Accessibility of O Antigens Shared between Salmonella Serovars Determines Antibody-Mediated Cross-Protection

Pathogenic Salmonellae are a cause of high morbidity and mortality in many countries around the world. The clinical outcomes produced by these pathogens include systemic infection typhoid and paratyphoid, localized gastroenteritis, and invasive nontyphoidal Salmonella (iNTS) disease (1). Typhoid produced by Salmonella enterica serotype Typhi (Salmonella Typhi) accounts for 27 million cases with 217,000 deaths annually (2). iNTS disease caused by Salmonella Typhimurium and Salmonella Enteritidis in immunocompromised individuals has an annual global burden of around 90 million cases with 155,000 deaths annually (3). Currently, there are no vaccines available for human use against iNTS disease (4). The genomes of typhoidal and nontyphoidal Salmonella serovars share a high degree of homology (5, 6), despite which these serovars produce different clinical manifestations in humans and exhibit extreme host specificity (7). S. Typhi produces typhoid almost exclusively in humans. In contrast, S. Typhimurium, which causes only self-limiting localized gastroenteritis in normal human subjects, produces a systemic infection in mice that is analogous to human typhoid. This experimental model of Salmonella infection has been extensively used to understand human typhoid (8, 9). Studies carried out in this model have shown that both Abs and cell-mediated immune responses are required for optimal immunity against Salmonella (10–14). The degree of similarity between different Salmonella serovars particularly in cell surface–associated molecules (15, 16), which are targets of Abs, suggests that the immune responses elicited upon infection or immunization with one Salmonella serovar might be cross-reactive and capable of imparting immunity against related serovars. However, there have not been any detailed investigations on the nature of such cross-reactive responses and their relevance to immunity. Studies (not very conclusive) in humans with currently available S. Typhi–derived live oral vaccine Ty21a have not indicated that this vaccine can provide protection against closely related serovars such as Salmonella Paratyphi, S. Typhimurium, or S. Enteritidis, the reasons for which have not been elucidated.

In the current study, we show that immunization of mice with S. Typhi generates T and B cell responses that exhibit differential capabilities to provide protection against two closely related nontyphoidal Salmonella serovars. Mice immunized with S. Typhi did not produce Abs capable of binding live intact S. Typhimurium, yet cross-reactive T cells significantly prevented S. Typhimurium replication in the spleen in a mouse model of infection but could not prevent death of infected animals. In contrast, immunization with S. Typhi not only brought about better bacterial clearance but also enabled survival of animals in a model of S. Enteritidis infection. The latter was mainly due to the presence of Abs that bound live intact S. Enteritidis. This cross-reactivity was dependent on the expression of accessible oligosaccharide (O) Ags that S. Typhi shares with S. Enteritidis. These results suggest that surface-accessible O Ag–dependent determinants may be major targets of protective anti-Salmonella Abs. These findings have relevance in designing effective vaccines against infections with Salmonellae.

Wild-type C57BL/6 mice were obtained from Jackson Laboratory (Bar Harbor, ME) and maintained in the small animal facility of the National Institute of Immunology. Animal experiments were carried out according to the guidelines provided by the institutional animal ethics committee.

The human cervical epithelial cell line Hela was obtained from the American Type Culture Collection. Salmonella enterica serovar Typhimurium SL1344 (S. Typhimurium) was provided by Emmanuelle Charpentier, University of Vienna, Vienna, Austria (now at the Max Planck Institute of Infection Biology, Berlin, Germany); S. Typhi (Vi-negative isolate) was provided by G. Mehta (Lady Hardinge Medical College, New Delhi, India); S. Paratyphi A was provided by A. Kapil, All India Institute of Medical Sciences, New Delhi, India; and S. Enteritidis was obtained from M. Suar, Kalinga Institute of Industrial Technology, Bhubaneswar, India. LPS from S. Typhi, S. Typhimurium, and S. Enteritidis were obtained from Sigma-Aldrich. mAbs against O9 and O12 have been described previously (17).

Mouse peritoneal macrophages were grown in RPMI-1640 supplemented with 10% heat-inactivated FBS (RPMI-10) in a humidified atmosphere at 37°C with 5% CO₂. Hela was cultured in DMEM supplemented with 10% FBS (DMEM-10). Bacteria were grown in Luria Bertani broth at 37°C with shaking (220 rpm).

Bacteria grown in Luria Bertani medium were pelleted by centrifugation at 8000 × g for 5 min. The pellet was washed three times with PBS, resuspended in ice-cold PBS and sonicated on ice with brief pulses of sonication and cooling, 1 min each. This cycle was repeated five times. The sonicate was centrifuged at 10,000 × g for 20 min at 4°C. The supernatant was filtered through a 0.22-μ membrane and used in immunoassays and cellular assays.

Six- to eight-week-old C57BL/6 mice were immunized i.p. with 5 × 10⁶ live Vi-negative S. Typhi, or 50 × 10⁶ gentamicin-treated S. Typhimurium or S. Enteritidis. Sera were collected on days 10, 21, and 31 and analyzed for reactivity by ELISA with LPS and sonicates prepared from S. Typhi, S. Typhimurium, S. Enteritidis, and S. Paratyphi A. The profile of reactivity with different bacterial Ags was analyzed by Western blotting. Mice were also immunized i.p. with LPS derived from S. Typhi or S. Typhimurium (obtained from Sigma Chemical), and Abs were analyzed in sera 7 d after immunization.

The interaction of Abs with live bacteria was studied by flow cytometry. Briefly, 10⁷ bacteria were incubated with preimmune or immune sera at different dilutions for 1 h at 4°C, washed three times with PBS and incubated with PE-conjugated anti-mouse Ig Ab for 40 min (BD Biosciences, San Jose, CA). Control bacteria were only incubated with PE-conjugated anti-mouse Ig Ab. Bacteria were washed, fixed with 4% paraformaldehyde for 10 min, and analyzed by flow cytometry (FACSCalibur; Becton Dickinson Immunocytometry Systems, San Jose, CA). Data were plotted using Flowjo software.

Splenocytes from infected mice were plated at a density of 0.3 × 10⁶ cells per well in a 96-well cell culture plate and stimulated with bacterial sonicates in RPMI-10 supplemented with gentamicin (100 μg/ml). Culture supernatants were collected 32 h poststimulation and analyzed for IL-2 and IFN-γ by ELISA. Cytokine ELISAs were performed per the manufacturer’s instructions.

HeLa and mouse peritoneal macrophages were seeded at a density of 0.2 × 10⁶ cells per well in a 24-well plate. Bacteria were preincubated for 30 min with heat-inactivated preimmune or immune serum, and infection of cells was carried out at different multiplicities of infection for 1 h at 37°C. Cells were washed to get rid of extracellular bacteria and cultured in RPMI-10 (DMEM-10 for Hela) supplemented with gentamicin (100 μg/ml). One hour and 24 h later, cells were washed to get rid of gentamicin, lysed with 0.2% Triton-X 100 (Sigma-Aldrich), and the intracellular bacterial load was quantified by plating cell lysates on Salmonella-Shigella (SS) agar plates. CFU were enumerated after overnight incubation at 37°C.

Six- to eight-week-old C57BL/6 mice were immunized with 5 × 10⁶ CFU of live Vi-negative S. Typhi, and 5 wk later, immune and nonimmune mice were challenged i.p. with 100 and 200 CFU of S. Typhimurium and S. Enteritidis, respectively. Mice were sacrificed on day 3 postinfection, and bacterial burden was determined in the spleens by plating serial dilutions of tissue lysates on SS agar. In some experiments, mice were infected with Salmonella that had been preincubated with anti–S. Typhi serum, and the splenic bacterial load was determined 3 d postinfection.

Mice immunized with live S. Typhi and challenged i.p. with S. Typhimurium or S. Enteritidis were observed for morbidity and mortality for 30 d.

The results are expressed as mean ± SEM. Data were analyzed by Student t test using GraphPad Prism, San Diego, CA. The survival data were analyzed using log-rank (Mantel–Cox) test. The results were considered statistically significant when a p value of <0.05 was obtained. In some experiments, p values were calculated by two-tailed test for unequal variances using Student t test in Microsoft Excel. Error bars represent SEM. The p values found are as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

To investigate whether cross-reactive anti-Salmonella immune responses elicited with one Salmonella serovar can provide protection against closely related serovars, we immunized mice with live capsule-negative S. Typhi and challenged them with S. Typhimurium and S. Enteritidis. There was an ∼60% reduction in splenic bacterial load in S. Typhi–immunized mice challenged with S. Typhimurium (Fig. 1A). However, these mice did not show significant delay in death when compared with unimmunized mice (Fig. 1B). In contrast, when challenged with S. Enteritidis, S. Typhi–immunized mice not only showed significantly greater reduction in tissue bacterial load but also survived for a long duration (Fig. 1C, 1D). These results demonstrated that anti-Salmonella immune responses generated with S. Typhi have a differential ability to protect against S. Typhimurium and S. Enteritidis.

To understand the reasons for reduced capability of S. Typhi–immunized mice to survive challenge with S. Typhimurium, B and T cell reactivities of these mice with Ags of S. Typhimurium were analyzed ex vivo. Abs obtained from these mice on days 10, 21, and 31 postinfection showed a high degree of cross-reactivity in ELISA with Ags extracted from S. Typhimurium (Fig. 2A). These Abs recognized a large number of Ags ranging from 10 to 200 KDa present in S. Typhi as well as S. Typhimurium, indicating that Abs were largely directed against conserved antigenic determinants shared by these two Salmonella serovars (Fig. 2B). Splenocytes from S. Typhi–immunized mice secreted comparable levels of IL-2 and IFN-γ upon ex vivo stimulation with Ags derived from S. Typhi, S. Typhimurium, and S. Enteritidis, revealing cross-reactivity at the level of T cell recognition as well (Fig. 2C). As protection by Abs is mediated through their interaction with live bacteria, we next analyzed binding of anti–S. Typhi Abs with bacteria by flow cytometry. This analysis showed dose-dependent binding of anti–S. Typhi Abs with live S. Typhi (Fig. 2D; similar binding profile was obtained with S. Typhi treated with antibiotics for 1 h, data not shown; live or antibiotic-treated bacteria have been described as “intact bacteria” throughout this article). Remarkably, however, these Abs showed negligible reactivity with intact S. Typhimurium, indicating that Abs capable of binding intact S. Typhi were highly serotype specific (Fig. 2D, Supplemental Fig. 1). This specificity was retained even after mice were boosted with live S. Typhi (Supplemental Fig. 2). The identity of surface-associated antigenic determinants recognized by Abs was investigated by preincubating sera with LPS and flagellin, two major cell surface Ags of Salmonella, before incubating with live bacteria. Preincubation of immune sera with S. Typhi–derived LPS but not with S. Typhimurium–derived LPS significantly reduced the binding of these Abs with intact bacteria, suggesting that a large proportion of surface reactive Abs might be directed against S. Typhi–specific determinants of LPS (Fig. 2E). Preincubation with flagellin did not significantly reduce the binding of Abs to S. Typhi, indicating a poor Ab response against flagellin (Fig. 2E). Anti–S. Typhi Abs, however, showed considerable reactivity with S. Typhimurium–derived LPS in ELISA and Western blotting (Fig. 2F, 2G). Their reactivity with LPS from Salmonella Minnesota, which does not share any O-antigenic determinants with S. Typhi or S. Typhimurium, was very poor (Supplemental Fig. 3). These results suggested that the O determinants on LPS that anti–S. Typhi Abs bound to on the surface of bacteria were either exclusively present in S. Typhi and/or accessible only on S. Typhi.

S. Typhi belongs to serotype D and its LPS carries O-antigenic determinants 9 and 12 whereas S. Typhimurium belongs to serotype B with its LPS carrying 1, 4, 5, and 12 O-antigenic determinants (18). The inability of anti–S. Typhi LPS Abs to interact with intact S. Typhimurium despite binding S. Typhimurium–derived LPS in ELISA suggested that these Abs may be specifically directed against O9 present in S. Typhi but not in S. Typhimurium, and that O12-antigenic determinant(s) shared by S. Typhi and S. Typhimurium may not be accessible on the surface of these two Salmonella serovars. We reasoned that accessibility of O9 could be one of the major reasons why S. Typhi–immunized mice were protected against challenge with S. Enteritidis because latter also belongs to serotype D and carries O-antigenic determinants 1, 9, and 12. Analysis with intact bacteria indeed showed significant reactivity of anti–S. Typhi Abs with S. Enteritidis (Fig. 3A). Importantly and consistent with this binding profile, mice infected with S. Enteritidis that had been preincubated with serum from S. Typhi–immunized mice showed reduced bacterial burden in the spleen that was associated with delay in death of infected animals (Fig. 3B). This reduction in bacterial load accompanied by delayed mortality was not seen in mice infected with S. Typhimurium preincubated with anti–S. Typhi serum (Fig. 3C). We also tested the binding of anti–S. Typhi Abs with S. Paratyphi A, which belongs to serotype A carrying O-antigenic determinants 1, 2, and 12, and found no detectable binding (Fig. 3A). These data established that O9 determinant was accessible for binding anti–S. Typhi Abs. This reactivity pattern was recapitulated with mAbs generated previously in our laboratory (17). Anti-O9 mAb bound S. Typhi and S. Enteritidis but not S. Typhimurium, whereas anti-O12 mAb did not bind intact bacteria of either serovar despite the fact that it reacted with LPS from all three serovars (Fig. 3D, 3E).

The phenomenon of selective accessibility of serogroup-restricted O-antigenic determinants was also observed with Abs generated against S. Typhimurium and S. Enteritidis. Abs obtained from S. Typhimurium–immunized mice did not show any detectable interaction with intact S. Typhi, S. Enteritidis, and S. Paratyphi A (Fig. 3F, Supplemental Fig. 4). The lack of reactivity with intact S. Typhi was not due to absence of LPS–cross-reactive Abs (Supplemental Fig. 4). Anti–S. Enteritidis Abs bound S. Typhi but did not interact with S. Typhimurium and S. Paratyphi A (Fig. 3F). Abs generated against any one of these serovars did, however, react in ELISA with Ags extracted from all four serovars (Fig. 3G).

The ability of anti-O Abs to bring about clearance of Salmonella in a serovar-specific fashion was also evaluated in mice immunized with LPS. As expected, the Ab response to LPS was low in magnitude consistent with previously established poor immunogenicity of O Ags (11), and it was predominantly IgM (Fig. 4A). However, the reactivity pattern of Abs from these mice with different Salmonella serovars was similar to the one seen with anti-O Abs generated upon immunization with live Salmonella (Figs. 3A, 4B). Anti–S. Typhi LPS Abs bound S. Typhi but did not show any detectable binding with S. Typhimurium (Fig. 4B). These also bound to S. Enteritidis because of the sharing of O9-antigenic determinants with S. Typhi, although the reactivity was several fold lower than that seen with S. Typhi (Fig. 4B). In contrast, anti–S. Typhimurium LPS Abs bound S. Typhimurium but not S. Typhi or S. Enteritidis (Fig. 4B). Consistent with this binding pattern, mice immunized with S. Typhi LPS showed a reduction in splenic bacterial load when challenged with S. Typhi (Fig. 4C). This reduction was not readily observed when these mice were challenged with S. Typhimurium (Fig. 4C). Similarly, mice immunized with S. Typhimurium LPS showed a reduction in splenic bacterial load when challenged with S. Typhimurium, which was not readily seen when these were challenged with S. Typhi (Fig. 4C). Importantly and in line with the results obtained with mice immunized with live S. Typhi, mice immunized with S. Typhi LPS also showed a reduction in splenic bacterial burden when these were challenged with S. Enteritidis (Figs. 1A, 4C). This reduction was not seen when S. Typhimurium LPS–immunized mice were challenged with S. Enteritidis (Fig. 4C). The degree of reduction in splenic S. Enteritidis in mice immunized with S. Typhi LPS was lesser than the one produced by anti-O Abs in mice immunized with live S. Typhi, which was due to considerable differences in the magnitude and quality of anti-O Ab response produced with the two immunization regimens (Fig. 4A).

The selective accessibility of serovar-restricted O-antigenic determinants for binding surface reactive anti-Salmonella Abs that might be involved in protection raised the possibility that these determinants may participate in a host–pathogen interaction. We therefore tested if Abs against these determinants could interfere with invasion of epithelial cells with Salmonella. Anti–S. Typhi Abs inhibited invasion of model epithelial cell line, Hela, with S. Typhi in a dose-dependent manner (Fig. 5A). Interestingly, at lower Ab concentration, the inhibition in invasion was not readily observed, yet Ab-decorated bacteria were cleared better intracellularly, as seen by the bacterial load at 24 h postinfection (Fig. 5A). As the induction of inflammatory responses with Salmonella in Hela cells is completely dependent on bacterial invasion, the inhibition of invasion with Abs also resulted in reduced cytokine secretion from infected cells (Fig. 5B). Consistent with their binding specificities, anti–S. Typhi Abs also partially inhibited the invasion of cells with S. Enteritidis but did not prevent invasion of cells with S. Typhimurium (Fig. 5C, 5D). The ability to inhibit invasion was reduced if Abs were preincubated with LPS derived from S. Typhi, suggesting that anti-LPS Abs directed against O9-antigenic determinant(s) were responsible for inhibiting invasion (Fig. 5E). In line with these results, the invasion of Hela with S. Typhi but not with S. Typhimurium was also inhibited specifically with anti-O9 mAb (Fig. 5F).

We also investigated the effect of anti–S. Typhi Abs on bacterial phagocytosis. S. Typhi–immune sera increased phagocytosis of S. Typhi and, to a lesser degree, that of S. Enteritidis but did not affect uptake of S. Typhimurium (Fig. 5G, 5H). The increase in phagocytosis was accompanied by better intracellular clearance of bacteria (Fig. 5G, 5H). The ability to increase phagocytosis specifically of S. Typhi and promote its intracellular clearance was also observed with anti-O9 mAb (Fig. 5I).

We used immunization with unmodified, capsule-negative, live S. Typhi followed by challenge with nontyphoidal serovars, S. Enteritidis and S. Typhimurium, in mice as a model to study cross-protection with pathogenic Salmonella. Our results showed that immunization of mice with S. Typhi generates T cell responses, which are highly cross-reactive and could bring about clearance of related serovars in experimental models of infection. The humoral response elicited with S. Typhi could be categorized into highly serovar-restricted and broadly cross-reactive Abs, with only the former contributing to protection. Serovar-restricted anti–S. Typhi Abs bound live intact bacteria, whereas cross-reactive Abs did not. The former group of Abs were directed predominantly against O Ag–dependent determinants of LPS and cross-reacted with intact S. Enteritidis that, like S. Typhi, belongs to serogroup D and shares O Ags 9 and 12 with S. Typhi but not with intact S. Typhimurium that belongs to serogroup B and shares O12 with S. Typhi or with S. Paratyphi A that belongs to serogroup A and also shares O12 with S. Typhi. The lack of binding of anti–S. Typhi Abs to S. Typhimurium and S. Paratyphi A was not due to the absence of cross-reactive Abs directed against O12-dependent determinant but due to inaccessibility of this antigenic determinant on the surface of Salmonella. Similar serogroup-restricted binding of Abs was observed upon immunization of mice with S. Typhimurium and S. Enteritidis. In fact, this kind of specific binding was also seen with Abs generated against purified LPS from S. Typhi. These results reveal that the O-antigenic determinants of Salmonella LPS may be organized such that only serogroup-restricted ones remain accessible on the surface of bacteria. These determinants may, therefore, be involved in specific interactions with the host. Indeed, in vitro analysis showed that Abs directed against surface-accessible determinants could inhibit bacterial invasion of epithelial cells and promote phagocytosis of bacteria in macrophages in a serogroup-specific fashion. Interestingly, at lower concentrations, anti–S. Typhi Abs did not inhibit the invasion of cells, yet bacteria coated with these Abs were cleared better, which may be because of engagement of TRIM21 that binds the Fc region of the Ab and has been shown to bring about clearance of intracellular bacteria through autophagy (19, 20). It is also possible that these Abs neutralize the inhibitory effect that O Ags have been shown to exert on antibacterial host defense in intestinal epithelial cells (21). Our results are consistent with previous studies that have shown protection against Salmonella infection with vaccine candidates based on O Ags (11, 12, 22–26) and a more recent report, which suggests that the targeting of protective Abs to Salmonella surface is influenced by the physical and chemical nature of O Ags (27).

Currently, there are no vaccines available for human use against nontyphoidal Salmonella serovars including S. Typhimurium and S. Enteritidis, which are a major cause of iNTS disease in humans (28, 29). Oral-attenuated S. Typhi Ty21a vaccine, which gives 60–70% protection against S. Typhi in humans, has the potential to elicit immune responses that can cross-react with related Salmonella serovars. However, whether those responses render protection against related serovars has not been conclusive. Studies carried out with sera obtained from Ty21a-vaccinated subjects suggested that this vaccine may generate cross-reactive Abs to Ags including LPS of S. Paratyphi A and S. Paratyphi B (30–32). Yet, data from immunized subjects did not suggest that this vaccine was sufficiently effective against these serovars (30, 33). The results presented in this study provide an explanation why immunization with S. Typhi–based vaccines may not generate immunity-relevant Abs against S. Paratyphi A and B and other closely related Salmonella serovars. In contrast, those vaccines would elicit immunity-relevant Abs against S. Enteritidis and other serogroup D Salmonellae. Our data also suggest that cross-reactive T cell responses produced upon immunization with S. Typhi can bring down bacterial load during infection with other Salmonella serovars even in the absence of protective Abs. Therefore, some degree of protection against infections with S. Paratyphi and other Salmonella serovars in subjects vaccinated with Ty21a might have been provided by cross-reactive T cells. Our results emphasize the importance of analyzing Ab responses against live intact unmodified bacteria. Analysis of Abs against extracted Ags including LPS by immunoassays does not provide a correct measurement of functionally relevant Abs (30, 34, 35). In fact, even coating on solid support opens up live bacteria for reactivity with cross-reactive protection-irrelevant Abs directed against antigenic determinants not exposed on the surface. Therefore, results from previous studies on protection-relevant B cell reactivities using ELIspot assays with coated bacteria should be interpreted cautiously (30, 34).

Taken together, our study suggests that anti-Salmonella Abs can impart heterologous protection provided the serovars share accessible O Ag–dependent surface determinants. Our findings advocate investigations on the use of Ty21a, which is a safe vaccine in humans, for immunization against iNTS disease produced by S. Enteritidis and also indicate that introduction of accessible O determinants of S. Typhimurium in Ty21a might be a good option to design one vaccine that could provide protection against S. Typhi and at least two major nontyphoidal Salmonella serovars. Ty21a may also offer additional beneficial effects of nonspecific immunity as suggested by a recent study (36).

We thank Drs. Rahul Pal, Devinder Sehgal, and Krishnamurthy Natarajan for valuable suggestions in the course of this study as well as members of the Ayub laboratory for insightful discussions.

The authors have no financial conflicts of interest.