INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Brucella species are facultative intracellular gram-negative bacterial pathogens that infect both phagocytic and nonphagocytic cells (42). Brucella abortus causes abortion and infertility in cattle and also various chronic zoonotic infections in humans (8, 42). The intracellular localization of these bacteria implies that the immunity against Brucella requires a cell-mediated immune response, which makes the Th1 arm of the response very crucial for controlling the infection (44).

Brucella abortus strain B19 is one of the most commonly used attenuated live vaccines against bovine brucellosis and induces high level of protection in cattle (15). The presence of smooth lipopolysaccharide in the vaccine strain B19 may interfere with the discrimination between infected and vaccinated individuals (32) and impair the test and slaughter strategy. Moreover, this strain can cause abortion when administered to pregnant cattle (9) and is still fully virulent for humans (42). In order to avoid these drawbacks, alternative vaccination approaches are needed. Among these, subcellular vaccines able to induce protective Th1 cell-mediated immune response are being developed. Recombinant antigens of Brucella spp. such as HtrA (40), GroEL (2, 30, 34), GroES (34), Cu,Zn superoxide dismutase (SOD) (47, 49), YajC (52), UvrA (34), and L7 and L12 (37) have been shown to induce humoral and cell-mediated immune responses in mice, but only L7/L12 (36) and peptides comprising certain epitopes of Cu,Zn SOD (38, 47, 54) induced some level of protection in a mouse model of infection. While the protection afforded could be improved using a multiple subunit vaccine, it remains also possible that a more powerful antigen or a better adjuvant or both may lead to protection with a monovalent subunit vaccine.

Our laboratory has previously described bacterioferritin (BFR) (13) and P39 (a putative periplasmic binding protein) (10, 11) as T-cell immunodominant Brucella antigens (12), eliciting both a strong delayed-type hypersensitivity (DTH) in guinea pigs sensitized with brucellin and in vitro proliferation or gamma interferon (IFN-) production by peripheral blood mononuclear cells (PBMC) from infected cattle. The potential of these antigens to induce a Th1-oriented immune response makes them attractive candidates as a subunitary vaccine against brucellosis.

While the type of antigen and microorganism administered to a host (as well as the dose and route of immunization) are important factors that influence the type of immune response, it is also well established that the presence of certain cytokines at the site of injection is crucial for orienting the emerging T-cell response (21, 33, 50). For this late concern, adjuvants are of paramount importance. It has been shown that a synthetic phosphorothioate oligodeoxynucleotide containing an unmethylated, consensus immunostimulatory CpG motif (5'-purine-purine-CpG-pyrimidine-pyrimidine-3' oligodeoxynucleotide [CpG ODN]) can act as an adjuvant which favors cell-mediated immune mechanisms (19, 20, 25, 26) with a Th1-like cytokine profile (7, 24). This suggests that CpG ODN could act as an adjuvant for the clearance of intracellular pathogens (14). We decided to take this newly described adjuvant and test its potentiating effect with the Brucella T-cell antigens we have previously described.

In this study, we evaluated the potential of P39 and BFR with CpG ODN as adjuvant in inducing a Th1 response and the efficiency of these vaccines to protect BALB/c mice against an infectious B. abortus 544 challenge.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteria. The B. abortus virulent strain 544 was obtained from J.-M. Verger (Institut National de la Recherche Agronomique, Pathologie Infecteuse et Immunologie, Nouzilly, France), and B. abortus vaccine strain B19 was obtained from J. Goodfroid. They were grown in 2YT medium (10 g of yeast extract, 10 g of tryptone, and 5 g of NaCl per liter) for 3 days at 37°C as described earlier (16). Escherichia coli BL21(DE3) was grown on Luria-Bertani medium that contained 100 µg of ampicillin per ml. This strain was used for production of the recombinant BFR protein using the pET-15b-bfr vector.

Oligodeoxynucleotides. Phosphorothioate-modified ODNs were synthesized at Eurogentec. The ODNs used in these studies were the immunostimulatory CpG 1826 (5'-TCCATGACGTTCCTGACGTT-3') and non-CpG 1745 (5'-TCCAATGAGCTTCCTGAGTCT-3'), which are nonstimulatory and used as a control. (CpG motifs or reversed non-CpG motifs are underlined.) CpG 1826 has been well characterized for adjuvant activity with protein antigen (7).

Purification of BFR and P39 proteins. The bfr gene of Brucella melitensis 16M was subcloned into a pET-15b expression vector (Novagen, Madison, Wis.), and the resulting plasmid pET-15b-bfr was introduced in E. coli BL21(DE3). After 2 to 4 of induction with 1 mM IPTG (isopropyl--D-thiogalactopyranoside) (Promega, Madison, Wis.), bacterial cells from a 100-ml culture were washed once and then sonicated (seven times for 25 s each time on ice). The lysate was centrifuged for 10 min at 9,000 × g at 4°C. The pellet was kept frozen at 70°C. After it had thawed, the pellet was resuspended in lysis buffer (10 mM Tris-HCl [pH 8], 5 mM EDTA) containing 100 µg of lysozyme (Sigma, St. Louis, Mo.) per ml. The resulting lysate was centrifuged at 9,000 × g for 20 min at 4°C. The insoluble fraction of the His₆-tagged BFR protein was solubilized in 50 ml of 6 M guanidine-HCl (pH 6.5) buffer containing 0.05% Triton X-100, and the extract was centrifuged at 9,000 × g for 30 min at 4°C. The supernatant was purified on an Ni-affinity chromatography column (Pharmacia Biotech). The protein was eluted from the column in elution buffer (1 M imidazole, 0.5 M NaCl, 20 mM Tris-HCl [pH 7.0]). The purified protein was divided into aliquots and stored at 70°C until use. The purification of P39 was as described previously (28).

SDS-PAGE and immunoblotting. SDS-PAGE and immunoblotting were performed as previously described (48).

Immunization of chicken. Sonicated E. coli BL21(DE3) lysate (150 µg of proteins) was injected into the breast muscle for the induction of E. coli-specific antibodies in the egg yolk. Inoculations were repeated 3 and 6 weeks later. The eggs were collected 7 days after the last injection. The egg yolk was diluted in H₂O (10 times the yolk weight) and then frozen at 20°C to precipitate the lipids. The sample was thawed and centrifuged at 2,500 × g at 4°C for 45 min. The supernatant was filtered on a 0.45-µm (pore-size) filter and mixed with ammonium sulfate (25% saturation, final concentration). After a 20-min incubation at room temperature, the sample was centrifuged at 2,500 × g at 4°C for 30 min. The supernatant was mixed with ammonium sulfate (40% saturation, final concentration) and processed as before. Finally, the pellet was resuspended in 1 ml of phosphate-buffered saline (PBS)-azide (0.1%).

Immunization of mice. Female BALB/c mice were obtained from IFFa Credo, Brussels, Belgium, at 4 weeks of age. Mice were separated into nine groups of 12 mice. Groups 1, 2, and 3 received PBS, CpG ODN, and non-CpG ODN, respectively, and served as negative controls. Groups 4 and 5 were injected with the purified P39 and BFR alone, respectively. Groups 6 and 7 were injected with the recombinant protein with CpG ODN adjuvant. Finally, groups 8 and 9 received the recombinant antigens with the non-CpG ODN. Vaccines were prepared in PBS and contained combinations of the following: 20 µg of recombinant protein and/or 20 µg of oligonucleotides when needed. Vaccines were given intramuscularly (i.m.) into the left tibial anterior muscles in a total volume of 50 µl three times at 3-week intervals. Three weeks after the last injection, four mice randomly selected in each group were sacrificed by cervical dislocation. Their sera were harvested to determine the humoral immune response. Their spleens were removed aseptically to investigate the cellular immune response.

Isotype-specific immunoglobulin ELISA assays. Specific murine IgG1 and IgG2a isotypes were assayed by enzyme-linked immunosorbent assay (ELISA) using microplates (Nunc, Roskilde, Denmark) coated overnight at 4°C with an optimal concentration of the different antigens in 50 µl of PBS-0.1% Thimerosal (Sigma). Plates were washed twice with PBS and blocked with PBS-2.5% casein for 2 h at room temperature (RT). After three washes in PBS, 100 µl of serial twofold dilutions starting at 1/100 in dilution buffer (PBS-0.05% Tween 20-1.25% casein) were performed and then loaded in microwells and incubated at RT for 1 h. The sera from nonimmunized mice were used as negative controls. After five washes with washing buffer (PBS-0.05% Tween 20), biotinylated goat anti-mouse IgG1 or anti-mouse IgG2a antibodies (Amersham) were added at an optimal dilution for 1 h at RT. Following five additional washes with PBS-Tween, the plates were incubated for 1 h with 50 µl of a 1:1,000 dilution of streptavidin-horseradish peroxidase (Amersham) at RT. Finally, the plates were washed five times and developed for 10 min in the dark with TMB (3,3',5,5'-tetramethylbenzidine) at 40 mg/ml in pH 4.0 citrate buffer containing 1.7 ml of hydrogen peroxide (KPL, Gaithersburg, Md.) per liter. The reaction was then stopped by the addition of 2 N H₂SO₄ to each well. The absorbance was determined at 450 nm (Bio Kinetics Reader EL-340). Titers were defined as the highest dilution of mouse serum which gave an optical reading of three times the reading of the negative control.

Lymphocyte proliferation assays. Spleens were homogenized with 2 ml of tissue culture medium (RPMI 1640-5% fetal bovine serum; Gibco-BRL), and erythrocytes were lysed with Gey's solution. Splenocytes at 2 × 10⁵ per well were stimulated with concanavalin A (ConA; 3 µg/ml), recombinant protein antigen (10 µg/ml), bacteria lysate (30 µg/ml), or no additive in culture medium for a total volume of 0.2 ml per well. Cell proliferation was determined in triplicate, based on the uptake of [³H]thymidine ([methyl-³H]thymidine; CNA). After 2 days of incubation at 37°C in 5% CO₂, the plates were pulsed with [³H]thymidine at 0.5 µCi per well and processed, 18 h later with a cell harvester (Skatron, Inc., Sterling, Va.) onto glass filter strips (Skatron, Inc.). Tritiated-thymidine incorporation was counted by use of liquid scintillation spectroscopy with a Betaplate counter (WALLAC Oy, Turku, Finland). The mean number of counts per minute and the standard error of the mean for each triplicate of cells were determined.

In vitro assay for cytokine production by spleen cells. Levels of IFN- and interleukin-5 (IL-5) in murine splenocyte culture supernatants were measured after 96 h of incubation with antigen or mitogen as described for the lymphocyte proliferation assay. IFN- and IL-5 were assayed by specific ELISA kits (Pharmingen, San Diego, Calif.). Samples were tested in duplicate. The concentrations of IFN- or IL-5 were calculated from a standard curve for recombinant mouse IFN- or IL-5 (Pharmingen) generated in each assay. Values of less than 40 and 10 pg/ml were considered negative for IFN- and IL-5, respectively.

Protection assay. Three weeks after the last injection, the remaining mice of each group were challenged by the intraperitoneal route (i.p.) with 5 × 10⁴ CFU of B. abortus strain 544 in 100 µl of PBS. An additional group of eight mice vaccinated i.p. with B19 (10⁵ CFU) was challenged 4 weeks later in the same way and served as a vaccinated control. Spleen colonization with the challenge strain was determined at 4 and 8 weeks postinfection. Spleens were homogenized in 2 ml of 0.1% Triton-PBS, and 100 µl of 10-fold serial dilutions were plated in triplicate onto 2YT agar for 3 to 4 days at 37°C with 10% CO₂, and the CFU were counted. The limit of detection of spleen counts is 20 CFU/spleen. For the B19-vaccinated group, dilutions were spread on 2YT agar alone or on 2YT plus 0.1% erythritol for differentiation of B. abortus B19 strain from strain 544 (43).

Statistical analysis. Significances of differences were determined by use of the Student's t test or Fisher exact test as appropriate. A P value of <0.05 was considered significant.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Expression and purification of the recombinant BFR protein. E. coli BL21(DE3) transformed with pET-15b-bfr and induced with IPTG were treaded as described in Materials and Methods. SDS-PAGE analysis of lysate of E. coli BL21(DE3) transformed with the pET-15b-bfr vector demonstrated that a protein was readily produced upon induction (Fig. 1). The size of the expression product corresponded to the calculated molecular mass of the recombinant fusion protein (21 kDa). The recombinant BFR was solublized and purified by the Ni-chelate affinity chromatography (Fig. 1, lane 3). Use of hen anti-E. coli BL21(DE3) antibodies showed that nearly no E. coli proteins contaminated the purified recombinant antigen (data not shown).

View larger version (142K): [in this window] [in a new window]   FIG. 1.   SDS-PAGE analysis, followed by Coomassie blue staining of whole-cell lysate of E. coli BL21(DE3) transformed by pET-15b-bfr before induction (lane 1) and after 2 h of induction (lane 2) and of the purified recombinant BFR (lane 3). Molecular mass standards (lane MM) are indicated on the right.

Antibody responses. The induction of specific IgG2a subclass during an immune response should give an idea of the Th1-Th2 balance. The mouse sera were collected 3 weeks after the third immunization and were assessed for IgG1 and IgG2a against the relevant recombinant antigen or B. abortus 544 extract. Regardless of the coating antigen considered, no mice from the groups immunized with the ODN (either CpG or non-CpG) or with the recombinant antigens alone gave a detectable antibody response (data not shown). As illustrated in Fig. 2, no specific IgG1 or IgG2a antigen response could be detected from mice immunized with P39 or BFR adjuvanted with non-CpG ODN. The CpG ODN adjuvanted BFR induced a significant IgG2a response but failed to elicit a detectable IgG1 response. On the contrary, the P39 protein mixed with CpG ODN elicited an IgG1 and an IgG2a response. The IgG2a titer (4.1 log) was higher than the IgG1 titer (3.5 log), indicating a shift toward a Th1 type of response. In both cases, proteins with the CpG ODN adjuvant elicited a response against the B. abortus 544 extract which paralleled the response against the nominal antigen but of lower amplitude and failed to induce a detectable response against the irrelevant antigen. Both antigens with CpG ODN appear to induce a Th1 response, with the P39 giving a stronger one. The third immunization did not result in a significant increase of the antibody titers compared to data obtained after the second immunization (data no shown).

View larger version (11K): [in this window] [in a new window]   FIG. 2.   Analysis of IgG1 (A) and IgG2a (B) antibody responses of BALB/c mice to the recombinant BFR and P39 antigens and to B. abortus 544 extract. Mice were immunized three times with BFR or P39 in different ODN combinations. Sera from four mice per group collected 3 weeks after the last immunization were assayed individually by ELISA. Antibody levels are expressed as mean titer.

Proliferative responses to purified antigens. Figure 3 shows the pattern of the proliferative response of splenocytes. Splenocytes from animals immunized with the recombinant protein (BFR or P39) with CpG ODN proliferated in response to the specific antigen and also in response to Brucella lysate (Fig. 3A). On the other hand spleen cells from mice immunized with BFR or P39 with non-CpG ODN or with the ODN, BFR or P39 alone, failed to raise a proliferative response to the specific antigen in vitro (Fig. 3A, 3B). All animals immunized responded to the polyclonal stimulant ConA as a positive proliferation control (data not shown).

View larger version (15K): [in this window] [in a new window]   FIG. 3.   Proliferation of splenocytes from BALB/c mice immunized with the recombinant proteins with CpG or non-CpG ODN (A) or with P39, BFR or ODN alone (B). Splenocytes were stimulated in vitro with BFR, P39 (10 µg/ml), or bacterial lysate (30 µg/ml) or were not stimulated (RPMI). For each mouse (n = 4) the assays were set up in triplicate. The data represent the mean ± the standard deviation of the triplicate from four mice. ND, not done.

Cytokine assay. To evaluate the production of cytokines by splenocytes collected 3 weeks after the last immunization, the levels of IFN- and IL-5 production were measured in the culture supernatants of cells stimulated with the specific antigens (BFR, P39, or B. abortus 544) and with the mitogen ConA as a positive control (Table 1). The animals that were immunized with recombinant proteins (BFR or P39) plus CpG ODN showed a high production of IFN- after stimulation with the specific antigen or with B. abortus 544, whereas mice treated with P39 plus non-CpG ODN failed to do so. Mice treated with BFR plus non-CpG ODN showed a very weak response to BFR or B. abortus 544. In addition, splenocytes from mice immunized with recombinant proteins, CpG, or non-CpG ODN alone released IFN- only in response to ConA stimulation.

View this table: [in this window] [in a new window]   TABLE 1.   IFN- production in spleen cell cultures from immunized mice.

Protection studies. The remaining immunized mice were challenged by B. abortus 544 to examine the protective activity of the induced immune response. In this experiment, protection was defined as a significant reduction in the number of bacteria in the spleen from immunized mice compared to the mice receiving PBS. The vaccine efficacy was calculated as the log₁₀ of protection. As expected the B19 vaccine offers a significant protection at both 4 and 8 weeks postchallenge with 2.45 and 2.92 log units of protection, respectively. The only protective antigen in this test is the P39 protein with the CpG adjuvant (Table 2) that shows approximately 2.5 logs of protection at 4 weeks postinfection (p.i.) and 1.2 logs of protection at 8 weeks p.i. When this vaccine (P39+CpG) is compared to B19 vaccine strain, the protection was equivalent at 4 weeks p.i., whereas it was 1.71 logs lower at 8 weeks p.i. (P < 0.001). None of the other combinations of antigens or adjuvant could induce protection against B. abortus 544 at any time point after challenge (4 and 8 weeks) (see Table 2).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The DTH test has been widely used for the diagnosis of brucellosis in ruminant (3). For this test, the Brucellergene (Rhone-Merieux) is the more commonly used allergene and consists of a mixture of 20 to 30 cytoplasmic proteins prepared from a rough strain of B. melitensis B115 (22). Previous work done in our laboratory identified among these proteins the P39 and the BFR proteins as Brucella T dominant antigens, which were shown to induce a positive DTH in infected guinea pigs and also to stimulate the production of IFN- by the blood cells of infected cattle (12). According to the fact that the protective immune response against Brucella infection is described as Th1 oriented (44, 58), these characteristics make both P39 and BFR attractive candidates for the development of a subunitary vaccine against brucellosis if properly combined with adjuvant. This potential was evaluated here by using these antigens as purified recombinant proteins and CpG ODN as an adjuvant in a mouse model of infection.

The choice of this adjuvant was dictated by the fact that it can stimulate multiple types of immune cells, leading to enhanced Th1 response characterized by the production of IFN-, IL-12, IL-6, IL-18, and tumor necrosis factor alpha (4, 7, 17). The production of these cytokines represents an early event in the defense mechanisms against intracellular pathogens such as Brucella spp. (44, 56, 57). It is also noteworthy that CpG also enhanced cytolytic CD8⁺ T cells (5). All of these mechanisms were shown to be involved in protection against Brucella (35).

In this study, we analyzed both the humoral immune response and the cell-mediated immune response induced by these vaccine preparations before assaying their protective efficacies. Since the subclass of IgG response is determined by the pattern of cytokines secreted by CD4 helper T cells, we measured the titers of both the IgG1 and the IgG2a antibodies raised against the relevant recombinant antigens and their abilities to react also against a Brucella extract. Significant levels of IgG1 and IgG2a could be detected only from the sera of mice immunized with P39 mixed with CpG ODN, whereas BFR with CpG ODN induced only IgG2a and to a lower titer. In these conditions, both antigens also induced antibodies able to react with the whole Brucella extracts. Equal amounts of non-CpG ODN mixed with the proteins failed to induce any detectable antibody response. Both antigens with CpG ODN adjuvant induced a response shifted toward IgG2a. This is consistent with the enhancement of IgG2a isotype switching previously reported for CpG ODN adjuvant (6).

This IgG2a isotype is important because the binding of their Fc portion to Fc receptors on the surface of phagocytes activates a broad spectrum of antimicrobial responses (e.g., phagocytosis, cytokine synthesis, release of inflammatory mediators, and generation of reactive oxidant species) (51). Nevertheless the differences in antibody response between mice vaccinated with P39 and mice vaccinated with BFR is striking. No clear explanation can be given, but it is worth noting that the P39 protein was also identified by our group as one of the few Brucella proteins potentially useful for the serological diagnosis of brucellosis (28).

Since the cellular arm of the Th1 response is essential for controlling intracellular pathogens (41), the splenocyte proliferative responses and the cytokines produced after the third immunization were examined. Splenocytes from mice vaccinated with the BFR or P39 proteins with CpG ODN adjuvant were able to proliferate and produced consistent amounts of IFN- when stimulated in vitro with their specific antigens (BFR or P39) or whole Brucella lysate. On the other hand, splenocytes from animals immunized with the recombinant proteins alone or with non-CpG ODN adjuvant did not induce any cellular immune response. Together with the serological data, these results clearly show that BFR or P39 with CpG adjuvant induced a Th1-oriented immune response in BALB/c mice. In addition, this response persisted for up to 12 weeks after immunization (data non shown). Like other studies, our results confirm that CpG ODN are excellent Th1 adjuvants and that the inversion from CpG to GpC eliminates this ability to induce an immune response in vivo (4, 23, 39).

The good and well-oriented immunogenicity of our vaccine preparations prompted us to test their protective efficacy against an infectious B. abortus 544 challenge that was given 3 weeks after the last injection. The protection was evaluated at 4 and 8 weeks after the challenge. The P39 mixed with CpG ODN induced protection in mice when inoculated at doses of 20 µg. On the contrary, mice inoculated with P39 alone or with non-CpG ODN as adjuvant are not protected compared to nonvaccinated mice. On the other hand, BFR with CpG ODN does not protect mice from B. abortus 544 infection. While we identified the BFR as a T dominant antigen and while we showed its ability to induce an appropriate immune response, the BFR protein does not appear to have an important role in the protective immunity. Other Brucella antigens behave in the same way, e.g., the 18-kDa outer membrane protein of B. abortus, which induced a good immune response but was not involved in mediating protective immunity (52). In addition, the combination BFR protein with P39 in CpG ODN did not increase the level of protection against B. abortus 544 (data not shown). As expected, the immunization of mice with recombinant proteins (BFR or P39) alone or with the nonstimulatory GpC ODN alone failed to induce a protective response.

Since it was shown that cytokines elicited by CpG ODN given alone could prevent the early spread of an intracellular pathogen such as Listeria monocytogens and Leishmania major (27, 45, 59), we were surprised that no level of protection could be detected in the group of mice that received 20 µg of CpG ODN three times. Reasons for the absence of adjuvant effect could be the route of immunization used here (i.m.) or more probably the longer interval between the last immunization and the challenge (3 weeks) compared to the conditions used in previous reports (up to 2 weeks). Elkins et al. have reported that the protection against an intracellular pathogen is optimal several days after DNA treatment and persists for about 2 weeks (14).

Altogether, these data indicate that the protection observed here is well linked to the specific combination of a good antigen (P39) and an adequate adjuvant (CpG ODN). None of them used alone was efficient.

Mice vaccinated with B. abortus B19 (used as a vaccinal control) were protected both at 4 weeks p.i. (2.45 logs) and at 8 weeks p.i. (2.92 logs). Surprisingly, the same level of protection (2.48 logs) was observed with the P39-CpG ODN vaccine 4 weeks after the challenge. To our knowledge, this is the first description for murine brucellosis of a subunitary vaccine offering a protection level similar to a live attenuated vaccine. In fact, at 4 weeks the protection level conferred by the L7/L12 protein with adjuvant was 1.5 logs less than the protection induced by the live vaccine, and no data were reported for the protection level at 8 weeks postchallenge (36). With regards to the SOD peptides, the protection was only studied at 2 weeks postchallenge and was not compared to a live vaccine but only to salt-extractable proteins (47). We think that the potency of the P39 candidate antigen is linked to the protocol of its identification, which was based on the careful selection of the more potent Brucella T antigen among those contained in the Brucellin (INRA) using three parallel models (DTH in infected guinea pigs, T-cell proliferation, and IFN- production from PBMC from infected cattle).

Nevertheless the protection induced by the CpG adjuvanted P39 is not comparable either in quality or in duration to the protection conferred by the B19 live vaccine, which was more effective at 8 weeks (2.92 logs) than the subunitary vaccine (1.21 logs).

The challenge by the infectious Brucella can only boost the anti-P39 humoral and cellular immunity, which appeared to be sufficient to confer a short-term protection (4 weeks) but is much less able to generate a long-term protection. In contrast, the immunity induced by the live smooth Brucella B19 strain involves not only protective cell-mediated immunity (1) against a panel of Brucella T-cell epitopes but also humoral responses against the LPS O chain and a variety of proteins. These antibodies were demonstrated as partially protective (18, 29, 31, 55). This immunity linked to the use of a live vaccine is less "monotone" than the one induced by the P39-CpG ODN, and this could be part of the explanation for the vanishing of the protection at 8 weeks p.i.

Other factors may also be involved, such as differences in antigen presentation. It is well known that major histocompatibility complex type I-dependent CD8⁺ cells are important for optimal resistance to Brucella infection (44). These kinds of effector cells could well be more easily induced by a live attenuated vaccine that is still able to replicate intracellularly than by a recombinant protein even with CpG ODN as adjuvant. Finally, the antigen persistence is also quite different, since the dose used here in BALB/c mice the vaccinal strain B19 is still detectable in the spleen 6 weeks after the vaccination (46) and is thus more prone to offer a long-term protection than is the protein with CpG adjuvant. These later points could be optimized by using other delivery systems for the P39 antigen. Naked DNA vaccine combines both the CpG adjuvant effect and an intracellular and persistent expression of the antigen. This is currently under investigation in our laboratory.