INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The whole-cell pertussis vaccine, as a component of the diphtheria-pertussis-tetanus toxoid (DPT) vaccine, has been shown to be very effective in protecting humans against Bordetella pertussis infection (35, 38). However, the reactogenicity of the whole-cell preparations (5) has led to resistance against vaccination campaigns, and some countries interrupted pertussis immunization, resulting in outbursts of pertussis infection (36). This stimulated the development of new acellular vaccines composed of one or more immunogenic components of the B. pertussis organism (24, 28, 35, 38). Several comparative studies on phase II efficacy trials have demonstrated that acellular pertussis vaccines are efficient and less reactogenic than the whole-cell DPT vaccine (12, 15). On the other hand, although acellular vaccines are a significant improvement on DPT vaccination, they still require multiple doses to achieve maximum efficiency and involve high-cost production. The development of a low-cost pertussis vaccine that immunizes efficiently with only one dose would be particularly important for developing countries, where difficult access to health centers hinders attaining complete vaccination of children.

Mycobacterium bovis BCG is considered a high-potential candidate as host for the presentation of heterologous antigens in the development of new live recombinant vaccines (13, 14). A multivalent vaccine based on recombinant BCG (rBCG) would have several advantages: (i) one dose would be sufficient to confer long-lasting cellular immunity; (ii) the World Health Organization recommends that BCG be administered at birth; (iii) it is the best-known adjuvant in animals and humans; and (iv) it has low production costs and is thermostable. Genes encoding immunodominant bacterial, viral, and parasitic antigens have been successfully expressed in BCG (1, 20, 25, 41, 42, 47). Although the expression levels of heterologous genes varied considerably with the expression vector and the gene used, many rBCG strains induced significant levels of humoral and/or cellular responses in vaccinated mice, some of which showed protection against challenge (1, 19, 25, 41). Recent developments in expression vectors have provided several options for the construction of rBCG strains based on different promoters and signal sequences. Besides the widely used hsp60 promoters (42, 47), pAN, pBlaF*, and the 18-kDa, 19-kDa and 85A antigen promoters, have been used to drive the expression of foreign genes (9, 19, 22, 43, 44, 46). The genes have been cloned under their native form or fused to mycobacterial exportation signals to direct the heterologous antigens to different BCG compartments. The upregulated M. fortuitum -lactamase promoter, pBlaF*, has been shown to be much stronger than other promoters; when its signal sequence was included, exportation of the heterologous protein was observed (21, 43).

A pertussis vaccine based on rBCG could improve general immunization, since BCG can be administered at birth and requires only one dose to induce long-lasting immunity. Most developing countries use BCG for immunization against tuberculosis. rBCG expressing pertussis antigens could induce protective immunity against B. pertussis infection. Pertussis toxin (PT) is the most important antigen characterized so far for B. pertussis (45) and is a component of all approved and commercially available acellular vaccines (12, 15, 38). Its diverse activities in animals led it to be considered the main factor responsible for the clinical manifestations of pertussis infection, suggesting that PT would itself be a pertussis vaccine candidate (34). It is composed of five subunits; S1 is the active domain, and S2 to S5 arrange to form the binding domain (40). S1 has been shown to be immunogenic and protective in experimental animals (23, 29). The gene for the S1 subunit has been modified by site-directed mutagenesis to eliminate its toxic properties, generating PT-9K/129G (33). This gene has been used to produce a B. pertussis whole-cell vaccine containing nontoxic PT or to compose acellular DTP (DTaP) vaccines (27, 28). Here we show that a recombinant BCG strain expressing the S1 subunit of PT (S1PT) in fusion with the M. fortuitum -lactamase signal sequence induces cellular and proliferative responses against PT and protects immunized mice against an intracerebral (i.c.) challenge with live B. pertussis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Animals and immunizations. Male 4-week-old Swiss or BALB/c mice (Instituto Butantan) were used for evaluation of the immune response against the rBCG strains. Mice were immunized intraperitoneally (i.p.) with 10⁶ CFU/0.5 ml of BCG or rBCG strain expressing S1PT. Positive and negative controls were DPT (Butantan, São Paulo, São Paulo, Brazil) at 0.8 of the human dose and saline, respectively. The animals were bled at 2 weeks or every 4 weeks after the first dose. A booster was given to BALB/c mice at 10 weeks under the same conditions.

Bacterial strains, growth conditions, and vaccine preparation. All cloning steps were performed in Escherichia coli DH5 grown in Luria-Bertani medium supplemented with ampicillin (100 µg/ml) or kanamycin (20 µg/ml). The M. bovis BCG Moreau strain was used to generate the rBCG strains. Liquid cultures of BCG strains were regularly grown in Middlebrook 7H9 medium supplemented with albumin-dextrose-catalase enrichment (Difco, Detroit, Mich.), with or without kanamycin (20 µg/ml) at 37°C, using stationary tissue culture flasks. The rBCG strains were cultured in Ungar's medium (17) for the heterologous protein localization assays.

Plasmid vectors. pUC18 was purchased from New England Biolabs (Beverly, Mass.). pUC-9K/129G, comprising the genetically detoxified PT gene cloned in the EcoRI site of pUC18, was generated as described previously (33). The mycobacterial expression vectors pLA71 and pLA73 were previously described (22). These vectors contain the E. coli and mycobacterium origins of replication, a kanamycin resistance gene, the upregulated M. fortuitum pBlaF*, its ATG initiation codon, and a multicloning site, which places the heterologous gene in fusion with either the -lactamase signal sequence or the whole -lactamase-encoding gene in pLA71 or pLA73, respectively.

Construction of the S1PT expression vector. To construct pNL71S1 and pNL73S1, a ~700-bp fragment containing the genetically detoxified S1PT gene was amplified by PCR from pUC-9K/129G, using primers 5'-TAGTAGTCTAGAGCGGCCGCCTAGAACGAATACGCGATGCT-3' containing XbaI (italic) and NotI (underlined) restriction sites and 5'-TAGTAGTCTAGAGGTACCGGACGACGATCCTCCCGCCACC-3' containing XbaI (italic) and KpnI (underlined) restriction sites (Bio-Synthesis, Inc., Lewisville, Tex.). The PCR fragment was then digested with XbaI and cloned into XbaI-restricted pUC18. The 720-bp KpnI/NotI fragment containing the S1 gene was then inserted into the pLA71 and pLA73 shuttle vector previously digested with KpnI and NotI. The resulting plasmids, pLN71S1 and pLN73S1 (Fig. 1), were electrotransfected into BCG, and transformants were selected by their resistance to kanamycin.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Schematic representation of the promoter and antigen regions of the shuttle vectors pNL71S1 and pNL73S1. Both vectors contain the E. coli and mycobacterial origins of replication and the kanamycin resistance gene. T, omega terminator; ssBlam, -lactamase initiation codon and signal sequence. S1PT is placed in fusion with ssBlam in pNL71S1 and with Blam, the whole -lactamase gene, in pNL73S1.

Western blotting. One or more kanamycin-resistant BCG clones were grown in 50-ml Middlebrook liquid cultures supplemented with kanamycin (20 µg/ml). The cells of 25-ml cultures were harvested at mid-log phase by centrifugation, washed once with 5 ml of Tris-EDTA, resuspended in 0.5 ml of Tris-EDTA, and disrupted on ice for 3 min with an Ultrasonic Processor GE 100 at half-maximum constant output. Protein concentration in the culture lysates was determined by the Bio-Rad protein assay, using bovine serum albumin (BSA) as a standard. Approximately 50-µg aliquots of protein extracts were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (10% gel). The proteins were then electrotransferred onto a nitrocellulose membrane (Protran; Schleicher & Schuell), and the membrane was saturated with 5% nonfat dry milk in PBS containing 0.1% (vol/vol) Tween 20 (Sigma, St. Louis, Mo.) (PBS-T). The presence of S1PT was detected using a mouse polyclonal antiserum (1:1,000) raised against PT detoxified by formaldehyde treatment (dPT; kindly provided by H. and Y. Sato, National Institute of Health, Tokyo, Japan); the immunoblots were developed with goat anti-mouse peroxidase-conjugated antibodies (1:1,000; Sigma) and visualized with an Amersham ECL kit.

Localization of heterologous proteins in rBCG. Clones of rBCG-S1PT expressing the heterologous protein were grown in 30-ml cultures of Ungar medium supplemented with kanamycin (20 µg/ml). The cells were harvested at mid-log phase by centrifugation. The proteins from the culture supernatants were concentrated by ultrafiltration (Centricon 3; Amicon Corp., Danvers, Mass.). The cell pellet was resuspended in PBS, adjusting cell density to equivalent values, and sonicated for 6 min as described above. Membranes were solubilized by the addition of 2% (vol/vol) Triton X-114. Insoluble material (cell wall-enriched fraction) was separated by centrifugation at 27,000 × g, and the supernatant was submitted to detergent phase partitioning, separating the membrane and cytosol fractions, as described elsewhere (31). Samples from each fraction were submitted to SDS-PAGE and Western blotting as described above.

ELISA. Serum antibody responses to S1PT in pooled sera of immunized animals were quantitated by enzyme-linked immunosorbent assay (ELISA). Briefly, Polysorp 96-well plates (Nunc International, Rochester, N.Y.) were coated with dPT (100 µl; 2 µg/ml in bicarbonate-carbonate buffer, pH 9.6; 4°C overnight), washed three times with PBS-T, and then blocked with 5% nonfat dry milk in PBS. The plates were then incubated with serial dilutions of mouse serum in PBS-1% BSA at 37°C for 1 h. The plates were washed as described above and incubated with peroxidase-conjugated goat anti-mouse immunoglobulin G (Sigma) in PBS-1% BSA at 37°C for 1 h. Following washing, antibodies were visualized by adding OPD substrate (0.04% o-phenylenediamine [OPD] in citrate phosphate buffer [pH 5] containing 0.01% H₂O₂). After color development (10 min), the reaction was interrupted with 8 M H₂SO₄, and the A₄₉₂ was determined. Absorbance values were plotted against dilutions.

Preparation of spleen cells. Spleens were removed from two Swiss mice 2 weeks after immunization with rBCG-pNL71S1 or the saline and BCG controls, and a single-cell suspension was prepared in RPMI 1640 medium (Sigma) supplemented with 2 mM L-glutamine (Merck, Darmstadt, Germany), 50 µg of gentamicin sulfate (Schering-Plough, Rio de Janeiro, Rio de Janeiro, Brazil) per ml, and 10 mM HEPES buffer (Sigma). The cells were washed once and treated with distilled water to remove erythrocytes. The cells were washed twice and resuspended in complete medium: RPMI 1640, as above, plus 10% fetal calf serum (FCS) (Cultilab; Campinas, São Paulo, Brazil) and 0.05 mM 2-mercaptoethanol (Pharmacia LKB Biotechnology AB, Uppsala, Sweden). Cells (2 × 10⁶/ml) were cultured in 24-well tissue culture plates (Corning Glass Works, Corning, N.Y.) in a volume of 1 ml per well, with dPT (1 µg/ml) as the antigen-specific stimulation and concanavalin A (5 µg/ml; Sigma) as a positive control for cell reactivity. Control cells were incubated only in complete medium, without antigen. Cultures were incubated for 3 days in a humidified 5% CO₂ incubator at 37°C. The culture supernatants were harvested at 24, 48, and 72 h, in order to determine the optimum time for lymphokine secretion, and stored at 20°C for further quantitation of gamma interferon (IFN-) and interleukin-4 (IL-4).

Lymphokine assays. IFN- and IL-4 levels in the supernatants of stimulated cells were quantitated by ELISA as described elsewhere (7). Purified rat anti-mouse IFN- (2 µg/ml) and anti-mouse IL-4 (1 µg/ml) monoclonal antibodies (PharMingen, San Diego, Calif.) were used as capture antibodies, coated overnight at 4°C in flat-bottom 96-well microtiter plates (Immunoplates-Maxisorp; Nunc International) in a volume of 50 µl of 0.1 M NaHCO₃, pH 8.2. Washes between each step (three to six times) were performed with PBS containing 0.05% Tween 20 (Sigma). The plates were blocked with PBS containing 10% FCS overnight at 4°C. Lymphocyte culture supernatants were added to the plates without dilution in quadruplicate, and the plates were incubated for 3 h at room temperature. Recombinant mouse cytokines IFN- and IL-4 (PharMingen) were used for the generation of the standard curve. Biotinylated rat anti-mouse IFN- (1 µg/ml) and anti-mouse IL-4 (0.5 µg/ml) monoclonal antibodies (PharMingen) were added in a volume of 50 µl of PBS containing 0.05% Tween 20 (Sigma) per well and incubated for 1 h at room temperature. Bound antibodies were detected by use of ExtrAvidin-horseradish peroxidase conjugates (Sigma). A dye-substrate buffer containing 0.2% (wt/vol) OPD and 0.015% (vol/vol) H₂O₂ diluted in 24.3% (vol/vol) 0.1 M citric acid-25.7% (vol/vol) 0.2 M Na₂HPO₄ was added and allowed to react for 15 to 30 min. Color development was interrupted by addition of 4.5 M H₂SO₄, and A₄₉₂ was determined. Sample concentrations from standard curves were obtained by nonlinear regression analysis (GraphPad Prism, version 2.0).

Proliferation assay. Spleen cells were harvested from 72-h cultures and prepared in RPMI 1640 supplemented with 2% FCS. The proliferative responses were determined on the basis of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Sigma) conversion to formazan as described elsewhere (11), with minor modifications. In brief, MTT solution (20 µl, 5 mg/ml) was added to 100 µl of medium and incubated for 3 h at 37°C. A solution of PBS containing 10% SDS and 0.01 M HCl (100 µl) was added, and the mixture was incubated for another 18 h. Color formation was measured by the A₅₉₅. One proliferative unit is equivalent to one optical density unit at 595 nm, and the value obtained for saline-inoculated mice was subtracted. Statistical analysis was performed by the Student-Newman-Keuls multiple-comparisons test.

B. pertussis i.c. challenge. Immunized mice were inoculated i.c. with 30 µl of a bacterial suspension of B. pertussis strain 18323 (18,500 or 30,000 CFU, depending on weight), and survival was monitored twice daily for 17 days. Mice not surviving the first 3 days were considered to have been improperly inoculated. The level of protection afforded was expressed as the percentage of surviving mice.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Expression and localization of S1PT in rBCG. We have expressed the nontoxic S1 subunit of PT-9K/129G in BCG, under control of the upregulated M. fortuitum pBlaF*, in fusion with either the -lactamase signal sequence (using pLA71) or the whole -lactamase gene (using pLA73), leading to rBCG-pNL71S1 or rBCG-pNL73S1, respectively (Fig. 1). Mycobacterial cell extracts from transformed BCG clones were submitted to immunoblot analysis using a polyclonal anti-PT antibody. As shown in Fig. 2, three clones of rBCG transformed with pNL71-S1 expressed the ~30-kDa fusion protein comprising the -lactamase signal sequence and S1PT. The free S1 subunit can be seen as the slower-migrating band with 26.2 kDa from purified PT. An rBCG clone transformed with pNL73S1 expressed the 57-kDa -lactamase-S1PT fusion protein at significantly higher levels, although even in this case it was not possible to detect the heterologous protein in Coomassie blue-stained SDS-polyacrylamide gels (comigration with major mycobacterial proteins may have hindered detection). The fusion protein produced by pNL73S1 showed few lower-molecular-weight bands, probably due to some degree of fusion protein degradation. Neither BCG nor BCG transformed with pLA73 showed any immunoreactive bands. Fractionation by Triton X-114 detergent phase partitioning showed that the heterologous proteins produced from both rBCG-pNL71S1 and rBCG-pNL73S1 were localized to the cell wall (results not shown). Very small amounts of extracellular S1 fusion proteins were detected in the culture supernatant, and none was found in the membrane fraction or cytosol.

View larger version (68K): [in this window] [in a new window]   FIG. 2.   Expression of S1PT in rBCG. Cell extracts of BCG or rBCG (50 mg) were analyzed by Western blotting using an anti-PT polyclonal antibody. Lane 1, BCG; lane 2, rBCG-pLA71; lanes 3 to 5, rBCG-pNL71S1; lane 6, rBCG-pLN73S1; lane 7, purified PT from wild-type B. pertussis.

In vivo stability of rBCG-S1PT. Swiss mice were inoculated i.p. with 10⁶ CFU of rBCG-pNL71S1 per animal. One and two months later, animals were sacrificed, spleens were separated and homogenized, and spleen extracts were plated in Middlebrook 7H10 in the presence or absence of kanamycin (20 µg/ml). Stability of the expression vector was determined as a ratio of kanamycin-resistant colonies to colonies obtained in the absence of antibiotics. At 1 month, the ratio had decreased to 0.86, and at 2 months it was still 0.85. The total number of colonies recovered from the spleens of inoculated animals at 8 weeks was approximately 530 for both BCG and rBCG-pNL71S1.

Humoral responses to rBCG-S1PT. Groups of five male BALB/c mice were immunized i.p. with 10⁶ CFU of rBCG-pNL71S1 or rBCG-pNL73S1. Control mice were immunized with saline, BCG, or DPT. A booster under the same conditions was administered at 10 weeks. Sera was collected every 4 weeks after immunization, pooled, and tested by ELISA for the induction of antibodies against dPT. As shown in Fig. 3, mice immunized with rBCG-pNL71S1 showed a twofold increase in anti-PT antibodies in relation to saline or BCG controls at 2 months from the initial vaccination. Antibody levels then decreased steadily, reaching levels comparable to the control level at 5 months. Neither rBCG-pNL73S1 nor DPT induced a major increase in anti-PT antibodies. However, in a short-term experiment in which the humoral response was monitored at weekly intervals, a twofold increase in anti-PT antibodies was observed at 1 week in rBCG-73S1PT-immunized animals, which was decreased by 2 weeks (results not shown). Also in this experiment, again DPT showed no humoral response under the same conditions (0.8 of the human dose), but when we used a lower dose (0.12 of the human dose), an antibody response comparable to that obtained by rBCG-73S1PT was observed (not shown). At 6 months, these immunized animals were challenged i.c. with live B. pertussis. The only animals surviving at 13 days postchallenge were two of five in the group vaccinated with rBCG-pNL71S1, and one of five in the group receiving rBCG-pNL73S1. The last surviving animal immunized with DPT died on the last day. This experiment demonstrated that the i.c. challenge is not appropriate to evaluate protection in older mice and indicated that the rBCG-S1PT vaccines may have induced some level of protection. Since rBCG-pNL71S1 also induced higher antibody levels, we investigated the induction of a cellular response by this strain.

View larger version (39K): [in this window] [in a new window]   FIG. 3.   Antibody response induced by rBCG-S1PT. BALB/c mice were immunized i.p. with rBCG-S1 strains rBCG-pNL71S1 and rBCG-pNL73S1 (106 CFU/0.5 ml) or the positive (DPT) and negative (saline [Sal] or BCG) controls, receiving a booster dose under the same conditions at 10 weeks as described in Materials and Methods. Sera were collected at the indicated times, pooled, and analyzed for anti-PT antibodies by ELISA. A492 of sera diluted 1:20 is shown.

Lymphokine production induced by rBCG-S1PT. Swiss mice were inoculated with rBCG-pNL71S1 or the saline and BCG controls. Two weeks later, spleen cells were isolated for quantification of IFN- and IL-4 production following stimulation with dPT. The in vitro lymphokine secretion profiles of dPT-stimulated spleen cells revealed marked differences among the groups (Fig. 4). Both IFN- and IL-4 levels peaked at 48 h in culture (not shown). Figure 4 shows that splenocytes from rBCG-S1PT-immunized mice induced the secretion of high levels of IFN- when stimulated with dPT. Cells derived from BCG-immunized mice also showed an increase in IFN- production, although to a lower degree. IL-4 secretion was also increased in the splenocytes from BCG-immunized mice, and the expression of S1PT decreased IL-4 secretion to control levels. On the whole, IFN- levels were higher than IL-4 levels.

View larger version (16K): [in this window] [in a new window]   FIG. 4.   PT-specific IL-4 and IFN- production in spleen cells derived from BCG- and rBCG-pNL71S1-immunized mice. Cytokine content was estimated as described in Materials and Methods after stimulation with dPT. The error bars indicate standard deviations. Sal, saline.

Proliferative response against PT. Spleen lymphocytes isolated from rBCG-pNL71S1-immunized mice showed 50% proliferation in response to stimulation with dPT compared to saline controls (P < 0.001) or 25% in relation to BCG-immunized controls (P < 0.05), as shown in Fig. 5. The proliferative response correlated with the cellular response, and BCG alone also induced some degree of proliferation in response to the stimulus with dPT.

View larger version (18K): [in this window] [in a new window]   FIG. 5.   Proliferation of spleen cells derived from BCG- and rBCG-pNL71S1-immunized mice upon PT stimulation. Proliferation 72 h after stimulation with detoxified PT was measured as MTT conversion to formazan as described in Materials and Methods. The values obtained for saline (0.88 ± 0.10) were subtracted from those obtained by BCG and rBCG-pNL71S1. The error bars indicate standard deviations, which were lower than 10% of the obtained values. Student's test: *, P < 0.01 compared with BCG; **, P < 0.001 compared with saline.

Protection against an i.c. challenge with live B. pertussis. Groups of five male Swiss mice were immunized i.p. with one of the following vaccine preparations: saline, BCG, rBCG-pNL71S1, rBCG-pNL73S1, and DPT. At day 14, all groups were challenged i.c. with a previously determined mean lethal dose of live B. pertussis under the same conditions as used to certify the potency of commercial whole-cell pertussis vaccines. As shown in Table 1, rBCG-pNL73S1, expressing S1PT in fusion with the whole -lactamase protein, induced complete protection of mice against challenge, comparable to that obtained with the commercial DPT vaccine. Animals immunized with rBCG-pNL71S1, expressing S1PT in fusion with the -lactamase signal sequence, also induced a high level of protection. BCG-immunized mice showed a slightly higher survival than the saline control group or a delay in the time of death. Previous experiments using saline, BCG, and rBCG-pLA71 showed that BCG and the vector strain were equivalent and that occasionally they induced low-level protection, consistent with the immunostimulatory properties of BCG (not shown).

View larger version (13K): [in this window] [in a new window]   FIG. 6.   Protection of mice immunized with rBCG-S1PT against B. pertussis i.c. challenge. Groups of 10 Swiss mice immunized i.p. 2 weeks before with rBCG-pNL71S1 (106 CFU/0.5 ml; squares), the negative control receiving 500 µl of saline (triangles), and the positive control receiving 400 µl of DPT (circles) were challenged i.c. with 1 mean lethal dose of live B. pertussis. Survival was monitored over 17 days.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Several attempts have been made to develop improved vaccines against B. pertussis. Since PT is considered the major immunogen of B. pertussis, most systems tested are based on PT, genetically detoxified or not (12, 28, 38). Following cloning and sequencing of the five subunits of PT, it was genetically detoxified by site-directed mutagenesis of the active subunit S1, generating PT-9K/129G. Mutant B. pertussis expressing this inactive PT showed immunogenicity equivalent to the whole-cell chemically detoxified vaccine (33), and purified PT-9K/129G was comparable to the acellular PT vaccine (27). Attempts to express the five subunits of PT in E. coli required fusion of the subunits to other proteins (29). A recombinant Salmonella enterica serovar Typhimurium aroA strain expressing the five subunits of PT failed to protect mice from a virulent challenge, probably due to incomplete processing of the subunits (8). Of the five subunits composing the PT holotoxin, S1 has been characterized as the most immunogenic moiety (10), and passive immunization with an anti-S1PT monoclonal antibody protected mice against a virulent challenge (39). The S1 subunit has been expressed in fusion or unfused and has been purified from recombinant E. coli or Bacillus subtilis, maintaining its activity and immunochemical properties (3, 23, 30, 37). Fusion of genetically detoxified S1PT to tetanus toxin fragment C (TTFc) has been shown to increase expression of the fusion protein in E. coli (6). With the growing evidences on the advantages of recombinant vaccines based on live vectors, the S1 subunit has been expressed alone or in fusion with TTFc in different vaccine strains of Salmonella, invasive E. coli, Streptococcus gordonii, and BCG, inducing protective antibodies or a cellular response against PT (2, 4, 21).

In this study we have constructed rBCG strains expressing the detoxified S1 subunit of PT under control of the upregulated pBlaF* in fusion either with the -lactamase exportation signal (BCG-pNL71S1) or with the whole -lactamase protein (BCG-pNL73S1). Expression levels of the S1 fusion proteins in rBCG-pNL73S1 were considerably higher than with rBCG-pNL71S1 (Fig. 2), confirming the importance of protein fusion in the expression levels of S1PT. On the other hand, fusion with smaller fragments of -lactamase would be more convenient for vaccine purposes. The fusion proteins were relatively stable, showing few degradation products in rBCG-pNL73S1. Both strains directed the fusion proteins to the cell wall, with very small amounts in the extracellular media, indicating that exportation was initiated but probably blocked by the thick cell wall components. Expression of S1PT in fusion with TTFc has previously been obtained in rBCG, using the 85A antigen promoter and signal sequence or the hsp60 promoter, the latter of which induced a higher level of expression (2). Fusion to the 85A signal sequence had not attained exportation of the heterologous protein, which was attributed to its large size. Although the -lactamase signal sequence has been shown to export other recombinant proteins expressed in BCG, and the fusion protein in BCG-pNL71S1 has a reduced size, complete exportation was not achieved, indicating that the size of the expressed protein is not the only factor hindering its translocation through the mycobacterial cell wall.

Besides the level of antigen expression and localization, another factor influencing the induction of an efficient immune response in live vaccine vectors is the stability of the expression system. Integrative vectors are considered to display lower expression but higher stability. On the other hand, plasmids are considered to provide higher expression levels but to be less stable in vivo. Although many mycobacterial expression systems comprising different antigens have been investigated, few have been characterized in relation to their in vivo plasmid stability. We have inoculated mice with rBCG-pNL71S1 and monitored mycobacterial recovery and kanamycin resistance for up to 2 months. At 1 month, the number of kanamycin-resistant colonies was 86% of that recovered in the absence of antibiotics. At 2 months, 85% of the BCG colonies recovered from the spleen were still resistant to kanamycin, indicating that this expression vector was relatively stable. The total number of rBCG clones recovered from the spleen was comparable to that obtained from nonrecombinant BCG-inoculated mice, showing that the transfected expression vector did not alter BCG persistence in mice.

BALB/c mice immunized with the rBCG-S1PT strains produced low levels of serum anti-PT antibodies, as well as the whole-cell DPT vaccine, when followed up to 6 months (Fig. 3). At 15 days from immunization with rBCG-S1PT strains or with DPT, Swiss mice showed no anti-PT antibodies (not shown). These experiments are in agreement with previous results suggesting that the humoral response against PT may not be the most important component for induction of protection (32), since the conventional DPT whole-cell vaccine is considered very efficient. On the other hand, increasing evidence suggests the importance of the induction of a cellular response against PT for protection against B. pertussis (26, 32). Previous studies have shown that rBCG expressing the S1-TTFc fusion protein under control of the hsp60 promoter induces a weak but significant cellular response against PT, as determined by the antigen-specific secretion of IL-2 (38). In this study, spleen cells obtained from mice immunized 2 weeks before with rBCG-pNL71S1 showed a significant increase in production of IFN- and in the proliferative response upon stimulation with detoxified PT compared to the saline and BCG controls (Fig. 4 and 5). On the whole, IFN- levels were higher than IL-4 levels, suggesting induction of a Th1-dominant response by these rBCG-S1PT strains. This cytokine pattern correlates with that observed in patients convalescing from whooping cough (32), and B. pertussis infection is known to provide a more efficient protection than vaccination.

Mice immunized with rBCG-S1PT strains displayed high-level protection against an i.c. challenge with live B. pertussis (Table 1; Fig. 6), which correlated with the induction of a PT-specific cellular immune response (IFN- production) and proliferation (Fig. 4 and 5). These results reinforce the importance of the induction of cell-mediated immunity in the protection against B. pertussis infection. Our results show that the pBlaF* and -lactamase signal sequences in rBCG can drive adequate expression levels and presentation of heterologous antigens to the immune system. A more effective vaccine against B. pertussis should also include antigens involved in colonization, and we are investigating the expression of complementary proteins in rBCG.

DPaT vaccines have been shown to be efficient and less reactogenic than whole-cell DPT vaccines, but the high costs of processing render them expensive, which will hinder widespread use in developing countries. BCG is used as a vaccine against tuberculosis in most developing countries, in which it is an important control mechanism. Safer and more efficient mycobacterial vaccines, addressing the problem of use in immunocompromised individuals, may be developed in the near future (15, 17); some of these may be used as carriers for heterologous antigens in much the same way as BCG. Combined rBCG-pertussis vaccines administered once, separate from the DT vaccination, would be a significant improvement over the currently used BCG-DPT and BCG-DaPT schedules. Even better would be a combined rBCG-DPT vaccine; we are currently working on the expression of tetanus and diphtheria antigens in rBCG.