Identification of Three Highly Attenuated Salmonella typhimurium Mutants That Are More Immunogenic and Protective in Mice than a Prototypical aroA Mutant

Abstract

A panel of Salmonella typhimurium 14028s mutants, which were previously shown to be highly attenuated in the BALB/c mouse model of infection, were analyzed for their potential as live Salmonella oral-vaccine candidates. A prototypical aroA mutant was chosen as a basis of comparison. From the panel of mutants initially chosen for this study, three mutants with comparable levels of attenuation elicited higher Salmonella-specific serum immunoglobulin G (IgG) and/or mucosal secretory-IgA antibody titers than the aroA vaccine strain. The three mutants, CL288, CL401, and CL554, also elicited a better protective immune response than the aroA control strain, after a single oral dose of 1 × 10⁹ to 2 × 10⁹ bacteria.

Salmonella typhi is the causative agent of typhoid fever in humans. The disease caused by Salmonella typhimurium in mice mimics typhoid fever in humans and is well accepted as a model for human typhoid. The murine typhoid model has been invaluable in the search for better typhoid vaccines as well as potential carrier vaccines, which deliver heterologous antigens to the immune system of the host. It provides a rapid and affordable means to assess the potential efficacy of a vaccine for use in humans. Currently, a variety of attenuated S. typhimurium strains have been characterized that endow protective immunity in mice. The best examples of genetically defined S. typhimurium vaccine strains include aroA or aroCD, crp/cya, phoPQ, ompR, and htrA mutants (5, 6, 8, 9, 11, 16, 28). The information gained from studies with mice has permitted the rational design of S. typhi-derived vaccines for use in humans. An example of a licensed live oral vaccine for typhoid fever is S. typhi Ty21a (17, 25). However, S. typhi Ty21a is not genetically defined and has not shown the level of efficacy first observed in earlier controlled large-scale human trials (26). More recently developed and genetically defined S. typhi strains, which carry aro or crp/cya mutations, have made significant progress in clinical trials (22, 34, 35). Other mutations, such as phoP/Q and htrA, either singly or in combination with aro mutations have more recently been introduced into S. typhi and tested in humans for their safety and immunogenic properties and have shown promising results (20, 21, 36).

A live Salmonella vaccine for use in humans needs to meet certain criteria for acceptability. Safety is a key issue, but a fine balance exists between nonreactogenic and immunogenic characteristics of the strain in question (29, 32). It is also of importance to construct a vaccine strain that contains two genetically unlinked attenuating mutations in order to prevent any possibility of reversion to wild type. To date, the S. typhi aroCD (CVD906 or -908) or crp/cya (chi3927) typhoid vaccines have demonstrated that these restrictions can be met with certain success (34). Yet even at this point, it would be imprudent to preclude further searches, by either examining novel combinations of known mutations or examining untested mutations, for improved “carrier” vaccines or typhoid vaccines.

The fact that Salmonella pathogenesis is a very complex system that is easy to genetically manipulate has resulted in a large collection of S. typhimurium mutants unable to elaborate full virulence within the murine typhoid model (1, 3, 12–14, 18, 19, 24, 27). Only a small proportion of mutants from this existing collection have been thoroughly examined for their potential as vaccines. In our laboratory, we have previously described two independently derived groups of S. typhimurium 14028s transposon mutants. The first group of mutants, generated by Tn10 mutagenesis, was initially screened for macrophage sensitivity (12). These macrophage-sensitive (MS) mutants were then assessed for virulence in mice, and a strong correlation between virulence and the ability to survive within macrophages was found. The second group of mutants, generated by MudJ mutagenesis, was screened directly for avirulence in mice (3). In this report, we describe our efforts to uncover novel mutants that could be used to further the development of either typhoid vaccines or vaccine systems that heterologously express antigen(s) derived from another important pathogen(s). The panel of mutants were compared to a prototypical aroA strain for the ability to elicit humoral and mucosal immune responses to Salmonella antigens in mice. A Salmonella aroA background was chosen as the basis of comparison since this background has demonstrated the best combination of safety, immunogenicity, and ability to induce a protective immune response in a variety of animal models, as well as in humans (7, 31). A short list of mutants were characterized further with respect to the immune responses evoked and the ability to protect against a virulent challenge.

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions.

All of the Salmonella strains used in this study were derived from the wild-type S. typhimurium 14028s strain from the American Type Culture Collection. The mutants MS1592, MS3792, MS4290, and MS9187 were generated by Tn10 mutagenesis (1). The mutants CL79, CL98, CL238, CL287, CL288, CL401, CL448, CL500, and CL554 were generated by MudJ mutagenesis and are described and characterized by Bowe et al. (3). The prototypical aroA strain PV4569 (aroA) was previously described (37). Mice were immunized with 0.2 ml of stationary-phase bacteria that had been grown aerobically overnight at 37°C in Luria-Bertani (LB) broth and then concentrated threefold in fresh LB broth to approximate a bacterial concentration of 5 × 10⁹ to 10 × 10⁹ per ml.

Mouse experiments. (i) Immunizations, LD₅₀ determinations, and virulent challenge.

Female BALB/c ByJ mice (6 to 8 weeks old) obtained from Jackson Laboratories (Bar Harbor, Maine) and housed under specific-pathogen-free conditions were used for all experiments. For each animal experiment, a single immunizing dose of 1 × 10⁹ to 2 × 10⁹ bacteria in LB broth was administered by gavage to mice (day 0). Control mice were given an equal volume of sterile LB broth.

The oral- and intraperitoneal (i.p.)-route 50% lethal doses (LD₅₀s) for the wild-type strain are 2 × 10⁶ to 5 × 10⁶ and <10 CFU/ml, respectively. The oral- and i.p.-route LD₅₀s for each mutant are reported in the accompanying paper by Bowe et al. (3). The i.p. LD₅₀ experiment for CL401 was performed as previously described (12). Five groups of four mice were given various doses of CL401 grown overnight in LB broth. The i.p. LD₅₀ was calculated as described by Reed and Muench (30).

When indicated, control mice and immunized mice were subjected to oral challenge with virulent 14028s at doses of 10⁷ or 10⁸ bacteria. The mice were observed daily for a total of 30 days postchallenge.

(ii) Determination of bacterial counts in mouse organs.

Groups of four or five mice per time point were sacrificed. Internal organs (the most distal Peyer’s patch of the small intestine, mesenteric lymph node, spleen, and liver) were collected and homogenized in 10 ml of sterile 1× phosphate-buffered saline (PBS) by using a stomacher (Tekmar). Dilutions were plated on LB plates containing kanamycin sulfate (60 μg/ml).

(iii) Immunological assays.

Blood and fecal samples were collected on day 0 (preimmune) and on days 14, 28, 42, and 56. Blood samples were incubated overnight at 4°C followed by microcentrifugation for 1 min before the serum was collected. Fecal samples were weighed, and 1.0 ml of PBS plus 0.1% sodium azide was added per 100 mg of feces (23). The fecal pellets were dispersed by microtip sonication for 10 to 20 bursts at 50% duty using Sonifier 450 (Branson Ultrasonics, Danbury, Conn.). The fecal suspensions were subsequently pelleted in a microcentrifuge for 5 min, and the supernatants were transferred to new tubes and frozen until assays were performed. Serum and fecal samples were pooled before being assayed for antibody content.

Enzyme-linked immunosorbent assays (ELISAs) were performed with Corning 96-well ELISA plates (catalog no. 25801) that were coated with Salmonella antigen (12.5 μg/well in a total volume of 0.05 ml). A stationary-phase culture of S. typhimurium 14028s (500 ml) was centrifuged and washed with 50 ml of 1× PBS. The washed cell pellet was resuspended in 5 ml of 1× PBS and sonicated (microtip) three times for 10 min at 50% duty while incubated in an ice water bath (Branson Ultrasonics). Cell debris was pelleted by low-speed centrifugation. The cleared cell lysate was adjusted to give 25 mg of protein per ml by using the Bradford assay kit (Bio-Rad) and frozen away in 0.5-ml aliquots. Antigen-coated plates were blocked with 0.2 ml of 3% BLOTTO (3% powdered skim milk, 0.04% anti-foam A, 0.05% Tween 20, and 0.1% sodium azide in 1× PBS) for 2 to 3 h at 37°C. Blocked plates were then washed once with H₂O and frozen if not used the same day. Samples (0.05 ml/well) were added in duplicate twofold serial dilutions with 3% BLOTTO as the diluent. The plates were incubated for 2 to 4 h at 37°C and then washed eight times with H₂O. The following alkaline phosphatase (AP)-conjugated antibodies were diluted 1:3,000 in 3% BLOTTO (0.05 ml/well): goat anti-mouse immunoglobulin G (IgG)-AP (Sigma product A-3562), goat anti-mouse IgA-AP (Sigma product A-4937), rat anti-mouse IgG1 (Pharmingen product 02003E), and rat anti-mouse IgG2a (Pharmingen product 02013E). Secondary antibodies were incubated for 2 to 4 h at 37°C and then washed eight times with H₂O. Detection was performed with 0.05 ml of p-nitrophenylphosphate (1 mg/ml; Sigma product N-2765) diluted in glycine buffer (0.1 M glycine, 1 mM MgCl₂, 1 mM ZnCl₂ [pH 10.4]). Reactions were stopped after a 60-min incubation at 25°C by addition of 0.05 ml of 0.1 M EDTA. An automated ELISA plate reader (MR700 microplate reader; Dynatech Laboratories, Inc., Chantilly, Va.) was used to measure A₄₀₅. Titers are expressed as the inverse of the highest dilution that gave an absorbance value greater than 0.1.

Delayed-type hypersensitivity (DTH) tests were done on groups of five mice by injecting Salmonella antigen into the ear pinna and measuring the degree of swelling at 24, 48, and 72 h with a spring-loaded-dial hand gauge (L. S. Starrett Co. [Athol, Mass.] product 1015MAZ). On day 65 postimmunization, control mice (unimmunized) and immunized mice were anesthetized before receiving 0.02 ml of sterile 1× PBS (right ear) and 2 × 10⁷ heat-killed Salmonella bacteria in a volume of 0.02 ml (left ear) with a 30-gauge needle. The values were determined by the following method. The left-pinna and right-pinna thicknesses were measured prior to sample injection to obtain a baseline value for the difference in thickness. On days 1, 2, and 3, the left and right pinnae were measured and the difference between them was calculated. The magnitude of swelling was calculated as the difference between day 1, 2, or 3 and day 0, with the values expressed in 10⁻² mm.

RESULTS

Preliminary screen for potential vaccine candidates.

From two transposon-generated banks of mutants, we chose 14 mutants to examine in a small-scale animal experiment. Our intention was to further characterize only those mutants that elicited a stronger Salmonella-specific immune response than that elicited by PV4569 (aroA). This rationale is based on the assumption that only strains which demonstrate stronger immunogenic characteristics could elicit a more protective immune response in mice. For the preliminary screen, each mutant was orally administered to two mice at a dose of 0.5 × 10⁹ to 2.1 × 10⁹ bacteria. By day 5, some groups of mice began to show signs of illness. S. typhimurium mutants which killed or caused obvious outward symptoms of disease at this dose (MS1592, MS3792, MS4290, MS9187, CL98, CL238, and CL287) were dropped from consideration without further analysis. The six groups of mice which received CL79, CL288, CL401, CL448, CL500, or CL554 showed no signs of disease. In these six groups, serum and fecal material were collected from the surviving mice after 4 weeks and Salmonella-specific antibody titers were measured (Table 1).

TABLE 1.

Preliminary characteristics of the murine immune response to S. typhimurium mutants

Strain	Titera
	Serum			Mucosal sIgA
	IgG	IgG1	IgG2a
PV4569	4,000	<1,000	4,000	160
CL79	8,500	<1,000	8,500	20
CL288	16,000	<1,000	16,000	288
CL401	160,000	2,500	160,000	24
CL448	34,000	<1,000	32,000	40
CL500	40,000	<1,000	40,000	160
CL554	16,000	<1,000	16,000	288

^aInverse of the highest dilution resulting in an A₄₀₅ of at least 0.1 after a 60-min incubation at 25°C. The data are arithmetic means for groups that had two healthy mice after immunization. 

The mutant CL401 elicited approximately 40-fold-higher levels of Salmonella-specific IgG than PV4569 (aroA). However, a similar increase in mucosal secretory IgA (sIgA) was not observed. The Salmonella-specific mucosal sIgA response actually decreased in comparison to the response elicited by PV4569 (aroA). Mutants CL288 and CL554 elicited Salmonella-specific IgG and mucosal sIgA titers that were higher than those elicited by PV4569 (aroA). The immune response elicited by CL79, CL448, or CL500 resembled that elicited by CL401. However, these three mutants elicited intermediate levels of Salmonella-specific IgG compared to PV4569 (aroA) and CL401.

An indirect measure of a Th1 or a cellular immune response is the differential between the IgG subclasses IgG1 and IgG2a, with the level of IgG2a significantly higher than that of IgG1 (33). As evidenced by the ratio of serum IgG1 to IgG2a in Table 1, all of the strains induced a systemic Th1 response. Overall, the observed differences in IgG and IgA antibody titers in each group of mice may not be significant, since small numbers of animals were used in the preliminary screen. Nevertheless, based on the preliminary results, CL401, CL448, CL288, and CL554 were chosen for further study.

CL288 and CL554 were included because of the potentially stronger mucosal sIgA response compared to PV4569 (aroA), while CL401 was included because of the significantly higher serum IgG response. CL448 was comparable to PV4569 (aroA) and was retained for further examination because of a membrane defect caused by the disruption of tolB (3, 15). This characteristic may be useful for a carrier strain when it is necessary to secrete a factor or antigen through both membranes of Salmonella. For example, secretion of an immune effector molecule such as interleukin-4, to the extracellular mileu, may require only a cleavable signal sequence for the general secretory pathway.

Systemic IgG and mucosal sIgA antibody responses.

To better assess the magnitude and kinetics of the Salmonella-specific immunity, the four S. typhimurium mutants CL288, CL401, CL448, and CL554 were compared to PV4569 (aroA) in a larger-scale time course experiment. Over the course of 56 days, it was apparent that the strains could be placed into two groups that differed in the magnitude of Salmonella-specific IgG elicited in mice (Fig. 1A). The mutants CL288, CL401, and CL554 elicited a 10-fold-higher titer of Salmonella-specific serum IgG antibody than the other two strains, PV4569 (aroA) and CL448. Interestingly, mice immunized with CL554 and CL288 never showed a decline in Salmonella-specific IgG titers through 56 days, unlike mice that received CL401, in which the Salmonella-specific IgG titers began to decline by day 42. The highest observed titer in each group still indicated that CL401 elicited the highest IgG titers, approximately 2-fold higher than those elicited by CL288 and CL554 and 15-fold higher than that elicited by PV4569 (aroA).

FIG. 1.

Salmonella-specific antibody levels in mice over time. Groups of 10 to 12 mice were immunized with the indicated strains. (A) For Salmonella-specific serum IgG, serum samples were pooled before being subjected to ELISA. Serum was prepared as described in Materials and Methods. Serum samples from control mice, which were given LB broth only, had no detectable Salmonella-specific IgG levels. (B) For Salmonella-specific mucosal IgA, fecal wash solutions were pooled before being subjected to ELISA.

The IgA responses by day 28 appeared to reflect the results observed for serum IgG responses (Fig. 1B). However, the levels of Salmonella-specific mucosal sIgA in mice that were given CL401 decreased to the levels observed in mice that received PV4569 (aroA) or CL448 by day 56. This result was in accordance with the preliminary screen in which the IgA titer in CL401-immunized mice was actually lower than the level in mice immunized with PV4569 (aroA). CL288 and CL554 elicited approximately 10-fold-higher Salmonella-specific IgA titers than PV4569 (aroA), and the level of Salmonella-specific sIgA continued to increase up to day 56.

Cellular response.

As mentioned earlier, the four CL mutants appear to induce a systemic Th1 response, as evidenced by the levels of Salmonella-specific IgG2a and IgG1 subclasses of IgG. To assess the relative magnitude of the cellular immune responses induced by the CL mutants compared to PV4569 (aroA), DTH assays were performed for mice in each group. To exclude the possibility of an Arthus reaction contributing to the degree of swelling, we reported only the data that was compiled 72 h after administration of antigen (10). In support of the serum data, the results summarized in Table 2 indicate that all of the strains were able to elicit a cellular response against Salmonella antigen compared to the control group immunized with LB broth only. Statistically significant differences between means were determined by an unpaired t test using a hypothesized difference of 0 and showed that CL288 appeared to elicit a better cellular response (P < 0.05) than the other three CL strains and PV4569 (aroA). CL401, CL448, and CL554 were not significantly different in their abilities to induce a DTH response in mice compared to PV4569 (aroA).

TABLE 2.

DTH responses to Salmonella antigens

Immunizing strain	Δ(L − R)a (10−2 mm)
None	5 ± 1.6
PV4569	33 ± 2.9
CL288	53 ± 5.5
CL401	45 ± 5.6
CL448	29 ± 6.1
CL554	41 ± 3.4

^aΔ(L − R), change in the difference between left- and right-pinna values at 72 h from the difference at time zero (see Materials and Methods). 

Protective immunity.

Based on the immunological data, it was predicted that the mice given CL288, CL401, or CL554 may show a more protective immune response against a lethal Salmonella challenge than the mice given the aroA strain or CL448. To test this hypothesis, each group of mice were split into two subgroups, with each subgroup receiving either 10⁷ or 10⁸ CFU of virulent Salmonella. As shown in Table 3, mice that received PV4569 (aroA) were only partially protected against a virulent challenge of either 10⁷ or 10⁸ bacteria. Not surprisingly, CL448 was less protective than PV4569 (aroA) at either challenge dose. Given these results, CL448 was dropped from further consideration. As expected, all mice immunized with sterile LB broth died within 8 days postchallenge. The other CL mutants appeared to elicit a better protective immune response than PV4569 (aroA). CL288-, CL401-, or CL554-immunized mice were completely protected against a challenge of 10⁷ virulent bacteria and showed at least 50% survival when challenged with 10⁸ virulent bacteria.

TABLE 3.

Protection against virulent 14028s challenge

Immunizing-strain CFUa	No. of survivorsb
None
107	0/4
108	0/4
PV4569
107	4/6
108	2/6
CL288
107	5/5
108	3/6
CL401
107	5/5
108	4/5
CL448
107	3/6
108	1/6
CL554
107	6/6
108	4/6

^aGroups of mice were immunized with one of the listed Salmonella mutants (day 0). On day 70, each group was divided into two groups in order to test two different challenge doses. The actual doses administered were 9.0 × 10⁶ and 1.2 × 10⁸ CFU. 

^bThe animals were monitored for 28 days postchallenge. 

Persistence in mouse organs.

The ability to colonize and persist in mouse tissues was examined to clarify the better immune response elicited by CL288, CL401, and CL554 and the continuous increase of antibody levels in mice immunized with CL288 or CL554 compared to PV4569 (aroA). One possible explanation for the continual increases in titers of antibodies against CL288 and CL554 is that a persistent infection was occurring in the mice, which resulted in continuous stimulation of the immune system with antigen.

All three CL strains were found at higher numbers in the spleen and liver in infected mice compared to PV4569 (aroA) (Fig. 2A). This was especially evident between days 3 and 21. CL401, but not CL288 or CL554, maintained relatively high numbers in the spleen throughout the experiment and was also detectable in the liver after 6 weeks, yet the serum IgG and, to a greater extent, the mucosal sIgA levels declined during the same period.

FIG. 2.

Persistence of Salmonella strains within different tissues of the mouse. Mice were infected as described in Materials and Methods. At the specified times, mice were euthanized by CO₂ asphyxiation and then the tissues were aseptically removed and processed for bacterial quantitation. (A) CFU per spleen and liver; (B) CFU per distal Peyer’s patch of the small intestine and mesenteric lymph node. The data are arithmetic means for three to five animals with standard errors of the means.

In contrast to what was observed in the spleen and liver, CL401 was not detected in Peyer’s patch tissue (Fig. 2B). This result, as far as we know, is unique among attenuated Salmonella strains reported to date. The other CL strains had patterns of persistence similar to that of PV4569 (aroA) in Peyer’s patch tissue. Although CL401 was not detected in the most distal Peyer’s patch at any time, the viable counts within the mesenteric lymph nodes were not significantly altered compared to the other three strains. In fact, only CL401 was still detectable in mesenteric lymph nodes by day 42. Nevertheless, there appears to be a defect in the ability to invade or persist within Peyer’s patch tissue. If, in fact, CL401 was unable to invade Peyer’s patches, then it still should be virulent by the i.p. route. However, earlier studies showed that CL401 was 100- to 1,000-fold attenuated when administered i.p. (3). The LD₅₀ for CL401 administered i.p. was confirmed in this study to be 8 × 10² CFU compared to <10 CFU for 14028s. Thus, CL401 is more likely to be affected in the ability to persist in Peyer’s patch tissue.

Sequence analysis.

Chromosomal DNA flanking the CL288, CL401, and CL554 MudJ insertion was obtained earlier by inverse PCR (3). The DNA sequence adjacent to the CL288 MudJ indicated that the interrupted gene was the homolog of the Escherichia coli mdoB gene (4). The DNA sequences from CL401 and CL554 were identified as homologs to an o660 open reading frame (ORF) and to a putative malate oxidoreductase of E. coli, respectively (2, 38).

DISCUSSION

We have identified three Salmonella mutants that are more immunogenic than the prototypical aroA background used as a basis of comparison. These mutants were CL288, CL401, and CL554. The mutant CL288 carried MudJ in the mdoB gene encoding an osmoregulated transferase that adds phosphoglycerol to periplasmic oligosaccharides (3). The mutant CL401 contains MudJ in an ORF that is homologous to an E. coli ORF designated o660. The putative polypeptide encoded by o660 in E. coli shows 30% overall homology to the methionyl-tRNA formyltransferase that is encoded by the fmt gene of E. coli. Finally, the mutant CL554 contains MudJ within the gene encoding the putative malate oxidoreductase (NAD linked) which is involved in gluconeogenesis. Specifically, the oxidoreductase is involved in a reversible decarboxylation reaction where malate is converted to pyruvate. The oxidoreductase can also recognize oxaloacetate as a substrate. The genes that were inactivated in each of the mutants are not unique to Salmonella and can be found in E. coli. It is unclear what role the respective loci play in CL288 and CL401 regarding Salmonella pathogenesis. The putative role for the mutated gene in CL554 indicates a metabolic defect in vivo and suggests that four carbon molecules such as malate and/or oxaloacetate may be important for growth in the murine host.

For all three immunological parameters examined in the mice (humoral, mucosal, and cellular immunity), the CL mutants were able to elicit Salmonella-specific immune responses that were as good as or better than PV4569 (aroA). Thus, the high level of attenuation did not diminish the immunogenic characteristics compared to the aroA control strain. There was no significant difference in the overall pattern of immunity elicited by CL288, CL401, and CL554 compared to PV4569 (aroA) in that both Salmonella-specific IgG and sIgA levels were induced in all cases. Quantitative differences of antibody titers were generally seen as early as the first time point, day 14. Mice infected with CL288, CL401, or CL554 had 10-fold-higher levels of Salmonella-specific IgG, while CL448-infected mice had IgG titers similar to those of mice infected with PV4569 (aroA). The same quantitative differences were noted for Salmonella-specific mucosal sIgA titers, except for mice immunized with CL401 that had very high levels at first, which declined by day 56 to the level observed in PV4569 (aroA)-immunized mice. The rapid decline in Salmonella-specific sIgA levels in CL401-immunized mice may be a consequence of CL401 not persisting in Peyer’s patch tissue, since the Peyer’s patch region plays a major role in the induction of mucosal immunity. Cellular immunity as determined by DTH assays showed no significant difference, with the possible exception of CL288-infected mice. However, due to the imprecise nature of the assay, further experiments to conclusively demonstrate whether CL288 can elicit stronger cellular responses compared to PV4569 (aroA) are warranted.

Thus, CL288, CL401, and CL554 were sufficiently attenuated without significant diminution in the immunogenic characteristics. In addition, these mutants appear to elicit a more effective Salmonella-specific immune response in the mice than PV4569 (aroA). This could be due to differences in the antigen profile caused by the unique mutations of each strain. Since this may be the case, this study could not provide insight as to whether these CL mutants are better than PV4569 (aroA) as carriers of heterologous antigens. A second explanation for the improved protective immune response is that the three CL mutants were able to attain higher bacterial numbers in the murine tissues as well as persist for longer periods. This would result in a more intense as well as a more prolonged period of stimulation with Salmonella antigen. The question of whether similar immune responses would be evoked against a heterologous antigen expressed in CL288, CL401, or CL554 compared to PV4569 (aroA) is currently under investigation.

Salmonella vaccine systems have considerable potential as mucosal vaccines. This is due to the fact that Salmonella specifically targets the gut-associated lymphoid tissue, which serves as a major site of induction of specific immunity. Presently, there are at least three promising vaccine strain backgrounds, aroCD, cya/crp, and aroA/htrA, undergoing human trials. This degree of progress does not preclude the search for more improved Salmonella vaccine systems. New combinations of mutations may result in improved vaccines. Except for aroA/htrA, each double mutant mentioned above contains two mutations that affect the same metabolic pathway(s), thereby preventing a more severe synergistic attenuating effect. This approach helps guard against excessive attenuation of a vaccine strain that usually results in unacceptably weak immunogenic characteristics (29, 32). In the end, the advantage of having a variety of vaccine strains to choose from is that each may have capabilities of eliciting immune responses with unique patterns. Therefore, one could tailor an immune response by using an appropriate vaccine targeted against mucosal pathogens or other transmissible diseases which require systemic as well as mucosal immune responses to prevent disease. This information will contribute to the advancement of efficacious typhoid vaccines as well as Salmonella carrier vaccines that target other clinically important pathogens.