Protective Properties of Rifampin-Resistant Rough Mutants of Brucella melitensis

INFECTION AND IMMUNITY, July 2005, p. 4198 4204 Vol. 73, No. 7 0019-9567/05/$08.00 0 doi:10.1128/iai.73.7.4198 4204.2005 Copyright 2005, American Society for Microbiology. All Rights Reserved. Protective Properties of Rifampin-Resistant Rough Mutants of Brucella melitensis R. Adone,* F. Ciuchini, C. Marianelli, M. Tarantino, C. Pistoia, G. Marcon, P. Petrucci, M. Francia, G. Riccardi, and P. Pasquali Istituto Superiore di Sanità, Dipartimento di Sanità Alimentare ed Animale, Viale Regina Elena 299, 00161 Rome, Italy Received 11 January 2005/Returned for modification 2 March 2005/Accepted 8 March 2005 Vaccination against Brucella infections in animals is usually performed by administration of live attenuated smooth B. abortus strain S19 and B. melitensis strain Rev1. They are proven effective vaccines against B. abortus in cattle and against B. melitensis and B. ovis in sheep and goats, respectively. However, both vaccines have the main drawback of inducing O-polysaccharide-specific antibodies that interfere with serologic diagnosis of disease. In addition, they retain residual virulence, being a cause of abortion in pregnant animals and infection in humans. To overcome these problems, one approach is to develop defined rough mutant Brucella strains lacking O antigen of lipopolysaccharide. B. abortus rough strain RB51, a rifampin-resistant mutant of virulent strain B. abortus 2308, is used as a vaccine against B. abortus infection in cattle in some countries. However, RB51 is not effective in sheep, and there is only preliminary evidence that it is effective in goats. In this study, we tested the efficacies of six rifampin-resistant rough strains of B. melitensis in protecting BALB/c mice exposed to B. melitensis infection. The protective properties, as well as both humoral and cellular immune responses, were assessed in comparison with those provided by B. melitensis Rev1 and B. abortus RB51 vaccines. The results indicated that these rough mutants were able to induce a very good level of protection against B. melitensis infection, similar to that provided by Rev1 and superior to that of RB51, without inducing antibodies to O antigen. In addition, all B. melitensis mutants were able to stimulate good production of gamma interferon. The characteristics of these strains encourage further evaluation of them as alternative vaccines to Rev1 in primary host species. * Corresponding author. Mailing address: Istituto Superiore di Sanità, Dipartimento di Sanità Alimentare ed Animale, Viale Regina Elena 299, 00161 Rome, Italy. Phone: 39 06 49902728. Fax: 39 06 49387077. E-mail: adone@iss.it. Brucellosis is a major zoonotic disease, widely distributed in humans and domestic and wild animals, especially in developing countries. Among the different species of the Brucella genus, B. abortus and B. melitensis are the most pathogenic and virulent, not only for cattle or for sheep and goats, respectively, but also for other animal species. The occurrence of the disease in humans is largely dependent on the animal reservoir, with the highest rate of human infection in areas where rates of brucellosis in sheep and goats are high (6, 35). Brucellosis vaccines are essential elements in control programs. Attenuated B. abortus strain 19 and B. melitensis strain Rev1 are proven effective vaccines; they induce good levels of protection against B. abortus in cattle and against B. melitensis in sheep and goats, preventing premature abortions (7, 30, 31). However, both vaccines have the drawback of inducing O- polysaccharide-specific antibodies that interfere with the discrimination between vaccinated and infected animals during serological screening (7, 31). In addition, they retain pathogenicity and sometimes cause abortion in vaccinated animals (10, 18, 41) and remain infectious for humans (3, 5, 26). The use of the conjunctival route when administering B. melitensis Rev1 vaccine significantly reduces the intensity and duration of serological interfering responses, but the safety and duration of the immunity conferred by this method are still under debate (11). One of several approaches for the development of alternative vaccines is to use live attenuated Brucella strains lacking O antigen (rough strains). The fact that B. abortus 45/20, a rough organism with little or no ability to induce O-chain antibodies, can induce significant protection against infection with B. abortus indicates that rough strains can be used to induce protective immune responses, avoiding the diagnostic problems described above. Unfortunately, B. abortus 45/20 is not totally devoid of an O chain (37, 38), and it tends to revert to the smooth, virulent form when used as a live vaccine (25, 46). Recently, B. abortus strain RB51 has been approved in the United States as a vaccine for bovine brucellosis. This strain, a rough rifampin-resistant B. abortus mutant derived from virulent B. abortus strain 2308, shows negligible interference with serological diagnosis and induces protective immunity in cattle similar to that afforded by B. abortus S19 (9, 34, 43). Data obtained using mice showed that RB51 is able to protect against infections with heterologous Brucella strains, including B. melitensis, B. ovis, B. abortus, and B. suis (19). However, protection against abortion and infection after a challenge with the virulent B. melitensis H38 strain in sheep is less than that provided by the conventional Rev1 vaccine (11). In addition, RB51 vaccine does not confer protection against B. ovis in rams (20). It has been speculated that rough mutants of B. melitensis or B. suis could induce protective immunity which could be superior to that induced by RB51 (40). For this purpose, the disruption mutants B. melitensis VTRM1 and B. suis VTRS1 were constructed using the Brucella wboa gene sequence. These mutants, tested in mice against infection with heterologous and 4198

VOL. 73, 2005 PROTECTIVE B. MELITENSIS MUTANTS 4199 Brucella species TABLE 1. Phenotypic characteristics and mutations identified in the rpob genes of the Brucella strains used in this study Strain Morphology Rifampin phenotype Amino acid(s) affected a Codon alteration(s) Amino acid change(s) B. abortus RB51 Rough Resistant 526 GAC3TAC Asp3Tyr B. melitensis RBM9 Rough Resistant 526, 574 GAC3AAC, CCG3CTG Asp3Asn, Pro3Leu RBM11 Rough Resistant 526 GAC3TAC Asp3Tyr RBM14 Rough Resistant 154 GTT3TTT Val3Phe RBM15 Rough Resistant 526, 539 GAC3AAC, CGC3AGC Asp3Asn, Arg3Ser RBM17 Rough Resistant 541 TCG3TTG Ser3Leu RBM19 Rough Resistant 526, 574 GAC3GGC, CCG3CTG Asp3Gly, Pro3Leu Rev1 Smooth Susceptible a The rpob codon numbering is based on the B. melitensis 16 M rpob numbering system (accession number AE009516). homologous Brucella species, showed good protection in comparison with that afforded by vaccines of killed cells in adjuvant (45). However, strain VTRM1 used as a vaccine in pregnant goats conferred only partial protection against infection and abortion following challenge (12). In a previous study, the genetic bases for rifampin resistance in Brucella spp. were investigated and different rpob genotypes associated with the rifampin resistance phenotype were determined (24). In order to develop defined rough vaccine strains, in this study we evaluated the efficacies of six rifampin-resistant rough mutants of B. melitensis with different rpob genotypes in protecting mice exposed to B. melitensis infection. The protective properties of these strains as well as both humoral and cellular immune responses were assessed in comparison with those afforded by B. melitensis Rev1 and B. abortus RB51 conventional vaccines. MATERIALS AND METHODS Animals. For this study, 8- to 9-week-old female BALB/c mice, weighing 21 to 25 g, were used. Mice were provided by Harlan Italy (Udine, Italy). They were maintained in barrier housing with filtered inflow air in a restricted-access room and under pathogen-limited conditions. They were fed a commercial diet, and water was provided ad libitum. All mice were acclimatized for a minimum of 1 week prior to experimentation. Bacterial strains and growth conditions. The following Brucella strains were used: (i) six rough rifampin-resistant (Rif r ) mutants of B. melitensis, named RBM9, RBM11, RBM14, RBM15, RBM17, and RBM19, showing different rpob gene sequences as indicated in Table 1 and produced as described below; (ii) B. melitensis vaccine strain Rev1, provided by the Veterinary Laboratories Agency of Weybridge, United Kingdom; (iii) B. abortus vaccine strain RB51, provided by the Cooper-Zeltia Veterinaria, S.A., Spain; and (iv) B. melitensis strain 16 M, for challenge exposure, provided by the Veterinary Laboratories Agency. All Brucella strains were cultured at 37 C on brucella agar medium (Oxoid Ltd., Hampshire, England) supplemented with 5% horse serum (BAS). Production and characterization of B. melitensis Rif r mutants. Rif r rough mutants were derived from two rifampin-susceptible (Rif s ) laboratory rough strains of B. melitensis according to published procedures with minor modifications (39). Briefly, parental Rif s strains were grown on BAS containing concentrations of rifampin rising from 40 to 400 g/ml. All clones obtained were inspected for roughness by crystal violet staining and by autoagglutination reaction with acriflavine solution (1/1,000) as described by Alton et al. (4). As described previously (24), all clones were inspected for the presence of mutations in the sequence of the rpob gene, which encodes the subunit of the DNAdependent RNA polymerase, the target of rifampin in prokaryotes (22, 23). Phenotypic and genetic characteristics of the rough strains are shown in Table 1. Phenotypic stability of B. melitensis rough mutants. Prior to testing as vaccines, B. melitensis rough strains were tested in vitro and in vivo to verify the tendency to revert to a virulent smooth phenotype. These strains were passaged 40 times at 37 C on BAS plates every 5 days of growth. After each passage, colonies were inspected for roughness as described above. In addition, colonies of each strain were inoculated intraperitoneally (i.p.) into BALB/c mice. Ten days later, the mice were killed by cervical dislocation and colonies were isolated from the spleen, inspected for the rough phenotype, and inoculated again into mice. This procedure was repeated three times. Preparation of cultures for vaccination and challenge experiments. All Brucella strains were cultured on BAS. In order to inoculate the same number of CFU for each strain, the exact concentration of all suspensions was determined by viable counts of 10-fold dilutions following incubation on BSA plates for 48 h. During this period, suspensions were kept at 4 C; in previous experiments we observed that no significant decrease of viable organisms occurs during this period under these storage conditions. After determination of counts, cultures were adjusted to the desired concentration. For the vaccination, suspensions of RB51 and B. melitensis mutant strains were adjusted to a concentration of 5 10 8 CFU/ml, while the Rev1 suspension was adjusted to a concentration of 5 10 5 CFU/ml. For challenge exposure, B. melitensis 16 M bacteria, previously inoculated into BALB/c mice to enhance their virulence, were adjusted to a concentration of 10 6 CFU/ml. Vaccination. BALB/c mice were divided into nine groups of 25 animals each. As described in previous studies (14, 19), groups I to VII were inoculated i.p. with 0.2 ml of sterile saline containing 10 8 CFU of strains RBM9, RBM11, RBM14, RBM15, RBM17, RBM19, and RB51, respectively. Mice of group VIII were inoculated subcutaneously with 0.2 ml of sterile saline containing 10 5 CFU of Rev1 vaccine. Group IX received saline and was kept as negative control (Table 2). Bacterial clearance. To evaluate the bacterial clearance, five mice of groups I to VII were bled and sacrificed at 7, 15, and 40 days postvaccination (dpv). At the same time, five mice from a group inoculated i.p. with 10 7 CFU of B. melitensis 16 M were sacrificed. Spleens of mice, removed aseptically, were weighed and homogenized in PBS. An aliquot was plated for Brucella organism detection in order to evaluate the bacterial clearance, while another aliquot was used to determine the gamma interferon (IFN- ) production as described below. Bacteriological and serological examinations were performed as described below. Protection assay. As indicated by the Office International des Epizooties for quality control of smooth live anti-brucella vaccines (33), the protective activity was evaluated by the immunogenicity test, which compares the abilities of mice TABLE 2. Protection of mice against B. melitensis 16 M infection a Group Vaccine Dose (CFU)/route Log 10 CFU Brucella per spleen (mean SD) Log 10 units of protection I RBM9 10 8 /i.p. 4.36 1.16 1.94 II RBM11 10 8 /i.p. 5.28 0.66 1.02 III RBM14 10 8 /i.p. 3.70 1.37 2.60 IV RBM15 10 8 /i.p. 4.43 1.29 1.87 V RBM17 10 8 /i.p. 3.47 1.07 2.83 VI RBM19 10 8 /i.p. 5.20 0.72 1.10 VII RB51 10 8 /i.p. 5.17 0.33 1.13 VIII Rev1 10 5 /s.c. b 3.80 0.17 2.50 IX Control Saline/i.p. 6.30 0.20 a Mice were challenged i.p. with 2 10 5 CFU/0.2 ml of B. melitensis 16Mat 40 days postvaccination and were killed 15 days postchallenge. b s.c., subcutaneous.

4200 ADONE ET AL. INFECT. IMMUN. FIG. 1. Splenic growth curves for BALB/c mice inoculated with B. melitensis rough mutants, RB51, and B. melitensis 16 M. BALB/c mice were vaccinated i.p. with 10 8 CFU/mouse of the corresponding B. melitensis rough mutant and RB51 and with 10 7 CFU/mouse of the virulent B. melitensis 16 M strain. Levels of infection are expressed as the means standard deviations (n 5) of individual log CFU/spleen. receiving the experimental vaccines, a reference vaccine, and a placebo (unvaccinated controls) to restrict the spleen infection after a standardized virulent challenge. According to the experimental conditions suggested, 10 mice of groups I to IX were challenged i.p. with B. melitensis 16 M, at 2 10 5 CFU per subject, at 40 dpv. Two weeks later, mice were killed by cervical dislocation and spleens were removed for bacteriological examination and for IFN- production evaluation as described below. A mean value for each spleen count was obtained after logarithmic conversion (27). Vaccine efficacy was expressed as log 10 units of protection. Units of protection were obtained by subtracting the mean logarithmic count for each vaccinated group from the mean logarithmic count for the unvaccinated control group. Bacteriological examination. To detect Brucella organisms, spleens were aseptically removed from sacrificed mice, individually weighed, and diluted 1/10 (wt/wt) in sterile phosphate-buffered saline. Further dilutions were made, and 0.1 ml of each dilution was plated in triplicate onto BAS medium and incubated at 37 C for 5 days. The Brucella isolates were identified by Gram staining, colony morphology, and Brucella-specific PCR procedures (8). Genetic stability of Rif r B. melitensis mutants. The whole rpob genes of the Brucella rough mutants were amplified and sequenced as previously described (24). The presence of the specific mutations of each genotype was confirmed after in vivo passages by sequencing the rpob genes of colonies recovered from spleens of vaccinated mice. Production of IFN-. For IFN- production, evaluation was done with mice killed at 7, 15, and 40 dpv and 15 days after challenge exposure. Spleen suspensions of sacrificed mice, previously diluted 1/10 (wt/wt) in saline, were tested without stimulation by using a mouse IFN- -specific antigen capture enzymelinked immunosorbent assay (Quantikine M kit; R&D Systems). All assays were performed in duplicate, and the concentration was calculated by using a linear regression equation obtained from the absorbance values of the standards according to the manufacturer s procedures. Serological examination. To evaluate antibody responses, blood samples were collected from the retro-orbital sinuses of mice under anesthesia and sera were stored at 20 C until use. Serum samples were tested by the complement fixation test (CFT) and Rose Bengal plate test (RBPT) with S-type B. abortus strain 99. Tests were done as described by Alton et al. (4). To detect antibodies induced by RB51 and B. melitensis rough strains, a CFT was performed using B. abortus RB51 as the antigen, which is deprived of the anticomplementary activity due to the rough phenotype, as previously described (1, 2). Statistical methods. Differences among groups were estimated by a nonparametric Mann-Whitney test. A P value of 0.06 was considered significant. RESULTS Stability of rough mutants. No reversion to the virulent smooth phenotype was observed in B. melitensis rough strains after passages in vitro and in vivo. In addition, bacteria of mutant strains recovered from spleens of vaccinated mice were genetically investigated, and the presence of the specific mutations in the rpob gene sequence was confirmed. Bacterial persistence. Bacterial clearance of the RBM9, RBM11, RBM14, RBM15, RBM17, RBM19, and RB51 strains in comparison with that of the virulent B. melitensis 16 M strain is shown in Fig. 1. As indicated, all B. melitensis rough strains replicated extensively in spleen, showing bacterial persistence similar to that of RB51. A marked decline of spleen colonization was observed at 40 dpv in all vaccinated groups; at this point, about 10 2 CFU was detected in mice vaccinated with RBM9, RBM11, RBM15, RBM17, and RBM19, while 10 3 and 10 CFU was recovered from RBM14- and RB51-vaccinated mice, respectively. At the same time, 2 10 6 CFU of brucellae was still detected in spleens of mice inoculated with the virulent B. melitensis 16 M strain. Antibody responses. Antibody responses induced by vaccine strains were evaluated at 7, 15, 30, and 40 dpv and 15 days after the challenge with B. melitensis 16 M (Fig. 2). As expected, mice vaccinated with the rough Brucella strains, tested prior to challenge exposure, did not react in serological tests (serum agglutination test, RBPT, and CFT) in which S-type B. abortus S99 is used as the antigen (data not shown). The results of the

VOL. 73, 2005 PROTECTIVE B. MELITENSIS MUTANTS 4201 FIG. 2. Percentage of seropositive mice in each group of mice vaccinated with B. melitensis rough strains and RB51. Mice were tested by CFT using S99 (dark shading) and RB51(light shading) antigens at 7, 15, and 40 dpv and at 15 days after challenge with B. melitensis 16 M. The arrow indicates the time of challenge. CFT performed using S99 and RB51 antigens, expressed as percentage of seropositive mice, are shown in Fig. 2. As indicated, no antibody response to RB51 was detected at 7 dpv in mice vaccinated with B. melitensis rough strains, while 50% of RB51-vaccinated mice were positive. At 15 dpv, responses to RB51 ranged from 20% to 100% in mice vaccinated with RBM9, RBM11, RBM14, RBM19, and RB51, while no reaction was found by testing mice vaccinated with RBM15 and RBM17. The highest percentage of reactors to RB51 was observed at 40 dpv; at this time, from 80% to 100% of all vaccinated mice seroconverted to RB51. As expected, after the challenge with B. melitensis 16 M, antibodies to the virulent strain were produced in all groups; the percentage of vaccinated mice reacting to S99, as measured 15 days after the challenge, ranged from 30% (RB51) to 100% (RBM17). At the same time, 30% of unvaccinated mice (control) subjected to the same challenge seroconverted to S99 antigen when tested by CFT, while 70% of these mice were positive by the RBPT (data not shown). Production of IFN-. The results for IFN- production are shown in Fig. 3. When IFN- production in spleen tissues was tested at 7, 15, and 40 days after vaccination, we found that all strains were able to induce IFN-. However, the production of IFN- seemed to be higher with most of the examined strains than with RB51, and it seemed to be quite precocious, peaking at 7 dpv. The production of IFN- in mice killed 15 days after challenge with B. melitensis 16 M was higher in the unvaccinated group than in the vaccinated group. Protection against challenge. Protection was defined as the difference between the number of viable B. melitensis bacteria recovered from spleens of immunized mice and that recovered from spleens of mice receiving saline. Vaccine efficacy was expressed as log 10 units of protection. The results are summarized in Table 2. The protection in mice vaccinated with all B. melitensis rough strains versus control mice was statistically significant (P 0.06). The vaccination with Rev1 and RB51 reference vaccines conferred, respectively, 2.5 and 1.13 protection units. A very good level of protection was observed in mice vaccinated with RBM14 and RBM17 (2.60 and 2.83 protection units, respectively), similar to that conferred by Rev1 (P 0.06) and significantly higher than that in RB51-vaccinated mice (P 0.06). The RBM9 and RBM15 strains also conferred good protection (1.94 and 1.87 protection units, respectively), similar to that of Rev1 (P 0.06). Strains RBM11 and RBM19 showed protective efficacy similar to that of RB51 (P 0.06), at 1.02 and 1.10 protection units, respectively (Table 2). Brucella organisms of challenge strain 16 M were recovered from all unvaccinated mice (infection rate, 100%). DISCUSSION A variety of strategies are being used for developing new Brucella vaccines, especially in obtaining attenuated strains, cell extracts, or recombinant proteins. Limited success has been obtained with DNA vaccines encoding known protective antigens. Among these strategies, attempts are being made to develop defined rough mutant vaccines to overcome the main problem of production of antibodies which interfere with diagnostic tests (5, 29, 40, 44). Rough Brucella organisms lack the polysaccharide O chain of the lipopolysaccharide (LPS) molecule. The O chain plays a central role in the serological diagnosis of brucellosis; it is an immunodominant antigen able to induce an antibody response in animals exposed to smooth Brucella organisms, and it is detected by standard serological tests. Recent data indicate that O-chain properties are firmly linked to Brucella virulence

4202 ADONE ET AL. INFECT. IMMUN. FIG. 3. Production of IFN- (pg/ml) by splenocytes of vaccinated and control mice evaluated at different times after vaccination. The arrow indicates the time of challenge with B. melitensis 16 M (at 40 dpv). The values are expressed as means standard deviations (n 10). (21). Usually, the change from smooth to rough phenotype is associated with a marked decline in virulence. In this study, we compared the vaccine properties of six rifampin-resistant rough mutants of B. melitensis, named RBM9, RBM11, RBM14, RBM15, RBM17, and RBM19, with those of the Rev1 and RB51 reference vaccines. Previous studies indicated that the addition of rifampin to the medium tended to turn B. abortus cultures rough and that organisms resistant to rifampin were less virulent than rifampin-susceptible strains (28). RB51 has also been used for comparative evaluation because of its well documented ability in mouse models to confer protection not only against B. abortus infection but also against B. melitensis and B. ovis (19). In fact, other authors compared B. melitensis rough mutants with S19, Rev1, and RB51 vaccines (45). In addition, unlike Rev1, strain RB51 has the same rough phenotype and the same rifampin resistance as the experimental Brucella strains used in this study. As described in Table 1, the B. melitensis rough mutants used in this study showed the same rifampin-resistant phenotype but different rpob genotypes. They proved to be stable in characteristics after repeated passages in vitro and in vivo: no tendency to revert to the virulent smooth phenotype was found, and the presence of specific mutations in the rpob gene sequence of each genotype was confirmed after many passages, indicating genetic stability. To evaluate virulence, experiments were performed with BALB/c mice to monitor bacterial clearance of each mutant compared to RB51 and to the virulent B. melitensis 16 M strain. To achieve this, as previously suggested (14), rough strains were given intraperitoneally, at 10 8 CFU/mouse, while the B. melitensis 16 M smooth strain was administered subcutaneously at 10 5 CFU/mouse. The splenic growth curves indicated that B. melitensis rough mutants replicated extensively in spleen and were significantly less virulent than the virulent B. melitensis 16 M strain (P 0.06); in fact, at 40 dpv the number of brucellae recovered from spleens of vaccinated mice was markedly reduced (about 10 2 CFU), while at the same time, 2 10 6 CFU of brucellae was still detected in mice infected with B. melitensis 16 M (Fig. 1). RB51, as previously observed (19), showed a lower ability to replicate in BALB/c mice. Strains RBM9, RBM15, RBM17, and RBM19 showed bacterial clearances similar to that of RB51 (P 0.06), while strains RBM11 and RBM14 were significantly different from RB51 (P 0.06). The reduced virulence of rough mutants is likely due to the absence of smooth LPS, promoting ingestion and killing by macrophages and a reduced ability to stimulate suppressor T-cell activity (16, 17, 42). The evaluation of serological responses induced by rough strains confirmed their inability to produce antibodies to LPS O antigen. In fact, no reaction was obtained in the CFT, RBPT, or serum agglutination test with the S-type B. abortus 99 strain as the antigen. However, according to previous studies (15, 36), rough mutants produce antibodies which can be detected by using RB51 as a rough antigen in a CFT. The production of rough antibodies appeared 1 week postvaccination in mice vaccinated with RB51 and at 2 weeks postvaccination in the other groups. At 6 weeks postvaccination, the percentage of reactors to RB51 ranged from 80 to 100% (Fig. 2). As expected, after the challenge exposure to B. melitensis 16 M, all vaccinated mice produced antibodies to LPS of O antigen. However, the percentage of mice that were seropositive to B. abortus S99 was higher in rough mutant-vaccinated groups (from 50 to 100%) than in RB51-vaccinated or control groups (30% for both); the highest percentage of reactors was observed in the RBM17-vaccinated

VOL. 73, 2005 PROTECTIVE B. MELITENSIS MUTANTS 4203 group (100%). The 30% seroconversion to S99 could seem to be a weak response; however, 70% of control mice were positive when tested with the RBPT (data not shown). In addition, Brucella organisms of challenge strain 16 M were recovered from all unvaccinated mice (infection rate, 100%), thus confirming that the experiment was performed under controlled conditions. The difference between the RBPT and CFT results could be due to different antibody subisotypes detected, in addition to individual variability. The results of the IFN- production evaluation indicated that the spleen colonization of rough mutants was, to an extent, enough to prompt an adequate cell-mediated immune response, which is essential for controlling intracellular pathogens. At 7 dpv, the IFN- level, evaluated ex vivo without stimulation of lymphocytes, was significantly higher in RBM14- and RBM15-vaccinated mice than in RB51-vaccinated mice; at 15 dpv, the IFN- production decreased, yet the mean value was still higher in mice vaccinated with rough mutants than in the RB51-vaccinated mice (Fig. 3). After challenge with B. melitensis 16 M, we found higher production of IFN- in unvaccinated animals than in vaccinated ones. It is not surprising that gamma interferon is virtually absent in vaccinated mice following challenge with the virulent strain. It has been observed that high levels of antibodies and antigen-specific IFN- are strongly related to active Brucella infection in which the pathogen is massively excreted. Similarly, the state of protection against the disease, but not necessarily against infection, provided by the Rev1 vaccine appears to determine the low level of response, except with some techniques (13). In effect, in the course of infection with intracellular bacteria such as Brucella, it is reasonable to expect that the protection would be accompanied by an high level of IFN-. However, many studies showed a lack of correlation between this technique in vitro and the immunity observed in the animals. In conclusion, as previously suggested (32), the magnitude and duration of detectable immune responses should decrease in immune animals, which eliminate infection sooner, thus removing the antigenic stimulus for a high and protracted response. In the protection assay, the protection in mice vaccinated with all B. melitensis rough strains versus control mice was statistically significant. In particular, mice vaccinated with RBM14 and RBM17 exhibited levels of protection similar to those conferred by Rev1 and significantly higher than those in RB51-vaccinated mice. RBM17 conferred the highest protection. RBM9 and RBM15 also conferred good protection similar to that of Rev1, while strains RBM11 and RBM19 showed protective activity similar to that of RB51. It has been known that the high protection against B. melitensis provided by the Rev1 vaccine can also be attributed to the induction of O-polysaccharide-specific antibodies, which confer a high degrees of protection against S-type challenge Brucella strains. However, the results of this study indicated that strains RBM17 and RBM14, in particular, conferred a very good level of protection despite their inability to induce a protective response by producing antibodies to O antigen. As previously indicated (43), it is unlikely that the rough antibodies induced by these strains, probably directed against outer membrane proteins of Brucella, could protect against S-type Brucella infection. In fact, the protection conferred by RB51was lower than that conferred by the other strains, despite its marked ability to induce rough antibodies. Here we showed that protective properties of rough strains may be due to their ability to induce a strong cellular Th1 response, as shown by the production of a high concentration of IFN-, which is considered to play a crucial role in protection (47). The presence of different mutations in the rpob sequences of all B. melitensis rough mutants did not affect the degree of virulence, since bacterial clearance was similar. However, because of the different protection provided by the strains in this study, in some cases significantly higher than that of RB51, protective properties could be affected by the rpob genotype. Variations in the rpob sequence, in fact, could be responsible either directly or indirectly for the induction of more potent cell-mediated responses. Additional studies are needed to define the role of rpob mutations in protective activity of Brucella spp. In conclusion, the induction of a level of protection against B. melitensis similar to that of Rev1 and significantly superior to that of RB51, in addition to the inability to produce antibodies to O antigen, encourages further studies especially on strains RBM14 and RBM17 as vaccine strains against B. melitensis infection. The active replication of these strains and their rapid elimination from mice provided a measure of the safety for primary host species. Finally, the stable mutations in the rpob gene represent specific genetic markers that would make these strains easy to differentiate from field isolates. REFERENCES 1. Adone, R., and F. Ciuchini. 1999. Complement fixation test to assess humoral immunity in cattle and sheep vaccinated with Brucella abortus RB51. Clin. Diagn. Lab. Immunol. 6:787 790. 2. Adone, R., F. Ciuchini, F., and S. C. Olsen. 2001. Field validation of RB51 antigen in a complement fixation test to identify calves vaccinated with Brucella abortus RB51. Clin. Diagn. Lab. Immunol. 8:385 387. 3. Alton, G. G., and S. S. Elberg. 1967. Rev. 1 Brucella melitensis vaccine: a review of ten years of study. Vet. Bull. 371:793 800. 4. Alton, G. G., L. M. Jones, R. D. Angus, and J. M. Verger. 1988. Techniques for the brucellosis laboratory. Institut National de la Recherche Agronomique, Paris, France. 5. Banai, M. 2002. Control of small ruminant brucellosis by use of Brucella melitensis Rev. 1 vaccine: laboratory aspects and field observations. Vet. Microbiol. 90:497 519. 6. Blasco, J. M., and R. Diaz. 1993. Brucella melitensis Rev. 1 as cause of human brucellosis. Lancet 342:805. 7. Blasco, J. M. 1997. A review of the use of Brucella melitensis Rev. 1 vaccine in adult sheep and goats. Prev. Vet. Med. 31:275 283. 8. Bricker, B., and S. M. Halling. 1994. Differentiation of Brucella abortus bv. 1, 2, and 4; Brucella melitensis; Brucella ovis; and Brucella suis bv. 1 by PCR. J. Clin. Microbiol. 32:2660 2666. 9. Cheville, N. F., M. G. Stevens, A. E. Jensen, F. M. Tatum, and S. M. Halling. 1993. Immune responses and protection against infection and abortion in cattle experimentally vaccinated with mutant strains of Brucella abortus. Am. J. Vet. Res. 54:1591 1597. 10. Corner, L. A., and G. G. Alton. 1981. Persistence of Brucella abortus strain in adult cattle vaccinated with reduced doses. Res. Vet. Sci. 31:342 344. 11. El Idrissi, A. H., A. Benkirane, M. El Maadoudi, M. Bouslikhane, J. Berrada, and A. Zerouali. 2001. Comparison of the efficacy of Brucella abortus strain RB51 and Brucella melitensis Rev. 1 live vaccines against experimental infection with Brucella melitensis in pregnant ewes. Rev. Sci. Tech. Off. Int. Epiz. 20:741 747. 12. Elzer, P. H., F. M. Enright, J. R. Mc Quiston, S. M. Boyle, G. G., and Schurig. 1998. Evaluation of a rough mutant of Brucella melitensis in pregnant goats. Res. Vet. Sci. 64:259 260. 13. Durán-Ferrer, M., L. León, K. Nielsen, V. Caporale, J. Mendoza, A. Osuna, A. Perales, P. Smith, C. De-Frutos, B. Goḿez-Martín, A. Lucas, R. Chico, O. D. Delgado, J. C. Escabias, L. Arrogante, R. Díaz-Parra, and F. Garrido. 2004. Antibody response and antigen-specific gamma interferon profiles of vaccinated and unvaccinated pregnant sheep experimentally infected with Brucella melitensis. Vet. Microbiol. 100:219 231. 14. González, D., M. J. Grilló, P. M. Muñoz, M. J. de Miguel, D. Monreal, C. M. Marín, L. López-Goñi, I. Moriyón, and J. M. Blasco. 2003. Comparison of Brucella melitensis R mutants with intact or defective lipopolysaccharide

4204 ADONE ET AL. INFECT. IMMUN. core as vaccines in BALB/c mice, p. 140. Abstracts of the Brucellosis 2003 International Research Conference, Pamplona, Spain, 15 17 September 2003. 15. Hamdy, M. E. R., S. M. El-Gibaly, and A. M. Montasser. 2002. Comparison between immune responses and resistance induced in BALB/c mice vaccinated with RB51and Rev. 1 vaccines and challenged with Brucella melitensis bv. 3. Vet. Microbiol. 88:85 94. 16. Harmon, B. G., L. G. Adams, and M. Frey. 1988. Survival of rough and smooth strains of B. abortus in bovine mammary gland macrophages. Am. J. Res. 49:1092 1097. 17. Haslov, K., A. Fomsgaard, K. Takayama, J. S. Fomsgaard, P. Isben, M. B. Fauntleroy, P. W. Stashak, C. E. Taylor, and P. J. Baker. 1992. Immunosuppressive effects induced by the polysaccharide moiety of some bacterial lipo-polysaccharides. Immunobiology 186:378 393. 18. Jiménez de Bagüés, M. P., C. M. Marín, M. Barberán, and J. M. Blasco. 1989. Responses of ewes to B. melitensis Rev. 1 vaccine administered by subcutaneous or conjunctival routes at different stages of pregnancy. Ann. Rech. Vet. 20:205 213. 19. Jiménez de Bagüés, M. P., P. H. Elzer, S. M. Jones, J. M. Blasco, F. M. Enright, G. G. Schurig, and A. J. Winter. 1994. Vaccination with Brucella abortus rough mutant RB51 protects BALB/c mice against virulent strains of Brucella abortus, Brucella melitensis, and Brucella ovis. Infect. Immun. 62: 4990 4996. 20. Jiménez de Bagüés, M. P., M. Barberán, C. M. Marín, and J. M. Blasco. 1995. The Brucella abortus RB51 vaccine does not confer protection against Brucella ovis in rams. Vaccine 13:301 304. 21. Jiménez de Bagüés, M. P., A. Terraza, A. Gross, and J. Dornand. 2004. Different responses of macrophages to smooth and rough Brucella spp.: relationship to virulence. Infect. Immun. 72:2429 2433. 22. Jin, D. J., and C. A. Gross. 1988. Mapping and sequencing of mutations in the Escherichia coli rpob gene that lead to rifampicin resistance. J. Mol. Biol. 202:45 58. 23. Levin, M. E., and G. F. Hatfull. 1993. Mycobacterium smegmatis RNA polymerase: DNA supercoiling, action of rifampin and mechanism of rifampin resistance. Mol. Microbiol. 8:277 285. 24. Marianelli, C., F. Ciuchini, M. Tarantino, P. Pasquali, and R. Adone. 2004. Genetic bases of rifampin resistance phenotype in Brucella spp. J. Clin. Microbiol. 42:5439 5443. 25. McEwen, A. D. 1940. The virulence of Brucella abortus for laboratory animals and pregnant cattle. Vet. Rec. 52:97 106. 26. Meyer, M. E. 1885. Characterization of Brucella abortus strain 19 isolated from human and bovine tissues and fluids. Am. J. Vet. Res. 46:902 904. 27. Montaraz, J. A., and A. J. Winter. 1986. Comparison of living and nonliving vaccines for Brucella abortus in BALB/c mice. Infect. Immun. 53:245 251. 28. Moorman, D. R., and G. L. Mandell. 1981. Characteristics of rifampinresistant variants obtained from clinical isolates of Staphylococcus aureus. Antimicrob. Agents Chemother. 20:709 713. 29. Moriyon, I., M. J. Grillo, D. Monreal, D.Gonzalez, C. Marin, I. Lopez-Goni, R. C. Mainar-Jaime, E. Moreno, and J. M. Blasco. 2004. Rough vaccines in animal brucellosis: structural and genetic basis and present status. Vet. Res. 35:1 38. 30. National Academy of Sciences. 1977. Brucellosis research: an evaluation, p. Editor: J. D. Clements 61 77. Report of the Subcommittee on Brucellosis Research, National Academy of Sciences. National Academy Press, Washington, D.C. 31. Nicoletti, P. 1990. Vaccination, p. 284 299. In K. Nielsen and J. R. Duncan (ed.), Animal brucellosis. CRC Press, Boca Raton, Fla. 32. Nicoletti, P., and A. J. Winter. 1990. Response to Brucella abortus, p. 83 95. In K. Nielsen and J. R. Duncan (ed.), Animal brucellosis. CRC Press, Boca Raton, Fla. 33. Office International des Epizooties. 2004. Manual of diagnostic tests and vaccines for terrestrial animals, 5th ed. Office International des Epizooties, Paris, France. 34. Olsen, S. C., B. Bricker, M. V. Palmer, A. E. Jensen, and N. F. Cheville. 1999. Responses of cattle to two dosages of Brucella abortus strain RB51: serology, clearance and efficacy. Res. Vet. Sci. 66:101 105. 35. Orduña-Domingo, A., M. A. Bratos-Pérez, R. Abad-Fernández, L. Ruiz- Garcia, M. De-Frutos-Serna, A. Rodriguez-Torres. 2001. La brucelosis. Etiologia y origen de la infección humana, p. 13 20. In A. Rodriguez-Torres and A. Orduña-Domingo (ed.), Manual de brucelosis. Junta de Castilla y León, Zamora, Spain. 36. Pasquali, P., R. Adone, C. Pistoia, P. Petrucci, and F. Ciuchini, F. 2003. Brucella abortus RB51 induces protection in mice orally infected with the virulent strain B. abortus 2308. Infect. Immun. 71:2326 2330. 37. Roop, R. M., II, D. P. Preston-Moore, T. Bagchi, and G. G. Schurig. 1987. Rapid identification of smooth Brucella species with a monoclonal antibody. J. Clin. Microbiol. 25:2090 2093. 38. Schurig, G. G., C. Hammerberg, and B. R. Finkler. 1984. Monoclonal antibodies to Brucella surface antigens associated with the smooth lipopolysaccharide complex. Am. J. Vet. Res. 45:967 971. 39. Schurig., G. G., R. Martin Roop II, T. Bagchi, S. Boyle, D. Buhrman, and N. Sriranganathan. 1991. Biological properties of RB51; a stable rough strain of Brucella abortus. Vet. Microbiol. 28:171 188. 40. Schurig, G. G., N. Sriranganathan, and M. J. Corbel. 2002. Brucellosis vaccines: past, present and future. Vet. Microbiol. 90:479 496. 41. Smith, L. D., and T. A. Ficht. 1990. Pathogenesis of Brucella. Crit. Rev. Microbiol. 17:209 230. 42. Stevens, M. G., S. C. Olsen, G. W. Pugh, and M. V. Palmer. 1994. Immune and pathologic responses in mice infected with Brucella abortus 19, RB51, or 2308. Infect. Immun. 62:3206 3212. 43. Stevens, M. G., and S. C. Olsen. 1996. Antibody responses to Brucella abortus 2308 in cattle vaccinated with B. abortus RB51. Infect. Immun. 64:1030 1034. 44. Ugalde, J. E., D. J. Comerci, M. S. Leguizamon, and R. A. Ugalde. 2003. Evaluation of Brucella abortus phosphoglucomutase (pgm) mutant as a new live rough-phenotype vaccine. Infect. Immun. 71:6264 6269. 45. Winter, A. J., G. G. Schurig, S. Boyle, N. Sriranganathan, J. S. Bevins, F. M. Enright, P. H. Elzer, and J. D. Kopec. 1996. Protection of BALB/c mice against homologous and heterologous species of Brucella by rough strain vaccines derived from Brucella melitensis and Brucella suis biovar 4. Am. J. Vet. Res. 57:677 683. 46. Worthington, R. W., and F. D. Horwell. 1974. An investigation of the efficacy of three Brucella vaccines in cattle. J. S. Afr. Vet. Assoc. 45:87 91. 47. Yingst, S., and D. L. Hoover. 2003. T cell immunity to brucellosis. Crit. Rev. Microbiol. 29:313 331.