INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Brucella abortus is an intracellular pathogen that causes abortion in bovines and can infect humans. Abortion in cattle is the consequence of the tropism that the bacterium has for the placenta of pregnant animals, in which it multiplies intracellularly (10). Brucellosis in humans is primarily a disease of the reticuloendothelial system, in which the bacteria multiply inside the phagocytic cell; the intermittent release of bacteria from the cells into the bloodstream causes undulant fever (17, 29). Brucellosis does not spread among humans; consequently, eradication of the disease from the natural reservoirs, cattle, pigs, sheep, goats, and other susceptible animals, will lead to elimination of human infection. In regions with high prevalence of the disease, the only way of controlling and eventually eradicating this zoonosis is by vaccination of all susceptible hosts and elimination of infected animals.

Vaccination represents an important tool for the control of bovine brucellosis. One of the most used vaccines is the attenuated strain B. abortus S19 obtained spontaneously from the virulent strain B. abortus 2308 (24, 25, 26, 29). Live attenuated B. abortus S19 has served for many years as an effective vaccine to prevent brucellosis in cattle (8, 18). The genetic defect that leads to attenuation of this strain has not yet been defined. B. abortus S19 has lost some essential unknown mechanism of virulence. Despite this fact, the vaccinal strain conserves some degree of virulence, being pathogenic for humans (37), and produces abortion and persistent infection in adult vaccinated cattle. Vaccination with B. abortus S19 is used only for sexually immature animals (25, 26). Brucella, Agrobacterium, and Rhizobium belong, according to 16S rRNA sequences, to the -2 subgroup of the Proteobacteria (16), and comparative studies of the virulence genes of the plant pathogen Agrobacterium and the endosymbiotic Rhizobium might give us new insights on Brucella virulence factors. The Brucella two-component regulatory system (30) is highly similar to the two-component regulatory system ChvG-ChvI of Agrobacterium tumefaciens (5) and ExoS-ChvI of Rhizobium meliloti (6). These two-component regulatory genes are equivalent to Salmonella PhoP-PhoQ (31) and Bordetella bronchiseptica BvgA-BvgS systems (32). In all these bacteria, the two-component sensory systems are involved in controlling virulence or, in the case of Rhizobium, in nodule invasion. B. abortus bvrS bvrR mutants display reduced invasiveness and virulence (22, 30).

A Brucella virB operon highly homologous to the A. tumefaciens virB operon was identified in Brucella suis (20) and in B. abortus (28). A B. abortus virB10 mutant lost the ability to multiply in HeLa cells and was not recovered from the spleens of infected BALB/c mice (28). The same results were obtained with a B. suis virB10 mutant (20), thus demonstrating that in Brucella, as in Agrobacterium, the virB operon is involved in virulence.

In a recent report, a highly conserved B. abortus homologue of the R. meliloti bacA gene, which encodes a putative cytoplasmic membrane transport protein required for symbiosis, was identified (14). The B. abortus bacA mutant shows decreased survival in macrophages and reduced virulence in BALB/c mouse infection (14).

Brucella, like Agrobacterium and Rhizobium, produces cyclic -1,2-glucans (34). chvB in A. tumefaciens and ndvB in R. meliloti were identified as the genes coding for the cyclic -1,2-glucan synthetase (12). We recently reported that in Brucella the biosynthesis of cyclic -1,2-glucan proceeds by the same mechanism as in Rhizobium and Agrobacterium (4). The cyclic glucan synthetase (Cgs) acts as an intermediate during the synthesis of the cyclic -1,2-glucan (12). So far, cyclic -1,2-glucan has been described only for bacteria that interact with plants as either pathogens or endosymbionts. This glucan is required for effective nodule invasion in symbiotic nitrogen-fixing R. meliloti and for crown gall tumor induction in A. tumefaciens (3). Agrobacterium cyclic -1,2-glucan mutants have several altered cell surface properties including loss of motility due to a defective assembly of flagella and increased sensitivity to certain antibiotics and detergents (3).

The B. abortus S19 gene that codes for a cyclic -1,2-glucan synthetase has previously been identified and sequenced (12). Brucella cgs, Agrobacterium chvB, and Rhizobium ndvB are interchangeable genes. Agrobacterium or Rhizobium cyclic -1,2-glucan mutants can be complemented by the Brucella cgs genes, indicating that their functions are highly conserved (11, 12). A preliminary characterization of B. abortus S19 cgs mutants showed that they had reduced survival in BALB/c mouse spleen tissues, thus suggesting that this glucan might be a virulence factor (12). In this study, we examined the virulence of B. abortus cgs mutants in mice and their intracellular replication in HeLa cells. The protection induced in mice by a B. abortus S19 cgs mutant against a challenge with the virulent strain B. abortus 2308 was also evaluated.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. Bacterial strains and plasmids used in this study are listed in Table 1. Brucella strains were grown in Brucella agar (BA) (Difco Laboratories, Detroit, Mich.). Escherichia coli strains were grown in Luria broth. Fuchsin, sodium dodecyl sulfate (SDS), Triton X-100, Zwittergent 316, and deoxycholic acid (DOC) sensitivity tests were carried out as previously described (1, 30). The absence of smooth-to-rough dissociation was checked by testing the sensitivity to smooth-specific phages (Tb, Wb, and Iz) (1).

HeLa cell culture and infection assay. HeLa cells were grown at 37°C in 5% CO₂ atmosphere in minimal essential medium (Gibco, Paisley, Scotland) supplemented with 5 mM glutamine and 5% fetal calf serum. Infection of cells with different Brucella strains was performed as previously described (22, 28).

Construction of B. abortus -1,2-glucan synthetase mutant and genetic complementation. Construction of B. abortus strains was carried out by gene replacement of the wild-type cgs gene with a Tn3-HoHo 1 mutated gene (12) (strain BAI129). Confirmation of transposon position was carried out by PCR and Southern blot hybridization. For genetic complementation of cgs mutants, plasmid pCD523 containing the A. tumefaciens chvB gene (9) or plasmid pBA19 containing the B. abortus cgs gene (12), were introduced in strains BAI129 or BvI129 by biparental mating using E. coli S17.1 as the donor strain (7, 12). Complemented B. abortus cgs mutants are described in Table 1.

Pathogenicity in mice. Nine-week-old female BALB/c mice were injected intraperitoneally with 0.2 ml of a suspension containing the appropriate number of viable brucellae. Stock cultures were grown for 48 h on BA plates, and cells were suspended in sterile 0.15 M NaCl and adjusted turbidimetrically to the selected concentration. The exact bacterial concentration was calculated retrospectively by viable counts. At selected times postinfection, groups of five mice were bled by cardiac puncture and sera were pooled and held frozen at 20°C until use. Mice were killed by cervical dislocation. Spleens were homogenized in 5 ml of 0.15 M NaCl, serially diluted, and plated by triplicate on BA plates with the appropriate antibiotic (15).

Vaccination and evaluation of protection after challenging with virulent strain 2308. Nine-week-old BALB/c mice were vaccinated with 10⁴ CFU of B. abortus S19 wild type or B. abortus S19 BAI129 cgs mutant (12). At 8 weeks postvaccination, no bacteria were isolated from the spleens of animals infected with S19 or BAI129 and animals were challenged intraperitoneally with different doses of B. abortus virulent strain 2308. One week after challenge, five animals per treatment were killed the numbers of viable Brucella recovered from the spleens and the weights of the spleens were determined as described (15). In order to distinguish strain S19 from strain 2308, counts were carried out on BA plates with 0.01% erythritol as previously described (15).

Semiquantitation of IFN- and IL-4 mRNA in spleens of vaccinated mice. BALB/c mice injected with 10⁴ CFU of B. abortus S19 or the B. abortus S19 cgs mutant were sacrificed after 4 or 8 days, and total RNA was extracted from spleens with TRIzol reagent (GIBCO BRI, Gaithersburg, Md.) as previously described (21). Concentration and purity of extracted RNA were determined by determining the A₂₆₀ and the A₂₆₀/A₂₈₀ ratio. Five micrograms of RNA were reverse transcribed using a commercial kit (Superscript II RT, GIBCO BRL) with a 12- to 18-mer deoxy-T oligonucleotide primer. Quantitation of reverse-transcribed mRNA by PCR was carried out as described previously (21). Plasmid pMus (kindly provided by D. Shire and F. Pitossi) harboring the same primer sequences for the amplification of several murine cytokines and some housekeeping mRNAs was used as a competitive fragment. The primer sequences used were as follows: for 2-microglobulin sense, TGACCGGCTTGTATGCTATC; for 2-microglobulin antisense, CAGTGTGAGCCAGGATATAG; for gamma interferon (IFN-) sense, GCTCTGAGACAATGAACGCT; for IFN- antisense, AAAGAGATAATCTGGCTCTGC; for interleukin-4 (IL-4) sense, TCGGCATTTTGAACGAGGTC; and for IL-4 antisense, GAAAAGCCCGAAAGAGTCTC. The expected sizes of the PCR products were 222 bp for 2-microglobulin, 227 bp for IFN-, and 216 bp for IL-4. The concentration of 2-microglobulin in each sample was determined to check reverse-transcription (RT) efficiency and the accuracy of the determination of RNA concentration. For semiquantitation of IL-4 and IFN- mRNA, 5 × 10⁴ molecules of pMus were coamplified in each PCR in the presence of 3 µCi of [-³²P]dCTP. The separation of amplicons was accomplished on agarose gel (1.2%) in Tris borate buffer in the presence of ethidium bromide. The bands of the gel were inspected at 365 nm and excised, and radioactivity was counted in a liquid scintillator. For each amplification, the ratio of counts per minute of cellular amplicon to counts per minute of standard amplicon was calculated. Each individual sample was amplified by PCR at least twice to exclude casual errors. Spleen RNA samples were obtained from three mice subjected to the same treatment and analyzed separately.

Determination of antibody titers by ELISA and KELA. Antibody titers against B. abortus lipopolysaccharide (LPS) were measured in an indirect, computer-assisted kinetics-based enzyme-linked assay (KELA), as described by Jimenez de Bagues et al. and Winter et al. (13, 36). The indirect enzyme-linked immunosorbent assay (ELISA) to measure antibodies against Brucella LPS in the sera of mice was performed as described by Nielsen et al. (19) with some modification (7).

TLC of cyclic -1,2-glucan and PAGE of membrane proteins. Cyclic (1-2) glucans were extracted by the ethanol method (70% ethanol for 1 h at 37°C) from cell pellets of the different strains. Ethanolic extracts were concentrated and submitted to thin-layer chromatography (TLC) as described previously (4). SDS-polyacrylamide gel electrophoresis (PAGE) of membrane proteins and fluorography were performed as previously described (4, 12).

Western blotting. SDS-PAGE and Western blot analysis of LPS O antigen were performed as described by Comerci et al. (7). Whole-cell lysates were solubilized in Laemmli buffer at 100°C, electrophoresed by SDS-12% PAGE and transferred to nitrocellulose filters. The filters were reacted with M84 anti-O chain monoclonal antibody (19) (kindly provided by K. Nielsen) diluted 1:5,000 and incubated with peroxidase-conjugated goat anti-mouse immunoglobulin (Sigma Chemicals Co., St. Louis, Mo.) diluted 1:10,000. Peroxidase activity was detected by the ECL Western blotting kit from Amersham Pharmacia Biotech.

Statistical analysis. Differences between the means of experimental and control groups were analyzed using the Student t test. Differences were considered significant at P values of <0.05.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Characterization of B. abortus cgs mutants. The B. abortus 2308 cgs mutant (strain BvI129) and the B. abortus S19 cgs mutant (strain BAI129) were obtained by targeted insertional mutagenesis as described in Materials and Methods. Both mutants were complemented with plasmid pBA19 containing the B. abortus cgs gene or with plasmid pCD523 containing the A. tumefaciens chvB gene. Some of the phenotypic characteristics of the mutants and complemented strains are shown in Table 2. B. abortus cgs mutants do not form cyclic -1,2-glucan, and the synthesis was restored by plasmid pBA19 (Fig. 1 and Table 2). Complementation of -1,2-glucan synthesis of cgs mutants was also achieved with plasmid pCD523, which contains the Agrobacterium cyclic -1,2-glucan synthetase gene (11), thus indicating that Agrobacterium Cgs is also active in the Brucella background. Brucella cgs mutants did not grow in media containing DOC, SDS, Zwittergent, or fuchsin (Table 2), suggesting that the lack of the Cgs membrane protein or the inability to produce cyclic -1,2-glucan determines a major membrane defect that affects susceptibility to these compounds. All these inhibitions were relieved when the mutants were complemented with Agrobacterium chvB or Brucella cgs genes.

View larger version (95K): [in this window] [in a new window]   FIG. 1.   TLC of cyclic -1,2-glucans accumulated by different B. abortus strains. Cyclic -1,2-glucans were extracted and submitted to TLC as described in Materials and Methods. Lanes: 1, B. abortus 2308; 2, B. abortus 2308 cgs mutant (BvI129 strain); 3, B. abortus strain BvI129 complemented with plasmid pBA19; 4, B. abortus S19; 5, B. abortus S19 cgs mutant (BAI129 strain); 6, B. abortus strain BAI129 complemented with plasmid pBA19. * and **, migration of charged and neutral B. abortus cyclic -1,2-glucans, respectively.

View larger version (28K): [in this window] [in a new window]   FIG. 2.   Western blot of different B. abortus strains. Whole-cell lysates of different B. abortus strains were electrophoresed and transferred, and LPS O antigen was detected as described in Materials and Methods. Lanes: 1, B. abortus 2308 (wild-type strain); 2, B. abortus 2308 BvI129 (cgs mutant); 3, B. abortus S19 (vaccinal strain); 4, B. abortus S19 BAI129 (cgs mutant); 5, B. abortus RB51.

B. abortus cgs mutants have reduced virulence in mice. The recovery of B. abortus 2308 and B. abortus S19 cgs mutants (strains BvI129 and BAI129) from the spleens of mice was studied as described in Materials and Methods. Twelve weeks postinfection, the persistence of the B. abortus 2308 cgs mutant (strain BvI129) in the spleens of mice inoculated with 10⁴ CFU per mouse was reduced by 1.5 logs (P < 0.01) compared to the parental wild-type strain (Fig. 3A), while the B. abortus 2308-infected mice remained relatively at the same value at 4 and 12 weeks.

View larger version (26K): [in this window] [in a new window]   FIG. 3.   Brucella persistence in spleens of mice inoculated with different strains of B. abortus. Mice were inoculated intraperitoneally as described in Materials and Methods with 104 CFU of B. abortus. (A) Shaded bars, B. abortus 2308 (wild-type strain); open bars, B. abortus 2308 BvI129 (cgs mutant). (B) Shaded bars, B. abortus S19 (vaccinal strain); open bars, B. abortus S19 BAI129 cgs mutant. At different times postinfection (2, 4, or 12 weeks) five mice per group were killed and their spleens were removed. The numbers of CFU in spleen tissues were determined as indicated in Materials and Methods.

Induction of splenomegaly. As shown in Table 3, B. abortus S19 induces a dose-dependent transient splenomegaly as a consequence of inflammatory response. At all the tested doses, B. abortus S19 cgs mutant induced a reduced response compared to that of the wild-type strain. Complementation of cgs mutant with plasmids pBA19 or pCD523 restored the induction of splenomegaly (Table 3). The pathogenic B. abortus strain 2308 at infection doses of 10⁴ CFU per animal induced a threefold increase in spleen weight 2 weeks postinfection (225 ± 25 mg [mean ± standard deviation]) which persisted after 12 weeks (249 ± 60 mg). Although the numbers of bacteria recovered from spleens of mice infected with wild-type strains and from those infected with cgs mutant strains were not significantly different, no induction of splenomegaly was observed with the B. abortus 2308 cgs mutant at 2 weeks postinfection (77 ± 30 mg). At 12 weeks postinfection the splenomegaly increased twofold (170 ± 27).

HeLa cells infection. B. abortus is able to infect and multiply inside HeLa cells. The infection process has two phases, invasion and replication. The invasion phase (0 to 8 h postinfection) is a period during which Brucella enters into the cells but does not multiply (28). During the replication phase, Brucella organisms reach the rough endoplasmic reticulum and start to replicate, causing no cytopathological cell signs (22).

View larger version (15K): [in this window] [in a new window]   FIG. 4.   Intracellular replication of different strains of B. abortus. HeLa cells were inoculated with 5 × 107 CFU of different strains of B. abortus. After 2 h of incubation at 37°C, cells were washed and streptomycin and gentamicin were added as described in Materials and Methods. Numbers of CFU were determined at the indicated times. Each determination is the average of two independent experiments carried out in duplicate. In all cases, the standard error for each point was less than 5%. Symbols: , B. abortus 2308; ----, B. abortus BvI129 (cgs mutant strain); , B. abortus S19; ····; B. abortus BAI129 (cgs mutant strain); , B. abortus BvI129 (pBA19); ----, B. abortus BAI129(pBA19).

B. abortus BAI129 as vaccine. One of the drawbacks of the vaccine strain B. abortus S19 is that in cattle, mice, and humans it displays some degree of virulence, causing abortion in pregnant cows (10). In order to determine if the less pathogenic B. abortus BAI129 (cgs mutant) remains immunogenic, experiments of protection in mice were carried out. The B. abortus S19 cgs mutant strain BAI129 protected mice against a challenge with the virulent strain B. abortus 2308 to the same extent as the wild-type B. abortus S19 strain (Table 4). This indicates that this mutant, although having reduced virulence and impaired ability to multiply intracellularly in HeLa cells, completely retained the ability to protect mice. As shown in Table 4, protection experiments with B. abortus S19 or B. abortus BAI129 were carried out and the animals were challenged with different doses of the wild-type pathogenic strain. No significant differences were observed between the B. abortus S19 wild type and the cgs mutant at any dose.

Quantitation of IFN- and IL-4 mRNA in spleens of infected mice. BALB/c mice were infected with 10⁴ CFU of B. abortus S19 or the B. abortus S19 cgs mutant and at different postinfection times spleens were removed and total RNA was extracted. Quantitation of IFN- and IL-4 mRNA was carried out by RT-PCR as described in Materials and Methods. At 4 days postinfection B. abortus S19 or the B. abortus S19 cgs mutant induced in the spleen an accumulation of IFN- mRNA that persisted after 8 days (Fig. 5). No induction of IL-4 mRNA was detected at any time with the mutant or the parental S19 strain (data not shown).

View larger version (31K): [in this window] [in a new window]   FIG. 5.   Quantitation of IFN- in spleen of infected mice. BALB/c mice were infected with 104 CFU of B. abortus S19 (shaded bars) or B. abortus cgs mutant (open bars). At different times postinfection, spleens were removed and total RNA was extracted and analyzed separately. (A) Quantitation of IFN- mRNA was carried out by RT-PCR as described in Materials and Methods. Error bars indicate standard deviations. (B) Agarose gel of the amplicon products from three mice at each time point. After the gel dried, radioactivity was detected by autoradiograph.

Patterns of immunoglobulins elicited in mice by B. abortus S19 and B. abortus BAI129 cgs mutant. The Th1 cytokine IFN- promotes switching to IgG2a, whereas IL-4, a product of Th2 cells, promotes switching to IgG1 and IgE (25, 37). An ELISA for IgG1 and IgG2a was carried out to estimate the Th1/Th2 ratio of host immune response to B. abortus S19 and the B. abortus S19 cgs mutant. Figure 6 shows the anti-LPS antibody titers of serum pools obtained from animals infected with 7 × 10⁸ CFU of B. abortus S19 and the B. abortus S19 cgs mutant BAI129 (Fig. 6A and B, respectively) at different postinfection times. It is observed that BAI129 induced a switching to IgG2a with low levels of IgG1, thus confirming a good IFN- response and the absence of IL-4 induction (see "Quantitation of IFN- and IL-4 mRNA in spleens of infected mice" above) (23, 35).

View larger version (23K): [in this window] [in a new window]   FIG. 6.   Pattern of immunoglobulins elicited by B. abortus S19 or the B. abortus BAI129 cgs mutant. KELA results for pooled sera from five mice inoculated with 7 × 108 CFU of B. abortus S19 (A) or B. abortus BAI129 (B) obtained at 1, 2, or 4 weeks postinfection are shown. IgG1 plus IgG2a give total KELA units.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we showed that the absence of cyclic -1,2-glucan in B. abortus is associated with a reduction of virulence in mice and defective intracellular multiplication in HeLa cells.

B. abortus cgs mutants display increased sensitivity to surfactant compounds. This association between reduction of virulence and sensitivity to surfactants was also observed in B. abortus mutants in the two-component regulatory system (30). With cgs mutants no correlation was observed between sensitivity to surfactants and virulence. B. abortus 2308 and S19 cgs mutants have the same sensitivity to DOC, SDS, and Zwittergent. However, in strain 2308 the effect on attenuation of virulence was lower than that observed with the attenuated vaccinal S19 strain, even though this may be the result of multiple defects in S19.

Some reports suggest that the presence of cyclic -1,2-glucan in the bacterial periplasm may stabilize membrane proteins against improper assembly or disassembly (33). For example, Swart et al. (33) showed that chvB mutants of A. tumefaciens produced an inactive form of the protein rhicadhesin which resulted in the attachment-minus phenotype of chvB mutant and that the normal phenotype was restored by the addition of the purified rhicadhesin from a wild-type strain. Banta et al. (2) showed that the chvB mutant of A. tumefaciens exhibits lower levels of the VirB10 protein than does the wild type. VirB10 is a transmembrane protein that is part of the type IV secretion system required for T-DNA delivery into plant cells; it has been recently demonstrated that this type IV secretion system is conserved in B. abortus and that virB10 null mutants are avirulent (28). Rhizobium and Agrobacterium cgs mutants are nonmotile due to a defective assembly of the flagella, a process that is known to take place in the periplasmic space (34). All these observations suggest that the absence of cyclic glucan and/or Cgs inner membrane may be important for virulence. On the other hand, it cannot be excluded that the Cgs 316-kDa inner membrane protein may have a direct effect on virulence by itself.

Our results demonstrate that the residual virulence of B. abortus S19 depends on the presence of cyclic -1,2-glucan, since BAI129 (cgs S19 mutant) was strongly attenuated in mice and displayed no intracellular multiplication in HeLa cells.

Montaraz and Winter (15) have shown that the number of CFU recovered from spleens of mice infected with B. abortus S19 peaked at 2 weeks postinfection with doses of 3.8 × 10⁴ or 3.8 × 10⁵ CFU. The behavior of the B. abortus S19 cgs mutant (strain BAI129) showed a different pattern of growth with a sharp declination of spleen counts after 2 weeks at all tested doses, thus indicating the inability of the mutant to replicate in the mice.

Despite the fact that the B. abortus S19 cgs mutant was cleared from the infected mice faster than the parental strain, due to reduction of virulence, our results showed that the mutant protected the mice against a challenge with the pathogenic B. abortus strain 2308 to the same extent as the parental vaccinal S19 strain.

In a recent study, Power et al. (23) proposed that the dose of the Mycobacterium bovis BCG vaccine is crucial in determining the Th1/Th2 nature of the immune response. They demonstrated that relatively low doses lead to an almost exclusive cell-mediated Th1 response, while higher doses induce a mixed Th1/Th2 response. We show that mice inoculated with the B. abortus S19 cgs mutant present equivalent levels of IFN- mRNA as well as lower titers of anti-LPS antibodies of IgG1 subtype than those observed in mice vaccinated with the parental S19 strain, suggesting an almost exclusively Th1 response. These results may be explained by the limited replication of the mutant in the spleen.

The decreased virulence with retention of the capacity to confer immunity suggests that exploring the cgs S19 mutant as a potential improvement of the B. abortus S19 vaccine might be worthwhile.