INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Enterotoxigenic Escherichia coli strains produce a plasmid-encoded heat-labile enterotoxin (LT) (15, 34) related to cholera toxin (CT) (9, 35). LT is composed of two subunits, A and B, which are exported to the periplasmic space, where they assemble into an AB₅ multimeric complex (16). Several mutants of LT-A have been constructed, and in particular, a nontoxic mutant which contains a substitution of serine 63 with lysine (LTK63) has been shown to maintain the structural and immunogenic properties of wild-type LT (21, 27, 28). LTK63 has also been found to display the strong mucosal adjuvant activity pertaining to wild-type LT. Efficient induction of mucosal immune response, specifically in the mouse vagina, has been achieved via the intranasal route of immunization (10). For the development of oral vaccines, however, it would be desirable to exploit the properties of LTK63 for enhancing antigen-specific immune response in the intestinal mucosa by means of oral delivery of the potent mucosal adjuvant.

Oral delivery of antigens by live vaccines is known to lead to a more effective production of antigen-specific antibodies in mucosal secretions than oral administration of the soluble antigen (36, 39). Several antigen delivery systems which use as carriers mutant intracellular pathogens that have lost the ability to persist and produce the disease while retaining limited growth in vivo have been developed. In particular, attenuated Salmonella mutants are suitable immunological carriers for virulence determinants from other enteric bacteria in that they can induce humoral immune response selectively at the site of colonization, the gut mucosa. Vaccine strains of Salmonella have been successfully attenuated by introducing different types of mutations (5, 8, 23, 26). Notably, Salmonella strains with a galactose epimerase (galE) mutation (18) or deletions in genes for the biosynthesis of aromatic compounds (aro mutants) (11, 12, 17, 19) or in the adenylate cyclase (cya) and cyclic AMP receptor protein (crp) genes (6) are the most extensively characterized.

Delivery of the B subunit of the E. coli enterotoxin (LT-B) by a galE mutant of Salmonella typhimurium has been shown to elicit low levels of anti-LT-B serum and mucosal antibodies. Since the vector used for expression of LT-B was rapidly lost in vivo, i.e., in the absence of the antibiotic required for selection of the plasmid, the level of immune response could be correlated only with the amount of antigen expressed during the initial phase of invasion (3).

Recently, direct comparison between the aroA aroD/pnirB and the cya crp asd/asd⁺ delivery systems for the ability to induce humoral and cellular immunity after a single immunization showed that the former vaccine strain had greater potential as a carrier for antigen delivery (20). However, the balanced lethal asd system for in vivo selection of plasmids expressing heterologous antigens in the attenuated cya crp asd strains is still very attractive in that asd⁺ plasmids do not require antibiotic resistance markers for selection while stably maintained in vivo (24). In addition, the cya crp asd/asd⁺ delivery system has been reported to induce protective immunity against several pathogens (25, 29, 40). Most of these studies have restricted analysis of the immune response to antigens expressed from the same asd⁺ plasmid carried by cya crp asd mutants usually of the same S. typhimurium serotype. In this work, we have analyzed the influence of heterologous gene dosage, and thus level of expression, as well as S. typhimurium strain variability on immune response toward both the heterologous antigen, a nontoxic mutant of E. coli LT, and the carrier Salmonella strain. Effects of a single integration into the host DNA and episomal vectors at different copy numbers were compared in S. typhimurium strains of two different cya crp asd serotypes, UK-1 and SR-11.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Strains, plasmids, and media. The bacterial strains and plasmids used are listed in Table 1. The cya crp asd S. typhimurium strains were kindly provided by Roy Curtiss III, Washington University, St. Louis, Mo.

DNA manipulations. Extraction and purification of plasmid DNA were carried out as described by Sambrook et al. (31). Genomic DNA extraction was performed as described by Rossolini et al. (30). Southern blot experiments were performed by bidirectional transfer of DNA from agarose gels to nylon membranes. Specific DNA probes were obtained amplifying DNA sequences by PCR and labeled with [-³²P]dATP, using the random-primers method.

SDS-PAGE and immunoblot analysis. Bacterial cultures were grown to the logarithmic or stationary phase in LB broth. Cells were collected and lysed by boiling (5 to 10 min) in LSB buffer (2% sodium dodecyl sulfate [SDS], 10% glycerol, 100 mM dithiothreitol, 62.5 mM Tris-HCl [pH 6.8]). Total proteins in cell extracts were quantified with a micro-bicinchoninic acid protein assay reagent kit (Pierce, Rockford, Ill.). Proteins were separated by SDS-PAGE (15% polyacrylamide gel) and transferred to nitrocellulose membranes. Mouse polyclonal anti-LT antiserum was used as the primary antibody. Bands were visualized by enhanced chemiluminescence (Amersham Life Science) or with 4-chloro-1-naphthol (Sigma) as the substrate. Evaluation of the amount of LT produced by the different strains was obtained by densitometric analysis of bands, using purified enterotoxin or CT as a reference.

Immunization and sampling. Groups of five or seven 6- to 8-week-old female BALB/c mice (Charles River, Calco, Como, Italy) were immunized intragastrically with two or four doses containing ca. 8 × 10⁸ CFU of each Salmonella strain. Mice were deprived of water and food for 12 h and fed with 30 µl of 0.2 M sodium bicarbonate to neutralize stomach acidity 30 min prior to immunization. Groups of mice were immunized on days 0 and 10 to 14 or on days 0, 10, 17, and 24.

ELISA for LT and Salmonella. Individual mouse serum and mucus samples were tested for immunoglobulin A (IgA) and IgG antibodies against LT or whole Salmonella cells by enzyme-linked immunosorbent assay (ELISA). A PBS solution of purified LT (a gift from M. Pizza) was absorbed to 96-well ELISA plates (Greiner-GmbH, Kremsmunster, Austria) at 0.2 µg per well overnight at 4°C. For anti-Salmonella antibody titration, bacteria were grown overnight, harvested by centrifugation, and resuspended in PBS at 3 × 10¹¹ CFU/ml. Bacteria were heat killed for 10 min at 80°C and stored at 20°C. Each sample well was coated with 0.1 ml of this suspension diluted 100-fold in 0.1 M carbonate buffer (pH 9.6). Plates were blocked with PBS containing 0.05% Tween 20 and 1% bovine serum albumin for 2 h at 37°C. After washing with PBS, an anti-isotype secondary antibody (Sigma) was added and plates were incubated for 2 h at 37°C. Serum samples were incubated with alkaline phosphatase-conjugated goat anti-mouse IgG (-chain specific) antibodies, mucus samples were incubated with biotin-conjugated goat anti-mouse IgA (-chain specific) antibodies. Plates were washed with PBS, developed with a solution of p-nitrophenyl phosphate (1 mg/ml) in 1 M diethanolamine buffer (pH 9.8), and read in an ELISA reader (Titertek Multiscan) at 405 nm. Titers were expressed as the last dilution that gave an optical density at 405 nm of 0.1. Under these experimental conditions, preimmune samples always gave an optical density at 405 nm of <0.1 from the first dilution.

Nucleotide sequence accession number. The nucleotide sequence of the mini-Tn5 insertion in the virulence plasmid described in this work has been assigned GenBank accession no. AF025956.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Influence of copy number on expression level of LT-K63 in vitro. Expression of LTK63 by S. typhimurium attenuated strains carrying the eltK63 gene cloned into the pYA plasmids listed in Table 1 was examined. Western blot analysis of total-cell extracts and periplasmic fractions from logarithmic- and stationary-growth-phase cultures of the different S. typhimurium strains showed, as expected, a good correlation between the amount of LTK63 produced by each strain and the copy number of the recombinant plasmid carried by the strain. Expression of the enterotoxin gene, which is under the control of its own promoter in these constructs, was constitutive in all strains analyzed and in all conditions tested.

View larger version (49K): [in this window] [in a new window]   FIG. 1.   Western blot analysis of expression of LTK63 in different S. typhimurium strains, using an anti-LT antibody. (a) Total-cell extracts of stationary-growth-phase cultures of strains 3987 (lane 1), 3987-18A (lane 2), 3987:pYA3074-LTK63 (lane 3), 3987:pYA3149-LTK63 (lane 4), and 3987:pYA3137-LTK63 (lane 5); (b) total-cell extracts of late-logarithmic-growth-phase cultures of eltK63 single-copy integrants. Total proteins in cell extracts were quantified, and 30 µg of each sample was used for the immunoblot. Purified CT (3 µg) and LT (3 µg) were used as a control and as a molecular mass marker for the A, A1, and B subunits of the toxins. (c) Nucleotide sequence of the integration site of the mini-Tn5 asd eltK63 into the virulence plasmid of strains 3987-18A and 4217-10E. The sequence of the mini-Tn5 19-bp O end flanking the eltK63 promoter (PeltK63; arrow) is boxed.

Serum antibody response induced by S. typhimurium 3987 expressing different levels of LTK63. Groups of five BALB/c mice were immunized orally with strain 3987 containing the low-copy-number plasmids pYA3074-LT and pYA3074-LTK63 or the high-copy-number plasmid pYA3137-LTK63. The same strain carrying the pYA3074 vector without the LTK63 insert and the parental attenuated asd⁺ strain 3985 were used as control strains. A single oral booster dose of 8 × 10⁸ S. typhimurium cells was administered 10 days after the first inoculation. The Salmonella-specific IgG titers in serum samples from immunized animals 24, 36, and 66 days after the first dosing were found to differ significantly particularly at day 24, in that strain 3987:pYA3137-LTK63 induced an average 10-fold-higher response. A diminished, but still significant, difference was observed at day 66 (Fig. 2). A possible explanation for the less pronounced difference at day 66 in mice receiving the high-copy-number plasmid strain could be that the higher anti-Salmonella titers elicited by this strain after the first immunization result in a reduced number of organisms able to colonize after the second inoculum.

View larger version (61K): [in this window] [in a new window]   FIG. 2.   Influence of amounts of LT or LTK63 synthesized by S. typhimurium on anti-Salmonella immune response. Serum IgG activity to S. typhimurium in mice immunized with two doses of strain 3987 containing the low-copy-number plasmids pYA3074-LT and pYA3074-LTK63, the high-copy-number plasmid pYA3137-LTK63, or the control plasmid pYA3074 without the enterotoxin gene insert was determined. The two doses were given to groups of five mice on days 0 and 10. IgG responses were detected by ELISA 24, 36, and 66 days after primary immunization. The standard error of the mean was calculated for all data, and mean values were compared by Student's t test. P values of 0.0001 were considered statistically significant.

Influence of Salmonella genetic background on the induction of serum antibody response. A similar immunization scheme was extended to include oral inoculation of groups of seven BALB/c mice with 8 × 10⁸ cells of the three S. typhimurium strains carrying a single-copy insertion of the mini-Tn5 asd eltK63 minitransposon into the genomes of S. typhimurium 3987 and 4217 (Table 1).

View larger version (47K): [in this window] [in a new window]   FIG. 3.   ELISA analysis of immune response induced by two different genetic backgrounds carrying LTK63 single-copy integrations. Groups of seven mice were inoculated with the two S. typhimurium strains on days 0 and 14. Samples were collected at 24, 36, and 66 days after primary immunization. (A) Serum anti-LTK63 and anti-Salmonella IgG antibody response; (B) mucosal anti-LTK63 and anti-Salmonella IgA titers. Mean values of each group were compared by Student's t test. P values of 0.05 were considered statistically significant.

Differential induction of serum antibody response by high LTK63 expression. Since the amount of enterotoxin produced by S. typhimurium seemed the most crucial parameter in determining a good immune response, an immunization experiment was carried out with four doses of the S. typhimurium strains containing the pYA3149-LTK63 and pYA3137-LTK63 constructs. The strains carrying the medium-high-copy-number plasmid were included in this experiment since strain 4072:pYA3137-LTK63 lysed after growth in static culture. In addition, in order to compare this immunization scheme with the previous ones, groups of seven mice were inoculated with only two doses of the same plasmid-bearing S. typhimurium strains used for immunizing mice with four doses. The immune response in these animals was evaluated only at day 66 from the first inoculum. The results presented in Fig. 4A show that both the anti-LTK63 and the anti-Salmonella IgG titers were highest for the group of mice immunized with strain 3987:pYA3137-LTK63.

View larger version (49K): [in this window] [in a new window]   FIG. 4.   ELISA analysis of immune response induced by UK-1 (3987) and SR-11 (4072) strains expressing high levels of LTK63. Groups of seven mice were immunized with two (on days 0 and 10) or four doses (on days 0, 10, 17, and 24) of strains 3987:pYA3137-LTK63, 3987:pYA3149-LTK63, and 4072:pYA3149-LTK63. Samples were collected at day 66 after primary immunization from mice immunized with two doses (66*) and at 24 and 66 days from mice immunized with four doses. (A) Serum anti-LTK63 or anti-Salmonella total IgG titers; (B) mucosal anti-LTK63 or anti-Salmonella total IgA titers. Mean values of each group were compared by Student's t test. P values of 0.001 (A) and 0.01 (B) were considered statistically significant.

Analysis of mucosal antibody response. Intestinal mucus samples from mice immunized with the S. typhimurium strains reported in Table 2 were analyzed for LTK63-specific and Salmonella-specific IgA responses. The mean IgA titers of each group of seven mice are shown in Fig. 3B and 4B. The anti-LTK63 secretory response was clearly higher in mice inoculated with the medium-high- and high-episomal-expression constructs (Fig. 4B) than in mice immunized with the eltK63 integrants (Fig. 3B), with the exception of the group immunized with strain 3987-18A. The high anti-LTK63 IgA titers in the latter group of mice correlate well with the anti-LTK63 serum IgG titers found in the same group (Fig. 3A), although the variation of titers in individual mice was greater for the IgA samples than for the IgG samples.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

As most studies on attenuated Salmonella carriers have analyzed differences in the induction of immunity between levels of antigen expression (3, 4), selection for vector maintenance in vivo, or attenuation of the delivery system (20, 37), we have compared the immune responses induced by the same antigen when delivered by different strain-plasmid combinations. In particular, we have analyzed a single integration into the host DNA and multicopy episomal vectors in three cya crp asd strains of different genetic backgrounds expressing a nontoxic mutant of the E. coli LT.

A first comparison of variation in antigen delivery by the same S. typhimurium strain in relation to different expression levels of the antigen was made by analyzing the systemic response in mice immunized with strain 3987 containing the low- and high-expression constructs. The lack of an LTK63-specific immune response in mice immunized with the low-expression construct, in contrast to high levels of IgG antibodies elicited by the high-expression vector, correlates with a fivefold difference in amount of LTK63 synthesized in vitro by the two constructs. This finding suggests that for this S. typhimurium strain, the threshold for induction of anti-LTK63 systemic response is high (see below). Alternatively, expression in vivo of LTK63 from the two plasmids may not correlate with what is seen in vitro, although expression in both constructs is driven by the same promoter.

An interesting finding of this experiment is that strain 3987:pYA3137-LTK63 induced anti-Salmonella IgG titers significantly higher, during the first 3 to 5 weeks postimmunization, than those induced by control strains or by low LTK63 producers. LTK63 has been previously shown to exert an adjuvant effect on immune response to different antigens administered to mice by different routes (10).

By comparing S. typhimurium strains with different genetic backgrounds and expressing the same amount of LTK63 from a single integration site, the influence of strain variability could be inferred from the anti-LTK63 antibody response, while both systemic and local immune responses to Salmonella were not significantly different for the different strains. Only strain 3987-18A could elicit high LTK63-specific IgG and IgA titers. Considering that strain 4217-10E contains the eltK63 gene integrated in the same locus of the virulence plasmid as strain 3987-18A and can persist longer in deep tissues, this result implies that very early events at the priming of the immune response are particularly critical in the induction of an anti-LT immune response (4). Furthermore, the 3987-18A genetic background may be intrinsically more effective in priming anti-LTK63 immune responses. The observation that high expression of the antigen is crucial during the early phase of the induction of the immune response is consistent with the finding by others that oral administration of Salmonella strains expressing LT-B induces a dramatic increase in the level of mRNA of some cytokines (e.g., interleukin-12) at mucosal sites within hours following immunization (1).

One of the advantages of the cya crp/asd system is the effectiveness of the asd selection for plasmid maintenance in vivo (7, 13, 24). As previously shown by other reports, the stability in vivo of the asd vectors used in this study for expression of LTK63 was found to be very high even for the high-copy-number plasmid pYA3137. It should be noted, however, that expression of LTK63 from the latter plasmid conferred an unusual phenotype to strain 4072 in vitro: lysis after static growth. For this reason, the analysis of immune response had to be extended to strains containing the medium-high-copy-number plasmid pYA3149. Comparison between both systemic and local antibody responses induced by UK-1 and SR-11 strains expressing LTK63 from the high- or the medium-high-copy-number plasmids clearly showed that, in addition to any influence of the S. typhimurium strain delivering the antigen, the level of expression of the heterologous antigen is critical for a high immune response. A 2-fold difference in amount of LTK63 produced by strain 3987:pYA3137-LTK63 was sufficient to enhance both IgG and IgA titers 10-fold. An adjuvant effect of LTK63 could be inferred also from these data, since higher Salmonella-specific titers correlated with higher expression of LTK63.

In conclusion, our data indicate that the same attenuation in different genetic background S. typhimurium strains, expressing the recombinant antigen at the same level, can produce significantly different immune responses. Moreover, delivery of the LTK63 antigen to the immune system by attenuated S. typhimurium strains is effective only when synthesis of the antigen is very high during the initial phase of invasion, while persistence of the S. typhimurium strain in deep tissues has only marginal influence.