (12). Stf fimbriae share homology with MR/P fimbriae of Proteus mirabilis and E. coli Pap fimbriae (13), and expression of stfA is induced during infection of bovine ileal loops (14). Long polar fimbriae (Lpf) are important for colonization of Peyer's patches in mice by mediating adherence to M cells (5). Lpf also plays a role in the early stages of biofilm formation on host epithelial cells (15) and is involved in intestinal persistence (16). Lpf synthesis is regulated by an on-off switch mechanism (phase variation) to avoid host immune responses (17). Some S. enterica fimbriae have been shown to serve functions beyond those required for interactions at the intestinal mucosal surface. For example, the Agf fimbriae are required for biofilm formation in the gallbladder (18,19). In addition, the Stg fimbriae of S. enterica serovar Typhi, required for adherence to epithelial cells, also serves to inhibit phagocytosis (20). In S. Typhimurium, most fimbriae are produced in vivo, since mice immunized with S. Typhimurium produce antibodies against fimbrial subunits AgfA, BcfA, FimA, LpfA, PefA, StbA, StcA, StdA, StfA, SthA, and StiA (6). Thus, it is likely that some of these uncharacterized fimbriae may be synthesized in extraintestinal tissues.
To investigate potential roles for S. Typhimurium fimbriae in the host, we utilized an in vivo expression technology (IVET) strategy (21). We identified four fimbrial operons that are actively expressed in the spleen, only one of which, agf, is synthesized during in vitro growth (at 26°C). We characterized the impact of deletion and constitutive expression of all four fimbriae on virulence and immunogenicity.

RESULTS
Identification of fimbrial operons expressed in the spleen by IVET. We constructed 12 S. Typhimurium strains, each harboring chromosomal transcriptional fusions of fimbrial promoter regions with aph lacZ reporter genes (Fig. 1). The stj operon is incomplete due to the apparent absence of any identifiable fimbrial subunit genes, so it was not included in our study (2). However, it is likely that this operon encodes a nonfimbrial or fibrillar structure (4). A mixture of all 12 fusion strains were orally administered to BALB/c mice. After infection, mice were treated orally and intraperitoneally with three doses of kanamycin to select for S. Typhimurium clones expressing the aph reporter gene in vivo. The experiment was performed twice, and 96 clones were obtained from pooled spleen samples in each experiment. Clones were identified by PCR using specific primers (see Table S2 in the supplemental material). In both experiments, we recovered the same four S. Typhimurium strains, strains 9451, 9453, 9456, and 9461, which contain aph lacZ reporter genes fused to the promoter regions of stiABCH, safABCD, agfBAC, and stcABCD operons, respectively ( Table 1). Each of these strains was sensitive to kanamycin when grown at 37°C on LB agar plates, indicating that these four fimbrial operons are expressed in the mouse host.
Virulence and immunogenicity of S. Typhimurium fimbrial mutants in BALB/c mice. Because these fimbrial promoters are active in the spleen, we hypothesized that the fimbriae may be important for virulence. Therefore, we constructed strains harboring single and multiple deletions of the four fimbrial operons, ΔstiABCH1225, ΔsafABCD31, Δ(agfC-agfG)-999, and ΔstcABCD36. BALB/c mice were orally administered graded doses of bacteria and monitored for 4 weeks. All single deletion mutants Fimbria Production in Live Salmonella Vaccines ® retained wild-type virulence (data not shown). Strains with any three of the four fimbrial operons deleted were also virulent ( Table 2). In contrast, two independently constructed quadruple deletion mutants, 11484 and 11599, were fully attenuated, with no deaths or disease symptoms occurring at the highest dose tested (50% lethal dose [LD 50 ] Ͼ~1 ϫ 10 9 to 2 ϫ 10 9 CFU).
We evaluated the immunogenicity of one of the strains, 11484, by determining its ability to confer protection against challenge with the virulent S. Typhimurium UK-1 strain 3761. The mice used in the virulence assay (above), which received graded doses of strain 11484 were challenged 4 weeks after immunization with S. Typhimurium 3761 (Table 3). A control group was given sterile buffer. Protection was achieved at all doses, and all mice that were immunized with at least 1.4 ϫ 10 5 CFU of 11484 survived challenge with the virulent S. Typhimurium strain (Table 3). Even mice inoculated with a single dose of only 8.4 ϫ 10 2 CFU were partially protected, indicating that this avirulent S. Typhimurium fimbrial quadruple mutant is highly immunogenic.
Colonization by S. Typhimurium fimbrial quadruple mutants. To evaluate the impact of the quadruple deletion on colonization, mice were orally inoculated with either strain 11484 or strain 11599. Peyer's patches, spleens, and livers were harvested 5 days later, and the bacteria in each tissue were enumerated. Both quadruple mutants colonized all tested organs as well as wild-type 3761 strain did (data not shown). To look more closely at spleen and liver colonization, we performed a competition assay. We chose to inoculate by the intraperitoneal route to eliminate any differences between strains that might be due to passage through the gastrointestinal tract. Thus, mice were inoculated parenterally with a mixture of S. Typhimurium wild-type 3761 and either 11484 or 11599. Each strain was marked with a stable low-copy-number chloramphenicol-resistant plasmid (pHSG576) or kanamycin-resistant plasmid (pWSK129). Groups of mice were eu-   thanized on days 1 and 3 postinfection. Samples of the spleens and livers were plated for enumeration of Salmonella. The total numbers of Salmonella recovered from each organ were consistent from mouse to mouse, between 10 4 and 10 6 CFU per g of tissue (data not shown). The ratio of the two strains in each organ was determined and compared to the input ratio to determine the competitive index (CI). On day 1 postinfection, there were no differences in spleen colonization between wild-type and mutant strains (Fig. 2B), while strain 11484, but not 11599, was outcompeted by the wild-type strain in the liver ( Fig. 2A) (P Ͻ 0.01). By day 3, the wild type had outcompeted both quadruple mutants in both the spleen and liver (P Ͻ 0.05), indicating an important role for saf, sti, stc, and agf in colonization of the spleen and liver in mice. In preliminary competition experiments comparing single deletion mutants and the wild type, no significant differences were observed between strains (data not shown), indicating that no single fimbria is responsible for this phenotype. Recombinant attenuated S. Typhimurium vaccine (RASV) strains producing fimbriae (Saf ؉ , Sti ؉ , Stc ؉ , and Agf ؉ ) in a constitutive manner. Our results showing that the saf, sti, stc and agf fimbrial operons are expressed in vivo led us to speculate as to whether these fimbriae could be exploited to enhance the immunogenicity and protective efficacy of Salmonella vaccine strains. For this work, we constructed derivatives of attenuated S. Typhimurium strain 9088 [ΔP fur33 ::TT araC P BAD fur Δpmi-2426 Δ(gmd-fcl)-26 ΔasdA33] (22) in which three fimbrial operons were deleted and the fourth was expressed from the constitutive P murA promoter (23). Consequently, the resulting strains, strains 11595, 11850, and 11851, have a genetic background that includes attenuating mutations, deletions in three fimbrial operons, and one deletioninsertion mutation ( Table 4). The agf genes are expressed from two divergent operons, agfDEFG and agfBAC, necessitating a different strategy. In this case, we introduced the previously described agfD812 mutation (24) to drive constitutive expression of the agf operon. Strain 12038 constitutively produced Agf fimbriae as indicated by the red, dry, 1.4 ϫ 10 9 6/6 6/6 (100) 1.4 ϫ 10 7 6/6 6/6 (100) 1.4 ϫ 10 5 6/6 6/6 (100) 8.4 ϫ 10 2 6/6 4/6 (67) a BALB/c mice were immunized orally with the indicated dose of strain 11484 (all mice survived) and challenged 4 weeks after immunization with~1 ϫ 10 9 CFU of S. Typhimurium wild-type strain (3761). Fimbria Production in Live Salmonella Vaccines ® and rough (rdar) colony morphology when grown on Congo red plates (7) at 37°C (data not shown). For a control, we also constructed strain 11606, which harbors deletions of all four fimbrial operons (agf, saf, sti, and stc).
To study the ability of these strains to elicit protective immune responses against heterologous antigens in mice, we introduced plasmid pYA4088 (25), carrying the gene encoding the Streptococcus pneumoniae protein PspA, into each strain. This pneumococcal protein has been extensively studied by our group (26) and others (27,28) and shown to elicit protective immunity against virulent S. pneumoniae challenge. For clarity, we will refer to these strains as 11595(pYA4088) (Sti ϩ ), 11850(pYA4088) (Saf ϩ ), 11851(pYA4088) (Stc ϩ ), 12038(pYA4088) (Agf ϩ ), and 11606(pYA4088) (Δ4). All strains were grown to mid-log phase in LB with appropriate supplements. Western blot analysis with specific anti-recombinant PspA (anti-rPspA) antibodies showed that all strains produced similar amounts of PspA (Fig. S1).

DISCUSSION
Fimbrial genes are widely distributed among bacteria, but only a few fimbriae are produced under standard laboratory conditions. Most bacterial fimbriae serve to present adhesins that assist in the adherence of bacteria to biotic and abiotic surfaces (1) and are produced in response to the appropriate environmental cues. Of the 13 known fimbriae in S. Typhimurium, only two, type 1 fimbriae and curli (Agf) are readily produced when grown in the laboratory. However, in one study, cells were coaxed to produce Pef, Bcf, Stb, Stc, Std, and Sth fimbriae after static growth in CFA broth at 32°C and Agf, Pef, Lpf, Stc, Stf, and Sth fimbriae in LB at pH 5.1 at 37°C, although the levels were low, as fimbriae were detected on Ͻ7% of the cells by a highly sensitive flow cytometry method (14). In the same study, fimbrial expression was further enhanced to around 10% of cells after growth for 8 h in bovine ileal loops. Type 1 fimbriae were detected in Ͼ20% of the cells under all three growth conditions. Thus, it is likely that a majority of the known S. Typhimurium fimbrial operons are expressed inside a mammalian host.
In the current study, we demonstrated that sti, saf, stc, and agf fimbrial genes are actively expressed in the mouse spleen (Table 1). In vivo expression of these genes is consistent with a previous study in which CBA mice inoculated with S. Typhimurium developed antibodies against recombinant His-tagged StiA, StcA, and AgfA (6). The mice were not evaluated for antibody responses against Saf fimbrial components. In another study, mice were protected from challenge with S. Typhimurium after injection with a mixture of purified recombinant His-tagged SafB, a putative chaperone, and recombinant SafD, the Saf adhesin, both produced in E. coli (29,30). In addition, transcription of saf fimbrial genes has been detected in blood samples from patients infected with S. Typhi (31) and S. Paratyphi A (32), supporting a role for these fimbriae in the human host.
SafA monomeric fimbriae were assembled in vitro in the presence of the chaperone protein SafB and crystallized (33). Subsequent crystallographic analysis showed that Saf fimbriae are composed of highly flexible fibers formed by globular subunits organized in a "beads on a string" arrangement (33). Characterization of the safABCD operon protein sequences suggest that SafA is the major structural protein, SafB is the periplasmic chaperone, and SafC is an outer membrane usher (29). The SafD protein is homologous to several other fimbrial adhesins and so is likely to be the Saf adhesin, believed to be present only at the tip of the fiber (29,33). In addition, the major fimbrial protein, SafA, exhibits similarity to the phage-encoded Bor protein that has been implicated in serum resistance of -infected hosts (34). Thus, it is possible that the saf fimbriae play a role in serum resistance.
Our results with agf seem to run counter to a previous report. Using a bioluminescence imaging technique, White et al. showed that agfB was not expressed during infection (35). The authors concluded that Agf fimbriae are not produced in vivo. However, their observations were based on results obtained from a single time point, while in our study, the bacteria were under constant selective pressure for 3 days. Thus, it is possible that there is a temporal component to agfB expression. Our data suggesting the in vivo production of Agf is also supported by the study we cite above in which anti-AgfA antibodies were detected in S. Typhimurium-infected mice (6).
In a previous study, strains with either agfAB or stcABCD deleted exhibited wild-type levels of spleen colonization in genetically resistant CBA mice (16). Consistent with those results, we observed that strains in which any single fimbrial operon (agf, saf, stc, or sti) or combination of three operons was deleted had no effect on virulence, while deletion of all four fimbrial operons resulted in a complete loss of virulence when mutant strains were administered by the oral route to genetically sensitive BALB/c mice ( Table 2). Our results suggest that these four fimbriae serve functionally redundant roles in mouse virulence. Interestingly, while a ΔstcABCD strain exhibits wild-type spleen colonization, it exhibits reduced fecal shedding, indicating a role for this fimbriae in long-term intestinal carriage (16).
Strain 11484 with sti, saf, stc, and agf deleted was immunogenic, protecting mice from a high-dose challenge with wild-type S. Typhimurium after a single immunizing dose as low as 1.4 ϫ 10 5 CFU (Table 3). We expanded our analysis of the roles of these genes in immunogenicity by examining the effect of constitutive production of each fimbriae individually in a previously characterized vaccine strain background (9088) in which we had also deleted the other three fimbrial operons. These vaccine strains were used to deliver the heterologous antigen, PspA. Our results indicate that constitutive production of Sti, Saf, or Stc, but not Agf, significantly enhanced protective immunity (Table 5), although they each had different impacts on the immune system.
Th1-type dominant immune responses are frequently observed after immunization with attenuated Salmonella (36), and most of the fimbrial deletion strains elicited a Th1-biased response. However, mice immunized with strain 12038(pYA4088) (Agf ϩ ) produced more of a mixed Th1/Th2 humoral response, indicating that overproduction of Agf fimbriae resulted in a reduced ability to stimulate Th1 helper cells to direct IgG class switching to IgG2a (37). IgG2a is the isotype with the greatest capacity to mediate complement deposition onto the surfaces of bacteria, and an increase in anti-PspA IgG2a has been correlated with increased C3 deposition on the S. pneumoniae cell surface (38).
The immune responses to PspA were examined by measuring the levels of IgG isotype subclasses. The anti-PspA IgG2a titers were higher than the IgG1 titers in all groups, indicating that all of the Salmonella vaccines induced a Th1-biased response against PspA (Fig. 3). Strain 11850(pYA4088) (Saf ϩ ) elicited high levels of anti-PspA IgG with a strong Th1 bias (Fig. 3). Thus, the strong Th1 responses observed in mice vaccinated with strain 11850(pYA4088) (Saf ϩ ) can explain why this strain was highly protective (Fig. 3 and Table 3). Strain 11595(pYA4088) (Sti ϩ ) produced a strong Th1 response by week 9 (Fig. 3A). In contrast, the strains that provided the weakest protection, 11606(pYA4088) (Δ4) and 12038(pYA4088) (Agf ϩ ), were deficient in either strong Th1-biased antibody responses.
The strong protection observed for mice immunized with strain 11851(pYA4088) (Stc ϩ ) does not fit as neatly into this interpretation, as this strain did not elicit a strong Th1 response at early time points (Fig. 3). However, by week 9, this strain elicited the greatest IgG2a/IgG1 ratio (Fig. 3C), which may have provided a humoral response that was adequate to control the S. pneumoniae challenge. This result, along with the results for strain 12038(pYA4088) (Agf ϩ ), which stimulated a low IgG2a/IgG1 ratio, indicates that production of IgG2a is the most important parameter for protection against pneumococcal challenge in this model. Deletion of all four fimbrial operons in strain 11484 resulted in complete attenuation (Table 2), while preserving its ability to elicit a protective response against challenge with wild-type S. Typhimurium at immunizing doses as low as 8.4 ϫ 10 2 CFU (Table 3). In contrast, deletion of these same four fimbrial operons in the 9088 background vectoring PspA compromised the ability of the strain to elicit protection against streptococcal challenge (Table 5). Since the Δ4 deletion is attenuating, combining these mutations with additional attenuating mutations could have resulted in overattenuation of the Salmonella vector strain, possibly due to a reduction in the ability of strain 11606(pYA4088) to colonize the spleen or other lymphoid organs. While the basis of this overattenuation is not clear, it does indicate that one must carefully consider the background genotype before combining Δ4 with other attenuating mutations.
This study demonstrates that in vivo-induced fimbriae play a role in spleen colonization and may be used to augment the immunogenicity of orally administered, live attenuated Salmonella vaccines. This represents a novel strategy for modulating host immune responses to strengthen Th1-biased immune responses and enhance protective immunity.

MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions. Bacterial strains and plasmids used in this study are listed in Table 4 and Table S1 in the supplemental material. Escherichia coli and Salmonella enterica serovar Typhimurium strains were routinely cultured at 37°C in LB broth (39) or on LB agar. Cultures of S. Typhimurium strain 9088 (22) and its derivatives were supplemented with 0.05% mannose (for Δpmi-2426) and 0.2% arabinose (for ΔP fur33 ::TT araC P BAD fur). Diaminopimelic acid (DAP) (50 g/ml) was added to LB medium for growing Δasd mutant strains. The following antibiotics were used as needed at the indicated concentrations: ampicillin, 100 g/ml; chloramphenicol, 15 g/ml; gentamicin, 20 g/ml; kanamycin, 50 g/ml; tetracycline, 10 g/ml. Carbohydrate-free nutrient broth (NB) was used for growth when determining lipopolysaccharide (LPS) profiles. LB agar without sodium chloride and with 7.5% sucrose was employed for sacB-based counterselection. MacConkey agar plates with 1% mannose were used to indicate sugar fermentation.
For animal experiments, S. Typhimurium strains were grown in LB broth with appropriate supplements. Overnight cultures were diluted 1:100 and grown with shaking (200 rpm) to an optical density at 600 nm of~0.8. Then, bacteria were centrifuged at 5,000 ϫ g for 15 min at room temperature and resuspended in phosphate-buffered saline (PBS) or buffered saline with 0.01% gelatin (BSG) (40). LB or Salmonella Shigella (SS) agar plates were used to enumerate S. Typhimurium recovered from tissues. Selenite cystine broth was employed to enrich samples for S. Typhimurium. Streptococcus pneumoniae WU2 was cultured on brain heart infusion agar containing 5% sheep blood or in Todd-Hewitt broth with 0.5% yeast extract (41). All media, antibiotics, and chemicals were purchased from BD Difco (Franklin Lakes, NJ) or Sigma-Aldrich (St. Louis, MO).
General DNA procedures. DNA manipulations, including plasmid and genomic DNA isolation, restriction enzyme digestions, ligations, and other DNA-modifying reactions, were conducted as described previously (42) or were performed according to the manufacturers' instructions (New England Biolabs, Ipswich, MA; Qiagen, Valencia, CA; Promega, Madison, WI). Synthesis of primers (Table S2)  Construction of transcriptional aph-lacZ fusions. DNA fragments containing the promoter regions of 12 fimbrial operons were amplified from the S. Typhimurium 3761 genome by PCR using the appropriate primers (Table S2). The PCR products were digested with ApaI and BamHI and cloned into the unique ApaI/BamHI sites of aph-lacZ fusion suicide vector pSG3 (43). The resulting plasmids were introduced by conjugation into S. Typhimurium strain 3761 to obtain fusions of selected promoter regions with aph-lacZ genes by a single-crossover event as previously described (43).
In vivo expression technology (IVET). Each S. Typhimurium aph-lacZ fusion strain was grown statically in LB broth at 37°C for 20 h. Bacterial cells were harvested by centrifugation at 5,000 ϫ g for 15 min at room temperature. The pellets were resuspended in BSG buffer. BALB/c mice were inoculated orally with~1 ϫ 10 9 CFU of the mixture of the 12 aph-lacZ fusion strains. Mice were treated with kanamycin at 3, 24, and 48 h postinoculation by oral administration (2 mg in 20 l) and intraperitoneal (10 mg in 100 l) injection. Three days after inoculation, the spleens were collected from the treated mice and homogenized. Dilutions of the homogenate were made in BSG and plated onto LB agar plates supplemented with gentamicin and incubated overnight at 37°C. Finally, selected clones were identified by PCR using specific primers (Table S2).
Construction of S. Typhimurium mutants. All S. Typhimurium mutants were derived from the highly virulent parent strain 3761 (45). The genealogy of constructed strains is shown in Table S1. All gene replacements were introduced by conjugational transfer of suicide plasmids using donor E. coli strain 7213 (46). All mutations were verified by PCR. We confirmed arabinose-regulated Fur production by Western blotting. The Δpmi mutation was confirmed by white colony phenotype on mannose-MacConkey agar. Lipopolysaccharide (LPS) profiles were examined by silver staining of 12% polyacrylamide gels as described previously (47).
SDS-PAGE and Western blotting. SDS-PAGE and Western blotting were performed by standard techniques. The blots were developed with nitroblue tetrazolium chloride/5-bromo-4-chloro-3=-indolyl phosphate (Amresco, Solon, OH) or Pierce ECL Western blotting substrate (Thermo Scientific), using rabbit polyclonal anti-rPspA serum as primary antibodies and mouse anti-rabbit IgG alkaline phosphatase conjugate (Sigma-Aldrich) as secondary antibodies.
Animal supply and housing. Female BALB/c mice (6 to 8 weeks old) were obtained from Charles River Laboratories (Wilmington, MA). Animals were allowed to acclimate for 1 week after arrival before starting the experiments. All animal procedures were carried out in compliance with the Institutional Animal Care and Use Committee (IACUC) at Arizona State University and the Animal Welfare Act.
Colonization of the mouse spleen and determination of the competitive index. BALB/c mice were inoculated intraperitoneally with a mixture containing~1 ϫ 10 4 of S. Typhimurium wild-type strain (3761) and either strain 11484 or strain 11599 suspended in 100 l of PBS. Wild-type and mutant strains were marked with low-copy-number chloramphenicol-or kanamycin-resistant plasmids: pHSG576 and pWSK129, respectively. On days 1 and 3 postinoculation, three mice in each group were euthanized, and the spleens and livers were collected to determine the colonization levels. The competitive index (CI) for each strain compared to the wild type was calculated by dividing the ratio of two strains from an organ divided by the same ratio in the suspension used for the infection.
Determination of the 50% lethal dose. Freshly grown bacterial cultures were pelleted by centrifugation at 5,000 ϫ g for 15 min at room temperature. Bacterial pellets were resuspended in BSG and adjusted to achieve a dose of~10 2 to~10 9 CFU in a volume of 20 l for orally inoculating BALB/c mice. Animals were observed for typhoid symptoms for 3 weeks postinoculation. Deaths were recorded daily. The 50% lethal dose (LD 50 ) was calculated using the Reed and Muench method (48).
Immunization and pneumococcal challenge. BALB/c mice were inoculated orally with 20 l of PBS containing~1 ϫ 10 8 CFU of the appropriate S. Typhimurium strain and boosted with the same strain and dose 6 weeks later. No food or water was provided for~4 h prior to immunizations. Groups of mice inoculated with PBS served as a control. At week 10 (i.e., 4 weeks after the booster), all mice were challenged by intraperitoneal injection with~1 ϫ 10 4 CFU of S. pneumoniae WU2 in 100 l of BSG (equivalent to 40 times the LD 50 ). Mice were monitored daily for 3 weeks.
Statistical analyses. All statistical analyses were performed using GraphPad Prism 6 (GraphPad Software, San Diego, CA). The significance of the different values obtained was appraised using two-way analysis of variance (ANOVA) followed by Dunnett's tests (for ELISA). For challenge experiments, log rank (Mantel-Cox) test was used to determine the significant differences between the survival curves. For CI assays, the geometric means of the CIs were determined, and a Student's t test was used to determine whether the logarithmically transformed ratios differed significantly from zero. P values of Ͻ0.05 were considered significant.