Effect of Preexisting Immunity to Salmonella on the Immune Response to Recombinant Salmonella enterica Serovar Typhimurium Expressing a Porphyromonas gingivalis Hemagglutinin

Abstract

Recombinant Salmonella strains expressing foreign heterologous genes have been extensively studied as live oral vaccine delivery vectors. We have investigated the mucosal and systemic immune responses following oral immunization with a recombinant Salmonella enterica serovar Typhimurium expressing the hemagglutinin HagB from Porphyromonas gingivalis, a suspected etiological agent of adult periodontal disease. We have previously shown a primary mucosal and systemic response following oral immunization with χ4072/pDMD1 and recall responses following boosting at 14 weeks after primary immunization. In this study, we examined the effects of earlier boosting as well as the effects of deliberately induced immunity to the Salmonella carrier strain on subsequent immune responses. Mice boosted at week 7 following immunization, a point which corresponded to the peak of the primary response, generally showed lower responses than those boosted at week 14. When mice were preimmunized with the Salmonella carrier alone and then immunized with the recombinant strain 7 or 14 weeks later, significant reductions were seen for serum immunoglobulin G (IgG) antibodies at week 14 and for salivary IgA at week 7. No reductions were seen in serum IgA or vaginal wash IgA antibodies. Mice appear to be refractory to boosting with orally administered salmonellae at 7 weeks. Deliberate immunization with the carrier strain did not appreciably affect recall responses at 14 weeks, with the exception of the serum IgG responses, nor did it affect colonization of the Peyer's patches.

Recombinant attenuated Salmonella strains have received much attention recently for their potential as antigen delivery systems for mucosal immunization (4, 14, 20). One concern with the use of Salmonella strains as a vaccine carrier expressing heterologous antigens is the effect of introduction into immunized hosts or the repeated use of the organism.

To date, few studies have focused on the effect of immunological experience on the immune response to foreign antigens delivered by recombinant Salmonella strains. Bao and Clements found that prior immunological experience potentiates the subsequent antibody response following oral immunization with recombinant salmonellae (2). Likewise, Whittle and Verma reported that not only did previous immunological experience with salmonellae not limit the immune response to a foreign antigen carried by the same organism but it also appeared to enhance the response (22), although in this case the intraperitoneal route was used. In contrast, Forrest reported impairment of immunogenicity of S. enterica serovar Typhi Ty21a due to preexisting cross-reacting intestinal antibodies in individuals from areas where the organism is not endemic (B. D. Forrest, Letter, J. Infect. Dis. 166:210–212, 1992). Similarly, Attridge et al. (1) and Roberts et al. (19) found reductions in serum responses in orally immunized mice as a consequence of prior exposure to the Salmonella carrier.

We have been studying the use of Salmonella enterica serovar Typhimurium χ4072 which is attenuated by mutations in the cya and crp genes (12). We have previously reported the expression of a Porphyromonas gingivalis hemagglutinin (HagB) in an active biological form (9) that has been shown to induce both systemic and mucosal immune responses specific to the HagB protein in orally immunized mice (8). In addition, we have demonstrated a recall response to HagB in mice following boosting at 14 weeks (16).

Because of conflicting results regarding prior exposure to the Salmonella vector, we investigated the role of preexisting immunity to the Salmonella carrier strain χ4072 and its effect on subsequent immune responses to the HagB protein following oral immunization with the recombinant serovar Typhimurium χ4072/pDMD1 by examining mice boosted during the peak of primary response and mice previously immunized with the carrier alone.

MATERIALS AND METHODS

Bacterial strains, plasmids, media, and culture conditions.

Serovar Typhimurium χ4072 and plasmid pYA292 (12) were provided by Roy Curtiss III (Washington University, St. Louis, Mo.). Plasmid pDMD1 containing the P. gingivalis hagB gene was constructed and introduced into serovar Typhimurium χ4072 as previously described (8). Strains were routinely grown at 37°C, and stock cultures were stored at −80°C in 15% glycerol as previously described (9).

Mouse immunization and sample collection.

Female BALB/c VAF/Plus mice (Charles River, Wilmington, Mass.) were housed in the Infectious Disease Isolation Unit of the University of Florida Animal Resource Center and given food and water ad libitum. Mice were immunized by gastric intubation three times on alternate days with 10⁹ CFU of the appropriate serovar Typhimurium strain or were sham immunized with diluent (0.1 M NaHCO₃). Boosting was carried out in the same manner. Serum, saliva, and vaginal washes were collected and processed as previously described (8, 16). The supernatant fluids were stored at −80°C.

Immunoassay methods.

Samples were assayed for immunoglobulin G (IgG) and IgA antibody to strain χ4072/pYA292 or HagB on microwell plates as described previously (8) using an enzyme-linked immunosorbent assay coated with either formalin-killed Salmonella carrier strain χ4072/pYA292 or purified HagB protein isolated with the QIAexpress system (Qiagen, Inc., Valencia, Calif.). Vaginal washes were normalized to total IgA, and salivary IgA anti-HagB antibody levels were normalized to amylase activity levels to account for variable dilution. Amylase activity was determined for each salivary sample with a colorimetric enzyme assay (3).

Colonization of Peyer's patches and spleen.

A subset of mice were orally immunized with a single dose with recombinant serovar Typhimurium χ4072/pDMD1 at week 7 (groups I and III) or week 14 (groups II and IV). Five days later, the mice were euthanized with sodium pentobarbital, followed by cervical dislocation. The spleen and a set of five Peyer's patches were surgically removed under sterile conditions. Spleen tissues were dispersed in 2.5 ml of phosphate-buffered saline (PBS) with sterile glass tissue homogenizers and stored on ice. Peyer's patches were dispersed in 1 ml of PBS with a handheld homogenizer and sterile disposable pestles and then stored on ice. Dilutions of each sample were plated on Luria-Bertani agar with nalidixic acid (20 μg/ml) to determine the number of CFU.

Statistical analysis.

For the time of boost experiments, data were logarithmically transformed, and the differences between week 6 and post-boost responses were compared using repeated-measures analysis of variance for each parameter. Multiple comparisons between means was done with Fisher's least-significant-difference method. For preexisting immunity and colonization experiments, the Wilcoxon ranked-sum test was used.

RESULTS

Effect of boosting during the peak of the primary response.

We have previously shown that boosting is possible at week 14 following primary immunization after the primary peak response had subsided (16). We had identified that the kinetics of the response to the Salmonella carrier paralleled the response to the expressed HagB (not shown); therefore, we examined the effect of boosting during the peak primary response at week 7 versus week 14. Groups of mice were immunized at week 0 and boosted at either week 7 or week 14. Serum, saliva, and vaginal washes were collected at the indicated time points and analyzed for specific anti-HagB IgG or IgA antibodies. Serum IgG anti-HagB recall responses were reduced in mice boosted at week 7 (Fig. 1A) compared to week 14 (Fig. 1B). Likewise, serum IgA anti-HagB recall responses were reduced in mice boosted at week 7 (Fig. 1C) compared to those boosted at week 14 (Fig. 1D).

FIG. 1.

Serum IgG (A and B) and IgA (C and D) anti-HagB levels following immunization (↓) at week 0 and boosting at week 7 (group I) or week 14 (group II) with serovar Typhimurium χ4072/pDMD1. Samples were taken at indicated times (↑). Antibody levels are expressed as means (n = 6). Error bars represent the standard error of the mean. Statistically significant increases (P < 0.05) in antibody levels from the primary response at week 6 are indicated for each group (∗).

In the mucosal compartment, a reduction in salivary IgA anti-HagB was seen in mice boosted at week 7 (Fig. 2A) compared to those boosted at week 14 (Fig. 2B), while little difference was seen in vaginal washes (Fig. 2A and B).

FIG. 2.

Mucosal salivary (A and B) and vaginal (C and D) IgA anti-HagB levels following immunization at week 0 and boosting (↓) at week 7 (group I) or week 14 (group II) with serovar Typhimurium χ4072/pDMD1. Samples were taken at the indicated times (↑). Antibody levels are expressed as means (n = 6). Error bars represent the standard error of the mean. Statistically significant increases (P < 0.05) in IgA levels from the primary response at week 6 are indicated for each group (∗).

Effect of prior immunization with the Salmonella carrier strain.

In order to address preexisting immunity more directly, we examined recall responses in mice previously immunized with the Salmonella carrier alone. The immunization and sampling schedule is shown in Fig. 3. Two groups of 10 mice (groups I and II) were orally immunized at week 0 with the carrier strain containing the plasmid cloning vector, serovar Typhimurium χ4072/pYA292, to induce immunity to the Salmonella carrier. Two additional groups (groups III and IV) were sham immunized with 0.1 M NaHCO₃. The mice were then orally immunized with serovar Typhimurium χ4072/pDMD1 at week 7 (groups I and III) or week 14 (groups II and IV). Samples of serum, saliva, and vaginal washes were collected and analyzed for specific anti-HagB IgG or IgA antibodies at the times indicated. Serum samples were also analyzed for IgG and IgA antibodies, and vaginal washes were analyzed for IgA antibodies to salmonellae at these sampling times. Four mice from each group were sacrificed 5 days after a single immunization to assess colonization of Peyer's patches and spleen.

FIG. 3.

Immunization and sampling schedule for effect of prior immunization with the Salmonella carrier strain. Mice were immunized and boosted at the times indicated (↓). Samples of serum, saliva, and vaginal washes were collected and analyzed for specific anti-HagB IgG or IgA antibodies (↑). Antibodies to salmonellae in serum and vaginal washes were also measured prior to boost(s). Four mice from each group were sacrificed (X) 5 days after a single immunization (single ↓) to assess colonization of Peyer's patches and spleen.

The serum IgG and IgA and the vaginal-wash IgA antisalmonella responses at week 6 in group I (preimmunized) were significantly elevated (P < 0.01) over the week-6 samples from group III (saline control). The differences were not significant at week 13 between groups II and IV since the antisalmonella levels had declined in group II. The anti-HagB IgG response in serum was significantly reduced (P < 0.05) in mice preimmunized with the serovar Typhimurium carrier and then immunized at week 14 (Fig. 4A). While there was some reduction in the mice immunized at week 7, the difference was not statistically significant. Preimmunization with the Salmonella carrier had no effect on serum IgA anti-HagB levels at either week 7 or week 14 (Fig. 4B).

FIG. 4.

Serum IgG (A) and IgA (B) anti-HagB responses from groups of BALB/c mice sham-immunized or immunized with the serovar Typhimurium carrier strain and then boosted with serovar Typhimurium χ4072/pDMD1 at week 7 or week 14. Responses at 5 weeks following boosting are shown. Antibody levels are expressed as means (n = 6). Error bars represent the standard error of the mean. Statistically significant decreases (P < 0.05) are indicated (∗).

With regard to mucosal responses, preimmunization with the carrier strain significantly reduced the salivary IgA anti-HagB response in animals immunized at week 7 (Fig. 5A). The levels were not significantly reduced in animals immunized at week 14. No statistically significant differences were seen in vaginal-wash IgA anti-HagB levels at either week 7 or week 14 (Fig. 5B).

FIG. 5.

Salivary IgA (A) and vaginal IgA (B) anti-HagB responses from groups of BALB/c mice sham immunized or immunized with the serovar Typhimurium carrier strain and then boosted with serovar Typhimurium χ4072/pDMD1 at week 7 or week 14. Responses at 5 weeks following boosting are shown. Antibody levels are expressed as means (n = 6). Error bars represent the standard error of the mean. Statistically significant decreases (P < 0.05) are indicated (∗).

Colonization of Peyer's patches and spleen.

Following a single oral dose of serovar Typhimurium χ4072/pDMD1 at week 7 or week 14, mice were sacrificed, and spleens and five Peyer's patches were removed for homogenization. No CFU were recovered from the spleens of any group. From the Peyer's patches, while there was some variability between the individual mice (Fig. 6), there was no statistically significant difference between the mean CFU of the two groups.

FIG. 6.

Colonization of Peyer's patches with serovar Typhimurium χ4072/pDMD1 to determine the effect of preexisting immunity to the Salmonella carrier. CFU from individual mice are shown (⧫), along with the mean (horizontal bars).

DISCUSSION

Recombinant Salmonella strains for delivery of heterologous genes have shown potential for use in oral vaccines against a variety of known pathogens (4, 20). There remains, however, a concern that reuse of the Salmonella carrier may lead to reduced effectiveness due to the induction of immunity to the carrier. Forrest showed an inverse correlation between the presence of cross-reacting intestinal sIgA antibody and the intestinal antibody response to serovar Typhi Ty21a in human subjects not residing in an area where Salmonellae are endemic, although the effect could be overcome with higher doses (B. D. Forrest, Letter). Ferguson and Sallam, however, showed that maximal intestinal responses in humans were seen when subjects had evidence of intestinal priming (A. Ferguson and J. Sallam, Letter, Lancet 339:179, 1992). Bao and Clements found that prior exposure in mice to the Salmonella carrier potentiated subsequent serum and mucosal immune responses (2), while Attridge et al., using different carrier strains, found inhibition of subsequent serum responses (1). It was suggested that the strain used in their study persisted longer in the mouse than in the earlier study, which may have led to a stronger immune response to the carrier.

In a recent study, Roberts et al. (19) reported that prior immunization with the carrier strain and, to a lesser extent with a heterologous strain, reduced subsequent immune responses to the recombinant expressed fragment C of tetanus toxin. The effects were reduced with a more immunogenic strain containing a different promoter. Responses were only examined after challenging 44 days (ca. 6 weeks) following preimmunization. We have determined that the peak of the response to the Salmonella carrier occurs at 5 to 7 weeks, in parallel with the response to HagB (unpublished observations). In the present study, we have shown that the effect of prior immunity decreases when the challenge is delayed until after the peak response (i.e., 14 weeks).

The characteristics of the carrier strain may affect the immunogenicity and suitability for vaccine use. A variety of mutations have been employed to attenuate Salmonella vaccine strains (5, 6, 13, 15, 17, 18, 21). Dunstan et al. compared six isogenic serovar Typhimurium strains with different attenuating mutations expressing the C fragment of tetanus toxin for immunogenicity and protection against challenge with tetanus toxin (7). Five of the six strains which colonized the Peyer's patches induced comparable immune responses and were protective, while differences in spleen colonization did not correlate with immunogenicity. The mutant which showed the least colonization of the Peyer's patches (ΔpurA) induced the lowest response and was not protective. Dunstan et al. concluded that immunogenicity did not correlate with relative invasiveness but with Peyer's patch colonization, attenuating mutation and strain background.

In our studies, inhibition of recall responses was seen when boosts were given early during the primary response. This could be due either to immune exclusion in the gut or to lack of time for the development of sufficient numbers of memory cells. In fact, antibody tends to inhibit the response of naive B cells to antigen and enhance the response of memory cells (11). Deliberate immunization with the carrier strain did not appreciably affect recall responses at 14 weeks, with the exception of the serum IgG responses, nor did it affect colonization of the Peyer's patches. This agrees with our previous findings (16), where recall responses were seen after boosting at week 14. The Salmonella delivery system also appears to be capable of inducing long-term immunity. We have recently shown that mice primed with S. enterica serovar Typhimurium χ4072/pDMD1 are capable of exhibiting a recall response at 52 weeks (submitted for publication), which represents nearly one-half the lifespan of a BALB/c mouse (10).

Thus, whether preexisting immunity interferes with subsequent use of live Salmonella vaccine vectors may depend on timing and the genetic characteristics and immunogenicity of the strain. With the appropriate vaccine strains, preexisting immunity should not preclude reuse of carriers or their use in areas where individuals have been previously exposed to salmonellae.