c mice

Microbes and Infection 5 (2003) 85–93 www.elsevier.com/locate/micinf

Original article

The recombinant Omp31 from Brucella melitensis alone or associated with rough lipopolysaccharide induces protection against Brucella ovis infection in BALB/c mice Silvia M. Estein a, Juliana Cassataro b, Nieves Vizcaíno c, Michel S. Zygmunt d, Axel Cloeckaert e, Raúl A. Bowden a,* a

Laboratorio de Inmunoquímica y Biotecnología, Departamento de Sanidad Animal y Medicina Preventiva, Facultad de Ciencias Veterinarias, UNICEN, 7000 Tandil, Argentina b Laboratorio de Inmunogenética, Hospital de Clínicas José de San Martín”, Facultad de Medicina, UBA, Buenos Aires, Argentina c Departamento de Microbiología y Genética, Universidad de Salamanca, 37007 Salamanca, Spain d Laboratoire de pathologie infectieuse et immunologie, Centre de recherches de Tours, Institut national de la recherche agronomique, 37380 Nouzilly, France e Unité de pathologie aviaire et parasitologie, Centre de recherches de Tours, Institut national de la recherche agronomique, 37380 Nouzilly, France Received 23 August 2002; accepted 15 October 2002

Abstract Immunogenicity and protective activity against Brucella ovis of detergent-extracted recombinant Omp31 (rOmp31 extract) from Brucella melitensis produced in Escherichia coli, purified rough lipopolysaccharide from B. ovis (R-LPS) and a mixture of rOmp31 extract and R-LPS (rOmp31 extract + R-LPS) were assessed in BALB/c mice. The experimental vaccines were compared with a hot saline extract (HS extract) from B. ovis mainly composed of outer membrane proteins (OMPs) and R-LPS, and known to be protective in mice against a B. ovis infection. Serum antibodies to Omp31 and R-LPS were detected in the corresponding mice using Western blotting with B. ovis whole-cell lysates and ELISA with purified antigens. Protection was evaluated by comparing the levels of infection in the spleens of vaccinated mice challenged with B. ovis. A significantly lower number of B. ovis colony-forming units in spleens relative to unimmunized (saline injected) controls were considered as protection. Mice immunized with rOmp31 extract or rOmp31 extract mixed with R-LPS developed antibodies that bound to the B. ovis surface with similar titers. Vaccination with rOmp31 extract plus R-LPS provided the best protection level, which was comparable with that given by HS extract. Similar protection was also obtained with rOmp31 extract alone and, to a lesser degree, with R-LPS. Comparisons between groups showed that an extract from E. coli-pUC19 (devoid of Omp31) provided no protection relative to either HS extract, rOmp31 extract or rOmp31 extract mixed with R-LPS. In conclusion, the recombinant Omp31 associated or not with B. ovis R-LPS, could be an interesting candidate for a subcellular vaccine against B. ovis infection. © 2003 Éditions scientifiques et médicales Elsevier SAS. All rights reserved. Keywords: Brucella; Vaccine; Outer membrane protein

1. Introduction Sheep can suffer from infection by two Brucella species, B. melitensis (which expresses smooth lipopolysaccharide, S-LPS) and B. ovis (which only expresses rough lipopolysaccharide, R-LPS). B. ovis causes a disease characterized by epididymitis and subfertility in rams and by placentitis and abortion in ewes, leading to severe economic losses [1,2]. * Corresponding author. Tel./fax: +54-2293-42-2357. E-mail address: [email protected] (R.A. Bowden). © 2003 Éditions scientifiques et médicales Elsevier SAS. All rights reserved. DOI: 1 0 . 1 0 1 6 / S 1 2 8 6 - 4 5 7 9 ( 0 2 ) 0 0 0 7 5 - 8

Control measures in flocks include vaccination when prevalence is high. Among live attenuated vaccines, B. melitensis Rev. 1, a smooth strain used to control B. melitensis infection in small ruminants, gives heterologous protection against B. ovis and is considered the best available vaccine for the prophylaxis of ovine brucellosis [3]. However, it elicits antibody responses interfering with serological diagnosis, is virulent for humans and its use is not allowed in countries free of B. melitensis. Regarding killed vaccines, the homologous bacterins of B. ovis with different adjuvants provided poor protection to

86

S.M. Estein et al. / Microbes and Infection 5 (2003) 85–93

rams [4]. In addition, they led to the development of local lesions and, as with live Rev. 1, intense antibody responses affecting the specificity of serodiagnosis [4]. Subcellular vaccines could avoid the drawbacks of live attenuated vaccines, being safer and not interfering with immunodiagnosis, provided that good protective antigens, different from those used for testing, are selected. A hot saline extract (HS extract) from B. ovis was shown to be protective in mice against a B. ovis infection [5], and to be as effective as the B. melitensis Rev. 1 vaccine in protecting rams against B. ovis [6]. In addition, transfer to naive mice of immune serum raised to HS extract demonstrated an important involvement of antibodies specific for HS extract molecules in protective immunity [5]. HS extract is a complex mixture of outer membrane molecules, mainly outer membrane proteins (OMPs) and R-LPS, and is especially rich in group-3 OMPs [7,8], which includes Omp25 and Omp31, the major OMPs of Brucella [reviewed in 9]. In addition to these OMPs, other proteins, namely Omp1, Omp2b as well as three lipoproteins, namely Omp10, Omp16 and Omp19, have been identified in HS extract [10]. The R-LPS, which is the only LPS form in rough Brucella, has been shown to induce low levels of protection against B. ovis infection in mice [5]. However, an HS extract depleted of R-LPS, was less protective than total HS extract in rams [6]. In addition, three monoclonal antibodies (mAbs) specific for R-LPS afforded protection to mice [10,11]. Accordingly, the R-LPS might be important for protective immunity against B. ovis by contributing to antibody-mediated immunity and/or by enhancing the protective activity of OMPs present in HS extract. Passive protection experiments in mice have shown that mixtures of mAbs, previously shown to bind individually to surface-exposed epitopes of several B. ovis OMPs and R-LPS, are protective against a B. ovis infection [10]; moreover, protection was better when an anti-Omp31 mAb was present in the mixture of mAbs [10,11] and this mAb, administered alone, conferred passive protection as strong as that obtained with an anti-HS extract sera. The cognate epitope of this mAb has been located on the most hydrophilic loop of Omp31 [12]. These features attracted our attention to Omp31 as one potential component for a subcellular vaccine against B. ovis. Problems were found in obtaining purified Brucella spp. OMPs free of other molecules, making the study of their genuine protective value as immunogens difficult. Indeed, although Omp25 and Omp31 have been extracted from B. ovis free of R-LPS with Triton X-114 followed by preparative SDS-polyacrylamide gel electrophoresis (PAGE), both proteins could not be successfully separated [13]. Molecular cloning techniques have made the isolation of individual antigens for testing as vaccines easier. B. melitensis Omp31 has been cloned and expressed on the surface of E. coli [12], which provides a way to obtain the protein free of other Brucella components. In addition, the cognate epitope of the protective mAb [10,11] is present in both B. melitensis and B.

ovis Omp31 [14]. In this paper, we examine the immunogenicity in mice of immunogens made of B. melitensis Omp31, detergent-extracted from recombinant E. coli and B. ovis R-LPS either alone or combined. The protective activity of these immunogens against B. ovis infection was also analyzed in mice.

2. Materials and methods 2.1. Bacteria and plasmids B. ovis virulent strains PA-76250 (PA, for short) and REO198 were obtained from the Brucella culture collection (INRA-Nouzilly, France). B. ovis PA was used as challenge strain after two serial passages in BALB/c mice and reisolation from spleens. Bacterial cultures were prepared as described previously [10]. For infection, contents of freshly thawed vials were diluted with phosphate-buffered saline (PBS) to the desired concentration [5]. Exact numbers of cells were established retrospectively. For immunization, the HS extract from B. ovis REO 198 was prepared as recommended [15]. Construction of plasmid pNV3123, which corresponds to the omp31 gene of B. melitensis 16M cloned into pUC19 (Pharmacia Biotech, Uppsala, Sweden), and expression of the Omp31 gene in recombinant E. coli JM109 strains have been described elsewhere [12,16]. E. coli (pUC19) was used as negative control. To check for Omp31 expression, the recombinants were subjected to SDS-PAGE and Western blotting with the anti-Omp31 mAb A59/10F09/G10 [12,16]. As expected, Omp31 was detected in E. coli (pNV3123) and not in E. coli (pUC19). These recombinant E. coli cells were used for obtaining the detergent extracts for immunizations. For antibody determinations in ELISA, a different construction for obtaining rOmp31 was created. A 687-bp B. melitensis DNA fragment encoding Omp31 [12] devoid of the putative signal peptide was cloned in Pet22+ vector (Novagen, Madison, WI, USA). Briefly, the sequence information previously reported [12] was used to design specific primers for the Omp31 with NdeI and XhoI restriction sites at the 5' ends. The primers were as follows: p1-5'TCCGCTCATATGGCCGACGTGGTTGTT3'. p2-5'GTTCGTCTCGAGGAACTTGTAGTTCAG3'. B. melitensis genomic DNA was used as template for PCR with Pfu DNA polymerase (Stratagene, La Jolla, CA, USA). The resultant plasmid (Pet-OMP31) contained the Omp31 gene, with the addition of a sequence for a poly-H tail. Competent E. coli BL21 (DE3) (Stratagene) was transformed with Pet-OMP31. Ampicillin-resistant colonies were grown until OD600 = 1.0 in LB medium containing 100 µg ml–1 of ampicillin, at 37 °C with agitation. Five milliliters of this culture were diluted to 500 ml and grown to reach an OD600 of 1.0. At this point, protein expression was induced by adding 1 mM IPTG and further incubating the transformed cells for 4 h. Bacteria were pelleted by centri-

S.M. Estein et al. / Microbes and Infection 5 (2003) 85–93

fuging (15 000g, 20 min, at 4 °C) and lysed by sonication in 1% Triton X-100, 5 mM EDTA, 50 mM Tris, pH 8.0. Inclusion bodies, containing Omp31, were recovered by centrifugation and solubilized (Tris 50 mM, EDTA 5.0 mM, urea 8.0 M, pH 8.0) during 24 h. After centrifuging (20 000g, 30 min at 4 °C), soluble protein was purified by chromatography through Ni-agarose (Quiagen, Dorking, UK). The presence of rOmp31 in eluates was checked by western blotting with specific mAb A59/10F09/G10 [17]. Purity was assessed by SDS-PAGE and Coomassie blue staining. This preparation contained no surface antigens of E. coli, and therefore was more suited for the antibody detection to rOmp31 in ELISA. 2.2. Detergent extraction of recombinant Omp31 used for mouse immunization Extraction/separation of recombinant B. melitensis Omp31 with a nonionic detergent was carried out as previously described [18] with minor modifications. Briefly, 1 g of each, E. coli (pNV3123) and E. coli (pUC19), was suspended in 10 ml of ice-cold Tris-buffered saline pH 7.0 containing 4% Triton X-114 (Sigma) and 20 mM phenylmethane-sulfonyl fluoride (Merck, Darmstaad, Germany). The mixtures were shaken overnight at 4 °C and centrifuged at 10 000g at 4 °C for 30 min. Supernatants of each mixture were heated at 37 °C for 20 min and centrifuged at 10 000g for 20 min at 25 °C. E. coli (pNV3123) and E. coli (pUC19) detergent layers were recovered and designated recombinant Omp31 extract” (rOmp31 extract) and E. coli (pUC19) extract”, respectively. Protein was precipitated by adding three volumes of cold acetone (–20 °C) [19] and recovered by centrifugation. Residual acetone was removed from the precipitate by evaporation under vacuum. Protein quantifications were performed by the bicinchoninic acid method (BCA Protein Assay, Pierce, Rockford, IL, USA). The rOmp31 extract and pellets of extracted E. coli cells were analyzed by Western blotting with mAb A59/10F09/G10, anti-E. coli (pUC19) and anti-HS extract mouse immune sera. The rOmp31 extract contained predominantly Omp31. No bands in the region corresponding to this protein were visible in the E. coli (pUC19) extract. When rOmp31 extract was reacted with an anti-E. coli serum, additional bands at approximately 14, 21, 48, 66 and 55 kDa were detected (data not shown). Protein concentrations in the extracts were relatively low, (0.12 mg ml–1 and 0.01 mg ml–1 for rOmp31 extract and E. coli (pUC19) extract, respectively).

87

ELISA was a gift from Dr. A. Vigliocco (CNEA, Argentina) and its preparation has already been described [21]. 2.4. Indirect ELISAs All ELISAs were carried out in 96-well polystyrene plates (NUNC 2-69620, Denmark). Blood samples were obtained by retroorbital puncture under ether anesthesia at 0, 21, 37 and 60 days after first immunization. Antibodies from immunized mice were titered in an indirect ELISA on whole B. ovis [10]. In addition, antibody specificity of individual sera from mice, at 60 days after the first immunization, was tested by indirect ELISA against R-LPS from B. ovis and purified rOmp31 from recombinant E. coli. 2.4.1. ELISA for B. ovis surface-binding antibodies B. ovis PA cells were grown for 48 h, harvested, washed and heat-inactivated (65 °C, 60 min). Threefold serum dilutions (1/30 to 1/177,147) in 0.05% Tween-20, PBS (PBS-T) were incubated for 2 h at 37 °C. Bound antibodies were detected by using a goat anti-mouse (whole-molecule) horseradish peroxidase conjugate (1/2,500; 1 h at 20 °C) (Sigma). Color was read at 405 nm after incubating 30 min at room temperature with a solution of 2,2'-azino-di-(3ethylbenzothiazoline-sulphonic acid) (ABTS)-H2O2, with continuous shaking. Titers were calculated by interpolating from logarithmic regression curves at an absorbance value of 1.0. 2.4.2. ELISA for R-LPS antibodies The plates were sensitized by incubation at 22–24 °C for 18 h with B. ovis R-LPS diluted in 0.06 M sodium carbonate buffer (pH 9.6) (1 µg ml–1, 50 µl well–1). Blocking was done with 150 µl of 1% bovine serum albumin (BSA) (Sigma) in PBS-T for 1 h at 37 °C. Sera (1/50 dilution in PBS-T-0.5% BSA, 50 µl well–1) was incubated for 1 h at room temperature. Bound antibodies were detected with rabbit anti-mouse IgG (whole-molecule) (1/2000, 50 µl well–1). After 1 h at 20 °C, the plate was further incubated for 1 h with protein A-peroxidase conjugate (1/2000, 50 µl well–1). The plates were shaken for 1 h at room temperature before read-out. The same substrate solution was used as that for the B. ovis ELISA. 2.4.3. ELISA for Omp31 antibodies The plates were sensitized with recombinant Omp31 purified from E. coli (Pet-Omp31) (10 µg ml–1, 50 µl well–1) overnight at 4 °C. The remaining steps were as for ELISA for R-LPS antibodies.

2.3. R-LPS preparations

2.5. SDS-PAGE and Western blotting

For inclusion in experimental vaccines, R-LPS was obtained by the phenol-ether-chloroform method from B. ovis REO 198 [20]. Residual protein in the extracts was measured by BCA. The R-LPS antigen from B. ovis used in the indirect

Specificity of serum antibodies from immunized mice was assessed by Western blotting onto B. ovis PA whole-cell lysate. B. ovis was boiled for 10 min in sample buffer, sonicated briefly and centrifuged. The supernatant was

88

S.M. Estein et al. / Microbes and Infection 5 (2003) 85–93

loaded onto a 12% acrylamide-bisacrylamide, and run and blotted as described previously [10]. Strips were incubated overnight with pooled sera from immunized mice diluted 1/50. The detection was done by incubating strips successively in rabbit anti-mouse immunoglobulin G diluted 1/500 and protein A-peroxidase (both from Sigma) diluted 1/2000 and developed as in [10]. mAbs to Brucella spp. R-LPS (A68/24G12/A08) and to Brucella spp. Omp31 (A59/10F09/G10), used as controls for antigen position, were incubated at a 1:500 dilution and detected as indicated for immune sera. 2.6. Immunizations and protection test Female BALB/c 5–6 weeks old (purchased from INTA, CICV-Castelar, Argentina) were distributed at random in groups of eight mice each and acclimated for 1 week before starting immunizations. Three mouse groups were injected intraperitoneally twice, at a 30-days interval, with 0.2 ml of rOmp31 extract (20 µg), E. coli (pUC19) extract (20 µg) and R-LPS (20 µg dry weight). Two other groups of mice were injected with mixtures of rOmp31 extract and R-LPS (rOmp31 extract-R-LPS) or E. coli (pUC19) extract and R-LPS (E. coli (pUC19) extract-R-LPS), respectively. These mixtures, prepared by sonication, contained the same amount of each immunogen as that used in the individual immunogen vaccines. The two remaining groups of mice were injected with 20 µg of HS extract (positive control group) or with PBS (negative control group). The HS extract was selected for immunization of the positive controls because it represents the best experimental subcellular immunogen available in terms of protective activity against B. ovis infection in mice [5]. Sera were obtained by retroorbital bleeding under ether anesthesia the day of the first immunization and 21, 37 and 60 days afterwards. Fifteen days after the booster, mice were inoculated i.v. with 1.6 × 104 colony-forming units (CFU) of B. ovis PA 76250. Mice were killed at 30 days after the challenge and their spleens were removed aseptically, weighed, and kept frozen (–20 °C) until further processing. To determine the infection level, spleens were thawed and individually homogenized in an appropriate volume of PBS in sterile plastic bags, serially diluted (ten-fold) and each dilution was seeded onto two plates of TSAYE medium. After 5 days of incubation in a 10% CO2 atmosphere, CFU were counted and expressed by the log10 per spleen value. The spleen infection in each mouse group was expressed as the log10 mean ± SD of B. ovis CFU of the mice included in each group. The weight of the spleen was determined, as it has been considered as an indicator of immunity against Brucella [22]. 2.7. Statistical analysis In ELISA and protection tests, the statistical significance of differences between treatments was calculated by ANOVA and Dunnett’s and Tukey’s post-hoc test using InStat soft-

Fig. 1. Antigens recognized by immune sera from mice immunized with subcellular vaccines investigated by Western blotting. Whole-cell lysates of B. ovis PA were analyzed in 12% SDS-PAGE and blotted. Blots were reacted with lane 1, mAb A59/10F09/G10 (specific to Omp31); lane 2, mAb A76/24G08/A09 (specific to R-LPS); lane 3, pooled sera from mice immunized with rOmp31 extract; lane 4, the same, immunized with E. coli (pUC19) extract; lane 5, the same, immunized with B. ovis R-LPS; lane 6, the same, immunized with rOmp31 extract + B. ovis R-LPS; lane 7 the same, immunized with E. coli (pUC19) extract + B. ovis R-LPS and lane 8, the same, injected with PBS. The position of antigens B. ovis Omp31 and B. ovis R-LPS recognized by the mAbs is indicated on the left.

ware (Graph-Pad, CA, USA). Graphs were done with Microcal Origin 6.0 software (Microcal Inc., MA, USA) and Prism 3.0 (Graph-Pad, CA, USA). The data passed the test for normality, so the only mathematical transformation used was the conversion of arithmetic values to log10 values. A correlation analysis between levels of serum antibody to B. ovis and protection was done using Prism 3.0 (Graph-Pad, CA, USA).

3. Results 3.1. Antibody response specific for B. ovis Omp31 and R-LPS induced in mice by the immunogens The capacity of mice immunized with the immunogens to produce antibodies to both Omp31 and R-LPS was investigated by Western blotting (Fig. 1) using a B. ovis PA wholecell lysate as antigen. Pooled sera collected 60 days after the first immunization from mice injected with the B. melitensis rOmp31 extract (lane 3) or purified B. ovis R-LPS (lane 5) recognized bands at 31 or 13 kDa, respectively, which appeared at the same molecular mass level as the bands revealed with the anti-Omp31 mAb (lane 1) and the anti-RLPS mAb (lane 2). Sera from animals immunized with rOmp31 extract mixed with B. ovis R-LPS revealed those two bands (lane 6), whereas mice immunized with E. coli

S.M. Estein et al. / Microbes and Infection 5 (2003) 85–93

89

The same sera were also investigated in ELISA using purified B. ovis R-LPS as coating antigen (Fig. 2, lower panel). Mice injected with E. coli (pUC19) extract or with saline exhibited low, baseline absorbances (below 0.2 units) similarly to animals immunized with rOmp31 extract alone. Mice immunized with B. ovis R-LPS presented the highest response to this antigen, significantly different from both control groups (P < 0.001) (i.e. saline and E. coli (pUC19) extract). In addition, in mice injected with rOmp31 extract plus R-LPS or with E. coli (pUC19) extract plus R-LPS, the intensity was similar to that obtained with R-LPS alone, and still significantly different from control groups (P < 0.001). However, reactivity was significantly lower for mice injected with E. coli (pUC19) extract plus R-LPS, and even lower in mice vaccinated with B. ovis HS extract, although the absorbance values in both groups of mice were significantly higher (P < 0.001) than those obtained for saline control mice. 3.2. Evaluation of antibody response to B. ovis PA in ELISA

Fig. 2. Antibody response to Omp31 and R-LPS. Antibodies to both antigens were determined in sera collected from each immunization group at day 60 after the first injection. Sera were diluted 1/50 and assayed against B. melitensis Omp31 purified from recombinant E. coli (pET-Omp31) (upper panel) or R-LPS extracted from B. ovis (lower panel). Data are expressed as mean ± SD absorbance units, and when appropriate, statistically significant differences in the activity of sera from control (PBS injected) group are marked by asterisks.

(pUC19) extract mixed with B. ovis R-LPS (lane 7) detected only the latter. Sera both from the mice immunized with control E. coli (pUC19) extract alone (lane 4) and those injected with PBS (lane 8) did not reveal any band. To measure the level of the antibody response specific for Omp31, sera from immunized mice, collected 60 days after the first injection, were tested in ELISA against B. melitensis recombinant Omp31 purified from E. coli (Pet-Omp31) (Fig. 2, upper panel). This time point was chosen as it was the nearest to the challenge moment. Mice injected with either saline, R-LPS, or E. coli (pUC19) extract with or without R-LPS, behaved similarly, exhibiting low, background absorbances. Mice immunized with rOmp31 extract presented significantly higher reactivity (P < 0.001) than controls; animals immunized with rOmp31 extract with R-LPS showed even higher absorbances, but they were not statistically different from those obtained with sera from mice immunized with the rOmp31 extract alone. Finally, mice injected with B. ovis HS extract also had significant (P < 0.001) antibody titers to recombinant Omp31, but lower than those found in mice immunized with rOmp31 extract either with or without R-LPS.

In order to evaluate the ability of antibodies induced by the immunogens to bind to B. ovis, sera from the seven groups of immunized mice were titered in an indirect ELISA using B. ovis PA whole-cells as antigen. Antibody IgG titers were determined at 0, 21, 37 and 60 days after the first immunization and expressed as mean ± SD of the individual log10 titers from each group of mice. At day 21 postimmunization, the level of B. ovis-specific antibodies in mice immunized with rOmp31 extract, R-LPS, rOmp31 extract with R-LPS and HS extract increased significantly in comparison to that obtained in the negative control group of mice injected with saline (P < 0.01). The four groups of mice showed similar levels of B. ovis-specific antibodies (2.33 log10, 1.94 log10, 2.61 log10 and 2.88 log10, respectively). Mice immunized with E. coli (pUC19) extract alone or mixed with B. ovis R-LPS showed lower reactivity, but titers were significantly higher than those of preimmune samples (P < 0.05) (1.19 log10 and 1.15 log10, respectively). At day 37 postimmunization, i.e. seven days after the booster, antibody responses in mice immunized with rOmp31 extract or rOmp31 extract with R-LPS reached similar titers (3.61 log10 and 3.32 log10, respectively) which were slightly lower that those obtained in mice immunized with HS extract (4.34 log10) (Fig. 3). Immunization with B. ovis R-LPS alone elicited lower responses (2.33 log10); higher however, than that detected in mice immunized with E. coli (pUC19) extract with R-LPS (1.41 log10). Sixty days postimmunization, antibody titers specific for B. ovis surface antigens further increased in mice vaccinated with HS extract (4.9 log10), rOmp31 extract (4.0 log10) and rOmp31 extract with R-LPS (3.9 log10) (Fig. 3). However, antibody levels in mice immunized with R-LPS remained stable. B. ovis-specific antibody titers in mice immunized with E. coli (pUC19) extract or E. coli (pUC19) extract with R-LPS remained low throughout the study. The rise in antibodies observed in sera from mice given E. coli (pUC19) extract could be due to cross-reactivity between

90

S.M. Estein et al. / Microbes and Infection 5 (2003) 85–93

Fig. 3. Antibody response to B. ovis surface-exposed antigens. Mice were first immunized i.p. at day 0 and boosted on 30 days with immunogens. Serum samples were obtained at 0, 21, 37 and 60 days. Serum antibodies (IgG) were measured by ELISA onto heat-killed whole-cells of B. ovis PA. Plotted values represented the mean and SD (n = 8) of serum titers obtained by interpolating from logarithmic linear regression curves at an absorbance value (405 nm) of 1.0.

E. coli and B. ovis antigens. Although such antigens could not be detected by Western blotting (Fig. 1) a higher analyte sensitivity of ELISA could account for the observed reactivity. Alternatively, denaturation in Western blotting could have affected the antigenicity of molecules, thus explaining their lack of reactivity. 3.3. Protective activity of the immunogens against a B. ovis infection in mice Mice were first immunized with the immunogens and then challenged with B. ovis PA as described in Section 2. Protec-

tive activity of the immunogens against a B. ovis infection was evaluated by measuring the number of B. ovis CFU in spleen 30 days after infection. Results are expressed as the mean ± SD of the log10 CFU per organ for the mice included in each treatment group; in addition, the confidence interval for each group is given in Table 1. B. ovis infection in nonimmunized mice led to a moderate splenomegally (mean spleen weight = 0.24 g) in spite of a high bacterial burden (6.27 ± 0.10 log10 CFU). Although mice immunized with rOmp31 extract, rOmp31 extract plus R-LPS and HS extract had smaller spleens (range 0.14–0.19 g; P > 0.05). This effect was only significant for B. ovis R-LPS (0.14 ± 0.02 g; P < 0.01) and E. coli (pUC19) extract plus R-LPS immunogens (0.13 ± 0.03; P < 0.01). In comparison with nonimmunized control mice (Dunnett’s test) as measured by the reduction in the B. ovis CFU in spleen, immunization with the rOmp31 extract provided significant protection against B. ovis PA challenge (2.06 log10 protection units; P < 0.01) (Table 1). Purified B. ovis R-LPS vaccine was also protective but less so than rOmp31 extract (1.63 log10 protection units, P < 0.01). Immunization with rOmp31 extract plus B. ovis R-LPS provided a high level of protection (2.26 log10 protection units; P < 0.01) close to that given by HS extract (2.57 log10 protection units; P < 0.01). The E. coli (pUC19) extract did not induce protection, whereas E. coli (pUC19) extract associated with R-LPS did (1.21 log10 protection units; P < 0.01) to a similar extent as the B. ovis R-LPS immunogen alone. We further wished to compare the protective activity between immunogens. This was done using Tukey’s test. When the activity of the best control immunogen (i.e. HS extract), with either immunogens rOmp31 extract, R-LPS, or rOmp31 extract plus B. ovis R-LPS, was compared no significant differences were detected. In contrast, the activity of HS extract was found significantly different (P < 0.001) from that exhibited by E. coli (pUC19) extract, although the comparison of E. coli (pUC19) extract plus R-LPS was nonsignificant (P > 0.05). Finally, whereas the protection provided by rOmp31 extract was significantly greater than that resulting from vaccination with E. coli (pUC19) extract

Table 1 Immunogenicity and protective activity in BALB/c mice against B. ovis by immunogens containing rOmp31 extract Immunogen a ROmp31 extract E. coli (pUC19) extract R-LPS ROmp31 extract + R-LPS e E. coli (pUC19) extract + R-LPS e HS extract Saline

Preinfection antibody titer b (mean ± SD) 2.93 ± 0.04 1.85 ± 0.16 2.72 ± 0.06 3.54 ± 0.14 2.50 ± 0.12 3.89 ± 0.07 0.69 ± 0.04

Splenic infection c, d log10 CFU/spleen (mean ± SD) 4.21 ± 0.10 *** 5.97 ± 0.10 4.64 ± 0.09 *** 4.01 ± 0.74 *** 5.06 ± 0.08*** 3.70 ± 0.11 *** 6.27 ± 0.10

Mean protection (CI) 2.06 (1.91–2.20) 0.30 (0.17–0.42) 1.63 (1.50–1.76) 2.26 (2.13–2.38) 1.21 (1.07–1.34) 2.57 (2.44–2.70) –

Mice were vaccinated twice over a 4-week interval and challenged i.v. with 1.6 × 104 CFU of B. ovis PA 6 weeks after the second vaccination. Antibody (IgG) titers as measured by ELISA onto whole B. ovis PA cells and expressed in log10 units (arithmetic group mean ± SD). c Four weeks after challenge mice were killed. Results are shown as log10 of CFU counts (group mean ± SD). d Differences between group values were estimated by anova and Dunnett’s test. e Extracts and R-LPS were mixed by sonication. *** P < 0.001. a

b

S.M. Estein et al. / Microbes and Infection 5 (2003) 85–93

(P < 0.001), it did not differ from that of R-LPS alone or E. coli (pUC19) extract plus R-LPS (P > 0.05).

4. Discussion In the present study, we investigated the immunogenicity and protective activity against B. ovis of B. melitensis Omp31, alone or associated with B. ovis R-LPS, in a mouse model. This was achieved by the use of the Omp31 protein synthesized in E. coli, which gave a preparation free of any contaminating Brucella antigen. This approach has been used previously to assess the immunogenicity and protective activity of both B. melitensis Omp25 [22] and B. melitensis Omp31 [16] against B. melitensis infection in mice. However, whereas in those reports whole recombinant E. coli synthesizing the B. melitensis OMP was used, in the present one, Omp31 was purified from recombinant E. coli taking advantage of the OMP-extracting ability of nonionic detergent TX-114 [13–18]. This gave preparations with much fewer contaminating molecules, and enriched in Omp31. R-LPS, a glycolipidic molecule easier to purify than OMPs, was obtained from B. ovis by the traditional extractive phenol-chloroform-ether method. Mice immunized with rOmp31extract produced IgG antibodies to this protein, as detected by Western blotting with a B. ovis lysate and ELISA with purified recombinant Omp31 (Figs. 1 and 2). Additionally, Omp31-specific antibodies were able to bind on the surface of B. ovis, as determined by ELISA with whole B. ovis cells (Fig. 3), which indicates that recombinant Omp31 was given to mice in a way that preserved the antigenicity of surface-exposed epitopes. The binding of OMP-specific antibodies has been associated with the accessibility of the epitopes/molecule at the surface of B. ovis [23]. In addition, a relationship between accessibility and protective activity was found, by immunization of mice with recombinant E. coli, for Omp25 in B. melitensis [22] and by passive protection experiments with anti-Omp25 mAbs in B. ovis [10,11]. In the present study, the level of IgG antibodies to B. ovis, irrespective of their specificity (i.e. anti-rOmp31, anti-R-LPS or anti-HS extract), correlated significantly with protective activity (r = 0.95 ± 0.32; P < 0.001) (data not shown), which is analogous with results obtained with smooth B. abortus infection in different mice strains [24]. Mechanisms mediating the protective effect of antibodies able to bind to B. ovis cell surface have not been elucidated, but preliminary experiments suggest that some antibodies specific to Brucella OMPs and R-LPS can mediate in-vitro lysis by complement (Estein et al., unpublished). Recombinant B. melitensis Omp31 extract administered alone or with B. ovis R-LPS induced protection in mice against a B. ovis infection. The protection level reached in spleen with both immunogens (about 2 log10 protection units (P < 0.01)) was very close to that obtained with HS extract (Table 1). Similar protection levels against B. ovis have been obtained by others with HS extract or delipidated HS extract

91

as vaccines [5] in an experiment with a comparable level of splenic infection in non-vaccinated mice. Accordingly, Omp31 emerges as a main protective immunogen of the HS extract vaccine against B. ovis infection in mice. However, both HS extract [5] and recombinant B. melitensis Omp31 [16] vaccines were ineffective against B. melitensis. Most probably, this could be attributed to (i) the presence of O-polysaccharide-bearing S-LPS in B. melitensis, shown to hinder deeper OMP and core-LPS epitopes in Brucella [17,26] and (ii) a predominant role of HS extractspecific antibodies relative to T cells as effectors of immunity to B. ovis [5]. The enhancement of protection by R-LPS cited by others [5,6], and marginally reflected by our results, might be due to different features. (i) It might result from an adjuvant effect of this molecule to the antibody response [25]; however, the antibody titers against whole B. ovis cells did not significantly differ (P > 0.05) between groups of mice immunized with rOmp31 extract alone or with R-LPS (Fig. 3). (ii) An adjuvant effect of R-LPS at the cellular level cannot be excluded, as part of the protective immunity against B. ovis in mouse has been shown to be transferable by T cells [5]. (iii) Appropriate folding of OMPs and exposure of critical epitopes influence their antigenicity [26]. Thus probably, an integration of Omp31 into vesicle-like particles, including amphipathic molecules from E. coli and the R-LPS from B. ovis, might also lead to a better immunogenicity of the complex, as strong and homologous as well as heterologous associations between B. ovis porins and LPS were also demonstrated [8,27]; supporting this, denatured B. melitensis Omp25, which had lost the binding capacity for some specific mAbs, has been refolded in vitro by mixing it with R-LPS which allowed the recovery of discontinuous B epitopes [26]. Although a similar experiment has not been done with Omp31, molecular analysis and epitope mapping with mAbs of this protein indicated the presence of conformational epitopes [14]. (iv) Antibodies to R-LPS might also account for protection. Indeed, mice produced antibodies to R-LPS when administered alone or in combination either with Omp31 extract or E. coli (pUC19) extract (Figs. 1 and 2). Their protective activity was reflected by the lower number of CFU, when R-LPS was injected alone or with E. coli (pUC19) extract, which by itself was non-protective. This possibility is supported by the fact that three mAbs specific to R-LPS were shown to afford protection in mice against B. ovis [10,11]. Although others did not find a protective effect for R-LPS against B. ovis infection in mice [5], R-LPS has been suggested to protect against B. ovis in rams [6]. Therefore, it will be interesting to revise the role of R-LPS in protective immunity against B. ovis as well as other ways of better presenting protective epitopes. In this regard, copolymers used as adjuvants for HS extract in mice were shown to greatly enhance its protective activity in both rough B. ovis and smooth B. abortus, thus probably involving protective mechanisms other than antibodies [28]. Thus, for a potential

92

S.M. Estein et al. / Microbes and Infection 5 (2003) 85–93

subcellular vaccine, appropriate adjuvants giving a wellbalanced immune response should be also carefully selected. During the preparation of this manuscript, minor differences in the Omp31 amino acid sequences between B. melitensis and B. ovis have been reported [14]. These differences have been shown to affect the antigenicity/conformation of the proteins, as detected by binding studies with mAbs and sera from B. ovis-infected rams. Thus, although the B. melitensis rOmp31 provides heterologous protection in mice against a B. ovis infection (this work), it would be interesting to determine the protective activity against B. ovis of the homologous recombinant Omp31. To be used in a control program, a subcellular vaccine containing protective antigens should be used together with a sensitive and specific diagnostic test. In the case of ram epididymitis, ELISA tests based on internal (cytoplasmic or periplasmic) antigens such as BLS [29] or bp26 [30] could work well if vaccines were based on outer membrane antigens, such as Omp31 and R-LPS. In conclusion, B. melitensis Omp31 is a protective immunogen against B. ovis in the mouse. Further experiments are needed to study both the cellular and antibody-dependent mechanisms involved in protection and to assess the immunogenicity and protective activity in sheep.

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

Acknowledgements [16]

We thank María R. Ortíz for media preparation and Sergio Islas (CIC), Fabián Amaya and Adrián Pérez (UNICEN, Argentina) for animal care. S.M. Estein was supported by CIC (Argentina). Work supported by UNICEN, and CIC (Argentina). We are indebted to Ana Vigliocco (CNEA, Argentina) for kindly supplying the reagents for ELISA. Work partially funded by UNICEN and CIC, Argentina.

[17]

[18]

References [19] [1] [2] [3]

[4] [5]

[6]

[7]

J.M. Blasco, Brucella ovis, in: K. Nielsen, J.R. Duncan (Eds.), Animal Brucellosis, CRC Press, Boca Ratón, Florida, 1990, pp. 351–378. M.S. Bulgin, Epididymitis in rams and lambs, Vet. Clin. North Am. Food Anim. Pract. 6 (1990) 683–690. C.M. Marín, M. Barberán, M.P. Jiménez de Bagüés, J.M. Blasco, Comparison of subcutaneous and conjunctival routes of Rev 1 vaccination with prophylaxis of Brucella ovis infection in rams, Res. Vet. Sci. 48 (1990) 209–215. B.L. Swift, R. Maki, Immunological studies of three ram epididymitis bacterins, Cornell Vet. 58 (1968) 659–665. M.P. Jiménez de Bagüés, P.H. Elzer, J.M. Blasco, C.M. Marín, C. Gamazo, A.J. Winter, Protective Immunity to Brucella ovis in BALB/c mice following recovery from primary infection or immunization with subcellular vaccines, Infect. Immun. 62 (1994) 632–638. J.M. Blasco, C. Gamazo, A.J. Winter, M.P. Jiménez de Bagüés, M. Barberán, I. Moriyón, B. Alonso-Urmeneta, R. Díaz, Evaluation of whole-cell and subcellular vaccines against Brucella ovis in rams, Vet. Immunol. Immunopathol. 37 (1993) 257–270. C.E. Suárez, G.A. Pacheco, A.M. Vigliocco, Characterization of Brucella ovis surface antigens, Vet. Microbiol. 18 (1988) 349–356.

[20] [21]

[22]

[23]

[24]

C. Gamazo, A.J. Winter, I. Moriyón, J.I. Riezu-Boj, J.M. Blasco, R. Díaz, Comparative analyses of proteins extracted by hot saline or released spontaneously into outer membrane blebs from field strains of Brucella ovis and Brucella melitensis, Infect. Immun. 57 (1989) 1419–1426. A. Cloeckaert, J.M. Verger, M. Grayon, M.S. Zygmunt, N. Vizcaíno, Minireview: molecular and immunological characterization of the major outer membrane proteins of Brucella, FEMS Microbiol. Lett. 145 (1996) 1–8. R.A. Bowden, S.M. Estein, M.S. Zygmunt, G. Dubray, A. Cloeckaert, Identification of protective outer membrane antigens of Brucella ovis by passive immunization of mice with monoclonal antibodies, Microbes Infect. 2 (2000) 481–488. R.A. Bowden, A. Cloeckaert, M.S. Zygmunt, G. Dubray, Outer membrane protein and rough lipopolysaccharide-specific monoclonal antibodies protect mice against Brucella ovis, J. Med. Microbiol. 43 (1995) 344–347. N. Vizcaíno, A. Cloeckaert, M.S. Zygmunt, G. Dubray, Cloning, nucleotide sequence, and expression of the Brucella melitensis omp31 gene coding for an immunogenic major outer membrane protein, Infect. Immun. 64 (1996) 3744–3751. R. Kittelberger, D.S. Diack, N. Vizcaíno, M.S. Zygmunt, A. Cloeckaert, Characterization of an immuno-dominant antigen in Brucella ovis and evaluation of its use in an enzyme-linked immunosorbent assay, Vet. Microbiol. 59 (1998) 213–227. N. Vizcaíno, R. Kittelberger, A. Cloeckaert, C. Marín, L. FernándezLago, Minor nucleotide substitutions in the omp31 gene of Brucella ovis result in antigenic differences in the major outer membrane protein that it encodes compared to those of the other Brucella species, Infect. Immun. 69 (2001) 7020–7026. G.G. Alton, L.M. Jones, R.D. Angus, J.M. Verger, Techniques for the brucellosis laboratory, INRA, Paris, 1988. L. Guilloteau, K. Laroucau, N. Vizcaíno, I. Jacques, G. Dubray, Immunogenicity of recombinant Escherichia coli expressing the omp31 gene of Brucella melitensis in BALB/c mice, Vaccine 17 (1999) 353–361. A. Cloeckaert, P. De Wergifosse, G. Dubray, J.N. Limet, Identification of seven surface-exposed Brucella outer membrane proteins by use of monoclonal antibodies: immunogold labeling for electron microscopy and enzyme-linked immunosorbent assay, Infect. Immun. 58 (1990) 3980–3987. R. Kittelberger, F. Hansen, M.F. Hilbink, G.W. De Lisle, A. Cloeckaert, Selective extraction of bacterial macromolecules by temperatureinduced phase separation in Triton X-114 solution, J. Microbiol. Meth. 24 (1995) 81–92. J.F. Robyt, B.J. White, Biochemical Techniques. Theory and Practice, Wadsworth, Belmont, 1987. C. Galanos, O. Lüderitz, O. Westphal, A new method for the extraction of R lipopolysaccharides, Eur. J. Biochem. 9 (1969) 245–249. A.M. Vigliocco, P.S. Silva Paulo, J. Mestre, G.C. Briones, G. Draghi, M. Tossi, K. Nielsen, Development and validation of an indirect enzyme immunoassay for detection of ovine antibody to Brucella ovis, Vet. Microbiol 54 (1997) 357–368. R.A. Bowden, A. Cloeckaert, M.S. Zygmunt, G. Dubray, Evaluation of immunogenicity and protective activity of the major 25 kDa outer membrane protein of Brucella melitensis expressed in Escherichia coli, J. Med. Microbiol. 47 (1998) 39–48. R.A. Bowden, A. Cloeckaert, M.S. Zygmunt, S. Bernard, G. Dubray, Surface exposure of outer membrane protein and lipopolysaccharide. Epitopes in Brucella species studied by ELISA and flow cytometry, Infect. Immun. 63 (1995) 3945–3952. G.W. Pugh Jr, E. Zehr, V. Meador, M. Phillips, T.J. McDonald, B. Deyoe, Immunologic, histopathologic and bacteriologic responses of five strains of mice to Brucella abortus strain 2308, Am. J. Vet. Res. 50 (1989) 323–328.

S.M. Estein et al. / Microbes and Infection 5 (2003) 85–93 [25] A. Berinstein, M. Pérez-Filgueira, A. Schudel, P. Zamorano, M. Borca, A. Sadir, Avridine and LPS from Brucella ovis: effect on the memory induced by foot-and-mouse disease virus vaccination in mice, Vaccine 11 (1993) 1295–1301. [26] A. Cloeckaert, M.S. Zygmunt, G. Bézard, G. Dubray, Purification and antigenic analysis of the major 25 kDa outer membrane protein of Brucella abortus, Res. Microbiol. 147 (1996) 225–235. [27] I. Moriyón, M.A. Montañés, In vitro interactions between lipopolysaccharides and heterologous outer membrane porin proteins, Curr. Microbiol. 12 (1985) 229–234.

93

[28] M. Murillo, M.J. Grillo, J. René, C.M. Marín, M. Barberán, M.M. Goñi, J.M. Blasco, J.M. Irache, C. Gamazo, A Brucella ovis antigenic complex bearing poly-epsilon-caprolactone microparticles confer protection against experimental brucellosis in mice, Vaccine 19 (2001) 4099–4106. [29] S.M. Estein, P.C. Baldi, R.A. Bowden, Comparison of serological tests based on outer membrane or internal antigens for detecting antibodies to Brucella ovis in infected flocks, J. Vet. Diagn. Invest. 14 (2002) 45–47. [30] M.S. Zygmunt, S. Baucheron, N. Vizcaíno, R.A. Bowden, A. Cloeckaert, Single-step purification and evaluation of recombinant BP26 protein for serological diagnosis of Brucella ovis infection in rams, Vet. Microbiol. 87 (2002) 213–220.