Identification and Characterization of a Brucella abortus ATP-Binding Cassette Transporter Homolog to Rhizobium meliloti ExsA and Its Role in Virulence and Protection in Mice

Abstract

Brucella abortus is a facultative intracellular bacterial pathogen that causes abortion in domestic animals and undulant fever in humans. The mechanism of virulence of Brucella spp. is not fully understood yet. Furthermore, genes that allow Brucella to reach the intracellular niche and to interact with host cells need to be identified. Using the genomic survey sequence (GSS) approach, we identified the gene encoding an ATP-binding cassette (ABC) transporter of B. abortus strain S2308. The deduced amino acid sequence encoded by this gene exhibited 69 and 67% identity with the sequences of the ABC transporters encoded by the exsA genes of Rhizobium meliloti and Mesorhizobium loti, respectively. Additionally, B. abortus ExsA, like R. meliloti and M. loti ExsA, possesses ATP-binding motifs and the ABC signature domain features of a typical ABC transporter. Furthermore, ortholog group analysis placed B. abortus ExsA in ortholog group 6 of ABC transporters more likely to be involved in bacterial pathogenesis. In R. meliloti, ExsA is an exopolysaccharide transporter essential for alfalfa root nodule invasion and establishment of infection. To test the role of ExsA in Brucella pathogenesis, an exsA deletion mutant was constructed. Replacement of the wild-type exsA by recombination was demonstrated by Southern blot analysis of Brucella genomic DNA. Decreased survival in mice of the Brucella ΔexsA mutant compared to the survival of parental strain S2308 demonstrated that ExsA is critical for full bacterial virulence. Additionally, the B. abortus exsA deletion mutant was used as a live vaccine. Challenge experiments revealed that the exsA mutant strain induced superior protective immunity in BALB/c mice compared to the protective immunity induced by strain S19 or RB51.

Brucellosis is a major zoonotic disease that causes abortion in domestic animals and undulant fever in humans. Brucella proliferates within macrophages of the host and thereby successfully bypasses the bactericidal effects of phagocytes (26, 34). Thus, Brucella virulence is associated with the capacity of the organisms to multiply inside the host cells. Therefore, cell-mediated immunity and subsequent activation of macrophages are essential for host clearance of infection (16). Once inside cells, Brucella prevents fusion of the phagosome with the lysosome by altering the intracellular traffic of the early phagosome vesicle (24). It has recently been demonstrated that brucellae replicate in a vesicle compartment containing reticuloendoplasmic markers reached after fusion between phagosomes and lysosomes is prevented (25). However, the genes that allow Brucella to invade and reach the appropriate intracellular replication niche remain to be identified.

To be successful in infection, a pathogenic intracellular bacterium requires four steps: adherence, invasion, establishment, and dissemination within the host (15). The interaction between host and pathogen also includes uptake and secretion of substances, which are facilitated by a family of proteins termed transporters. ATP-binding cassette (ABC) transporters are some of the active transport systems that are common in bacteria and eukaryotic cells (14). ABC transporters use the free energy of ATP hydrolysis to pump substances across the membrane against a concentration gradient into or out of cells (27). These transporters can use a variety of substrates, such as amino acids, sugars, inorganic ions, polysaccharides, peptides, and proteins like toxins. In Rhizobium meliloti, a member of the alpha subgroup of the Proteobacteria closely related to Brucella, the exsA gene was identified as a gene that encodes an ABC transporter of the exopolysaccharide succinoglycan (EPS I) (4). EPS I is essential for the invasion of alfalfa root nodules by R. meliloti. Mutants of R. meliloti devoid of EPS I are unable to establish an effective symbiosis with alfalfa (22). Since EPS I from R. meliloti is a surface polysaccharide, like Brucella lipopolysaccharide (LPS), which is considered an important virulence factor involved in many host-pathogen interactions (11) and therefore pathogenesis, we decided to isolate and further characterize the B. abortus exsA gene and the product that it encodes.

In this study, we identified the gene encoding B. abortus ExsA and performed a nucleotide and deduced amino acid sequence analysis. The amino acid sequence analysis revealed a high degree of identity among the B. abortus ABC transporter and R. meliloti and Mesorhizobium loti ExsA proteins. The sequence similarity and the presence of ABC transporter motifs in the Brucella ExsA sequence suggest a role for this molecule in the polysaccharide transport function critical in bacterial pathogenesis (11, 33). Furthermore, a mutant with a defined mutation in the exsA gene of B. abortus was obtained by gene replacement. The survival of the mutant was analyzed in the mouse model, and it was determined that ExsA is required for full virulence during Brucella infection. Additionally, the B. abortus ΔexsA mutant induced greater protective immunity in BALB/c mice than the commercially available strain S19 vaccine induced.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. B. abortus virulent strain S2308 and B. abortus vaccine strains RB51 and S19 were obtained from G. Splitter (University of Wisconsin-Madison, Madison, Wis.). They were grown in brucella broth medium (Becton Dickinson, Sparks, Md.) for 3 days at 37°C. If necessary, the medium was supplemented with ampicillin or kanamycin at a concentration of 25 μg/ml and with 0.1% erythritol. Escherichia coli DH5α was cultured at 37°C in Luria-Bertani medium containing kanamycin (50 μg/ml) or ampicillin (100 μg/ml) as needed (28).

TABLE 1.

Bacterial strains and vectors used in this study

Strain or plasmid	Characteristics	Source
E. coli	rfbD endAl hsdRl7 supE44 thi-l F′	Gibco BRL
Brucella strains
B. abortus S2308	Wild-type, smooth, virulent	Laboratory stock
B. abortus S19	Vaccine strain, smooth	Laboratory stock
B. abortus RB51	Rifr, rough mutant of S2308	Laboratory stock
B. abortus ΔexsA	Kanr, ΔexsA mutant of S2308	This study
Plasmids
pUC4K	ColE1, Ampr Kanr	Amersham Pharmacia
pUC12	ColE1, Ampr	Gibco BRL
pUC64	pUC12 plus 2.1 kb containing B. abortus exsA gene	This study
pGR64	pUC64 containing kanamycin resistance gene	This study

Isolation of B. abortus exsA and DNA and amino acid sequence analysis.

The exsA gene was isolated in a gene identification program by using the genomic survey sequence strategy. This approach was based on end sequencing of random clones of a B. abortus S2308 genomic library constructed in our laboratory by using the plasmid vector pUC12. Five hundred clones were sequenced, and a clone designated clone 64 (pUC64) with high homology to the R. meliloti exsA gene was identified. This clone, which was approximately 2.1 kb long, was sequenced by primer walking. Double-stranded DNA sequencing was performed by the dideoxy chain termination method (29) by using an automated laser fluorescence DNA sequencer from Amersham Pharmacia Biotech (Piscataway, N.J.). The clone was sequenced with a Thermo Sequenase kit, and the primers used were M13 reverse sequence, M13 universal sequence from Amersham Pharmacia Biotech, and specific primers purchased commercially. The sequences of the specific primers used were as follows: 1F, 5′-CTG ATG GGC CTG ATG GTG-3′; 2F, 5′-CGC AAC CTG CAA CGC TCG-3′; 3F, 5′-TCA CAG GCC GAG ATC GAG-3′; 1R, 5′-GTT TTC CGC CAG CGT CCG-3′; 2R, 5′-TGT CCC CTG GAT GAA ACG-3′; and 3R, 5′-GTC GGT CGA AAA GCG TTG-3′. The sequence data were compiled and analyzed by using the sequence analysis program DNASIS V5.00 (Hitachi Software). Subsequent homology searches were performed by using BLAST programs (2). Amino acid sequences were aligned by using the ClustalW service of the European Bioinformatics Institute available on the internet (http://www2.ebi.ac.uk/clustalw). The potential transmembrane domains of the ExsA protein were predicted by the Kyte-Doolittle method by using a computer program available at the Tokyo University of Agriculture & Technology website (http://sosui.proteome.bio.tuat.ac.jp/sosuiframe0E.html).

Determination of the exsA gene copy number in the B. abortus genome.

Genomic DNA from B. abortus strain S2308 was isolated as described previously (17). For DNA hybridization, 10 μg of genomic DNA was digested with BamHI, EcoRI, and EcoRV. BamHI and EcoRI did not cut within the open reading frame (ORF), whereas EcoRV did cut in the middle of the exsA gene. The fragments obtained were separated by 1% agarose gel electrophoresis, passively transferred onto a nylon membrane (Hybond-N; Amersham Pharmacia Biotech), and UV cross-linked. The exsA fragment was used as a DNA probe. This fragment was labeled with [α-³²P]dCTP by random priming (Gibco BRL). Hybridization was performed under stringent conditions (38°C, 50% formamide) as described previously (28). The nylon membrane was subjected to autoradiography for 18 h at −80°C.

Construction of B. abortus ΔexsA deletion mutant.

The deletion plasmid used for exsA gene replacement in B. abortus was constructed as follows. Plasmid pUC4K (Amersham Pharmacia Biotech) was digested with EcoRI, which produced a 1.2-kb fragment containing the kanamycin resistance cassette. This DNA fragment was blunt ended with T4 DNA polymerase (Amersham Pharmacia Biotech) and ligated into the unique EcoRV site of the exsA gene present in pUC64 to generate plasmid pGR64 (Fig. 1). To prepare B. abortus S2308 competent cells, bacteria were grown in 200 ml of brucella broth for 6 h at 37°C to the log phase (optical density at 600 nm, 0.4 to 0.6). The bacterial cells were harvested by centrifugation at 5,000 × g for 10 min, and they were washed three times with cold double-distilled water plus 10% glycerol and resuspended in 0.65 ml of chilled 10% glycerol. After that, 0.05-ml aliquots were stored immediately at −80°C. Five micrograms of pGR64 plasmid DNA was added to 50 μl of B. abortus competent cells in sterile electroporation cuvettes with 0.2-cm electrode gaps (Bio-Rad Laboratories, Richmond, Calif.), and then electroporation was performed with a Gene Pulser II transfection apparatus (Bio-Rad Laboratories) at 25 μF, 2.5 kV, and 400 Ω. One milliliter of SOC (2% tryptone, 0.5% yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl₂, 20 mM glucose; Gibco BRL Life Technologies, Inc., Gaithersburg, Md.) was added, and the cells were grown with agitation at 37°C for 16 h and plated on brucella agar containing 25 μg of kanamycin per ml. After 3 to 5 days of incubation at 37°C, transformants were examined by replica plating on brucella agar containing 25 μg of kanamycin per ml and 25 μg of ampicillin per ml. Then colonies that were Kan^r Amp^s and Kan^r Amp^r were selected as colonies in which double and single crossover events had occurred, respectively.

FIG. 1.

Schematic map of clone 64 encoding the exsA gene of B. abortus and construction of the pGR64 target vector for generation of the ΔexsA mutant by gene replacement. The arrow indicates the translation direction, and the cross-hatched bar indicates the exsA gene. Kan^r, kanamycin resistance gene.

Southern blot analysis of the B. abortus ΔexsA deletion mutant.

To provide genetic evidence that the wild-type exsA gene was replaced by an exsA gene interrupted by the Kan^r cassette, 10 μg of genomic DNA isolated from both the mutant strain and the wild-type strain (S2308) was digested with EcoRI and then loaded onto a 0.8% agarose gel to perform Southern blotting. The genomic DNA was transferred to a nylon membrane by capillary transfer overnight in 10× SSC buffer (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). DNA bound to the membrane was exposed to UV cross-linking and then prehybridized at 42°C for 4 h in 25 ml of a prehybridization solution (6× SSC, 0.2% sodium dodecyl sulfate, 10× Denhardt's solution, 100 μg of salmon sperm per ml, 50% formamide). The exsA, Kan^r, and Amp^r genes were used as probes and were labeled with [α-³²P]dCTP by using a random-primed labeling kit (Gibco BRL), and unincorporated ³²P was removed by Sephadex G-50 column chromatography. Hybridization was conducted at 42°C in the same prehybridization solution for 16 h. The blots were washed three times with 2× SSC-0.1% sodium dodecyl sulfate at 60°C for 30 min. After the last wash, the membrane was subjected to autoradiography for 6 h at −80°C.

Survival in the mouse model.

Virulence was determined by measuring the numbers of CFU of wild-type and mutant strains in mouse spleens. Female BALB/c mice that were 6 to 8 weeks old were injected intraperitoneally (i.p.) with 1 × 10⁵ CFU of brucellae in 0.1 ml of phosphate-buffered saline (PBS) (145 mM NaCl, 49 mM KH₂PO₄, 21 mM Na₂HPO₄; pH 7.2). Groups of eight mice were injected with either B. abortus S2308 or the B. abortus ΔexsA mutant strain. At 1, 2, and 6 weeks postinoculation, all mice in each group were killed, and bacterial survival was determined following homogenization of the mouse spleens in 10 ml of PBS. Tenfold serial dilutions of the homogenized spleens were plated on brucella agar containing kanamycin to determine the number of ΔexsA CFU per spleen compared to the number of wild-type CFU.

Immunization of mice with the B. abortus ΔexsA mutant.

Female BALB/c mice that were 6 to 8 weeks old were immunized i.p. with 1 × 10⁵ CFU of brucellae in 0.2 ml of PBS. Groups containing eight mice each were immunized with either B. abortus S19, B. abortus RB51, or B. abortus ΔexsA. Nonimmunized, control mice were injected i.p. with 0.2 ml of PBS. Six weeks after immunization, all mice in each group were challenged by i.p. injection of 1 × 10⁵ CFU of B. abortus S2308. Experimentally infected mice were killed 2 weeks later by cervical dislocation, and the spleens were collected and disrupted in 10 ml of PBS. Tenfold serial dilutions were plated on brucella agar containing kanamycin or 0.1% erythritol for differentiation of B. abortus ΔexsA, B. abortus strain S19, and B. abortus strain S2308. After 3 days of incubation at 37°C, colonies were visualized, and the number of CFU of B. abortus S2308 per spleen was determined after the number of B. abortus S19 or B. abortus ΔexsA found by replica plating was subtracted. The degrees of protection in immunized animals and controls were expressed as the mean CFU of B. abortus S2308 for each mouse group obtained after challenge and log₁₀ conversion. Log₁₀ units of protection were obtained by subtracting the mean log₁₀ CFU for the experimental group from the mean log₁₀ CFU for the control group, as previously described (19).

Statistical analyses.

Statistical analyses were performed with Student's t test by using the MINITAB computer software package (Minitab Inc., State College, Pa.).

Nucleotide sequence accession number.

The nucleotide sequence of the exsA gene of B. abortus has been deposited in the GenBank database under accession number AF218367.

RESULTS

B. abortus ABC transporter is a homolog of R. meliloti ExsA.

A 1,791-bp ORF was identified in a 2.1-kb clone (Fig. 1). The nucleotide sequence of the B. abortus exsA gene and the amino acid sequence encoded by this gene are shown in Fig. 2. The coding region was translated by using the computer program DNASIS V5.00. This ORF and this protein sequence were subjected to nucleic acid and protein homology searches by using the Basic Local Alignment Search Tool (BLAST) service of the National Center for Biotechnology Information. These searches revealed that the ORF encoding the Brucella ABC transporter amino acid sequence has 69 and 67% identity with the ABC transporter-encoding exsA genes of R. meliloti and M. loti, respectively. On the basis of the alignment of the Brucella ExsA amino acid sequence with R. meliltoti and M. loti ExsA sequences, we determined that Brucella ExsA is probably missing 5 to 15 amino acids, including the start codon, in our clone (Fig. 3). The Brucella exsA gene encodes a polypeptide consisting of at least 596 amino acids with a predicted molecular mass of approximately 66 kDa.

FIG. 2.

Nucleotide sequence of the B. abortus exsA gene and deduced amino acid sequence. The asterisk indicates the termination code.

FIG. 3.

Alignment of ABC transporter protein (ExsA) amino acid sequences from R. meliloti (Rm), M. loti (Ml), and B. abortus (Ba). Three typical ABC transporter motifs, Walker A and B and the SGG (Q) ABC signature, are indicated by gray boxes. Gaps are indicated by dashes. Asterisks indicate identical residues, colons indicate conserved substitutions, and periods indicate semiconserved substitutions. The percentages of identity with the amino acid sequence of the ABC transporter protein (ExsA) from B. abortus are indicated in parentheses.

B. abortus ExsA contains all ABC transporter motifs.

Analysis of the B. abortus ExsA sequence revealed the presence of Walker A and B motifs and an ABC signature that are conserved ATP-binding domains in the ABC transporter protein family (Fig. 3). B. abortus ExsA has conserved features of a typical ABC transporter. First, ABC transporters possess typical ATP-binding motifs, such as Walker A (GXSGXGKST), Walker B (hhhhDEXT), and the ABC signature, where X and h indicate any amino acid and hydrophobic amino acid residues, respectively (Fig. 3). Furthermore, the Kyte-Doolittle analysis of hydrophobicity revealed that Brucella ExsA is a typical transmembrane protein containing five membrane-spanning segments at the amino acid terminal region, like R. meliloti and M. loti ExsA (Fig. 4). Additionally, an ABC transporter ortholog group analysis performed by Junko Yabuzaki and Minoru Kanehisa (Kyoto University, Kyoto, Japan) revealed that B. abortus ExsA belongs to ortholog group 6 (data not shown), which includes toxin and glucan transporters which are important bacterial virulence factors (18, 35). Therefore, according to Tomii and Kanehisa (32), ABC transporters classified in ortholog group 6 are more likely to be involved in bacterial pathogenesis than transporters which belong to other ABC transporter ortholog groups.

FIG. 4.

Hydrophobic profiles of ExsA proteins of B. abortus (A), R. meliloti (B), and M. loti (C). The analysis was performed by using the Kyte-Doolittle method. The vertical axis indicates the relative hydrophobicity (above the horizontal line) or hydrophilicity (below the horizontal line). The boxes indicate the potential transmembrane regions, and the numbers at the bottom are the amino acid positions.

Characterization of B. abortus ΔexsA deletion mutant.

Before constructing a B. abortus ΔexsA deletion mutant, we determined the copy number of the exsA gene in B. abortus. Southern blot analysis revealed that a single copy of the exsA gene is present in the B. abortus genome (data not shown). Furthermore, a defined Kan^r Amp^s ΔexsA deletion mutant of B. abortus S2308 was constructed by chromosomal gene replacement. After electroporation, transformants were examined by replica plating on brucella agar containing kanamycin and ampicillin or kanamycin alone. Ten clones of the two phenotypes, Kan^r Amp^r and Kan^r Amp^s, were selected to confirm that the exsA-deficient (Kan^r Amp^s) clones were the result of an allelic exchange. Chromosomal DNA was isolated from these clones and from the parental strain for Southern blot analysis. The same hybridization profile was observed for all transformants selected from each different phenotype group, as shown in Fig. 5. DNA hybridization of EcoRI-digested chromosomal DNA by using the exsA probe produced one fragment at approximately 4.8 kb for B. abortus S2308 (Fig. 5A, lane 4) and a 6.0-kb band for the Kan^r Amp^s ΔexsA mutant (lane 1). The 1.2-kb difference between the exsA hybridizing fragment present in the Kan^r Amp^s mutant (lane 1) and the fragment in B. abortus S2308 (lane 4) was due to insertion of the kanamycin cassette. This profile indicates that a double crossover between the deletion plasmid and homologous chromosomal DNA occurred. A single recombination event gave rise to clones (lanes 2 and 3) that produced two fragments, one that corresponded to the wild-type exsA gene and one that comprised the deletion plasmid. These two bands appeared in the Southern blot profile because the pGR64 deletion plasmid had one EcoRI restriction site present in the polylinker. This result was confirmed when the Amp^r probe hybridized to only one fragment corresponding to integration of the deletion plasmid (pGR64) in the chromosome (Fig. 5B). When the kanamycin cassette was used as a probe, it hybridized to genomic DNA from Kan^r Amp^s or Kan^r Amp^r clones but not to genomic DNA of wild-type B. abortus S2308, which was used as a negative control (Fig. 5C). The size difference in the fragments observed in Fig. 5C between the Kan^r Amp^s and Kan^r Amp^r clones again was due to integration of the deletion plasmid in the ampicillin-resistant bacterial cells.

FIG. 5.

Characterization of the B. abortus ΔexsA mutant by Southern blot analysis. EcoRI-digested genomic DNA was probed with the exsA gene (A), Amp^r (B), or Kan^r (C) DNA fragments. Lanes 1, B. abortus ΔexsA mutant; lanes 2 and 3, clones in which a single recombination event took place; lanes 4, B. abortus S2308. Lanes MW contained molecular weight markers.

Attenuation of B. abortus ΔexsA mutant in BALB/c mice.

The ability of B. abortus to persist within BALB/c mice has been shown to correlate with virulence in the natural host (12). Groups of mice were inoculated i.p. with the ΔexsA mutant or B. abortus parental strain S2308 to determine differences in persistence in the mouse model. The Brucella CFU were evaluated 1, 2, and 6 weeks postinfection in the spleen of each animal. The B. abortus ΔexsA mutant strain displayed reduced virulence at all times tested compared to the virulence of the wild type (Fig. 6). The smallest difference in CFU between the ΔexsA mutant and S2308 was observed at week 2 (log 0.23), and the greatest difference was observed at week 6 (log 0.92). Nevertheless, the log CFU difference between these two strains was statistically significant at all times studied.

FIG. 6.

Persistence of B. abortus S2308 and the B. abortus ΔexsA mutant in BALB/c mice. Eight mice were infected i.p. with a dose of 10⁵ CFU of bacteria. Spleens were harvested at different times, and the number of CFU in disrupted tissue was determined by 10-fold serial dilution and plating. The values are means ± standard deviations. The asterisks indicate statistically significant differences between the results obtained for the group that received the mutant and the results obtained for the group that received B. abortus parental strain S2308 (P < 0.05).

Immunoprotection conferred by vaccination with the B. abortus ΔexsA mutant strain.

To determine if the B. abortus ΔexsA mutant strain is able to induce protective immunity against infection, mice immunized with the ΔexsA mutant or with strain S19 or RB51 were challenged with the B. abortus virulent strain (S2308). The numbers of bacterial CFU in the spleens were determined 6 weeks after challenge, since Araya et al. (3) showed that nonspecific resistance to infection with unrelated bacteria is very low 6 weeks after immunization with Brucella. At this time, mice immunized with the B. abortus ΔexsA mutant strain had significantly (P < 0.05) fewer splenic brucellae than nonimmunized animals (Table 2). Additionally, we observed higher log units of protection in mice immunized with the ΔexsA mutant strain (2.74 log units) than in mice immunized with commercial vaccine strain S19 (1.44 log units) or RB51 (0.84 log unit) following challenge. Thus, the B. abortus ΔexsA mutant strain induced significantly enhanced resistance to experimental infection.

TABLE 2.

Protection of mice against challenge with B. abortus S2308 after immunization with the B. abortus ΔexsA mutant or with the B. abortus S19 or RB51 vaccine strain

Vaccine	Log10 CFU of B. abortus S2308 in spleen (mean ± SD)a	Log10 units of protection
PBS control	7.38 ± 0.16
B. abortus RB51	6.54 ± 0.30	0.84b
B. abortus S19	5.94 ± 0.78	1.44b
B. abortus ΔexsA	4.64 ± 0.77	2.74b,c

^aMice were immunized with 10⁵ CFU and 6 weeks later were challenged i.p. with 10⁵ CFU of strain S2308.

^bSignificantly different from value for the PBS control group (P < 0.05).

^cSignificantly different from value for the group immunized with S19 and from value for the group immunized with RB51 (P < 0.05).

DISCUSSION

Bacterial pathogens modulate their gene expression to adapt to the various environments to which they are exposed during the course of infection. The mechanism of Brucella virulence is not yet fully understood. Brucella sensing unfavorable conditions produces diverse gene products that contribute to the survival and multiplication of this bacterium inside host cells (13). Understanding the molecular events that allow brucellae to reach an intracellular niche where the bacteria freely multiply should help determine the mechanism of pathogenesis. Little is known regarding the nature of bacterial factors responsible for Brucella entry and multiplication in different cell types. An increasing number of pathogenic bacteria are being found to secrete effector molecules that affect host cell functions (8, 10). In Brucella infection, the type IV secretion system with which this bacterium transports macromolecular complexes across its envelope has been described as an important virulence factor (5, 23). Therefore, the cell secretion machinery that uses bacterial transporters is critical for the maintenance of Brucella pathogenesis.

In this study, we identified, sequenced, and disrupted the exsA gene of B. abortus. The deduced amino acid sequence for the ORF from clone 64 exhibited homology to ABC transporters of different bacterial species, mainly ExsA of R. meliloti and M. loti, with which the levels of sequence identity were 69 and 67%, respectively. The ABC transporters are a major class of cellular translocation machinery in all bacterial species (32). Based on computer analysis of the predicted amino acid sequence, Walker and ABC signature motifs that could form an ATP-binding site were identified in the C-terminal domain of Brucella ExsA. Additionally, the ABC transporter transmembrane region in the highly hydrophobic segment was predicted. In R. meliloti, the exsA gene encodes an ABC transporter involved in EPS I transport (4). EPS I is essential for the invasion of alfalfa root nodules by R. meliloti. Mutants of R. meliloti devoid of EPS I are unable to establish an effective symbiosis with alfalfa (22). Since like Rhizobium and Mesorhizobium, Brucella belongs to the alpha-2 subgroup of the Proteobacteria (21), whose members are characterized by their ability to interact pericellularly or intracellularly with eukaryotic cells, we suggest that Brucella ExsA functions as a polysaccharide transporter and may be crucial for the maintenance of Brucella pathogenesis.

To address the role of Brucella ExsA in bacterial virulence, a mutant was constructed. In mice the ΔexsA mutant exhibited a different level of spleen colonization than wild-type strain S2308, indicating that virulence in vivo was altered by the absence of ExsA (Fig. 6). Brucella ExsA also showed a high degree of homology to the Sinorhizobium meliloti MsbA-like saccharide-exporting ABC transporter (data not shown). An msbA mutant strain of Francisella novicida, a facultatively intracellular bacterium like Brucella that survives and grows in macrophages by preventing phagolysosomal fusion, was unable to grow in macrophages and mice, and it was found to be sensitive to serum (20). Therefore, we speculate that the Brucella ΔexsA mutant has an altered polysaccharide architecture, as demonstrated previously for E. coli (9, 11), and therefore is sensitive to complement-mediated lysis and/or phagocyte-mediated killing mechanisms. Macrophages and neuthophils are potent killers of Brucella, and this bacterium is extremely sensitive to the myeloperoxidase-halide-peroxide system found in the host phagocyte system (7). This phenomenon has been demonstrated in B. abortus LPS mutants (1). Even though changes in polysaccharide structure enhanced bacterial clearance, the B. abortus ΔexsA mutant has a smooth phenotype, like parental strain S2308 (data not shown).

Besides its importance in helping understand Brucella pathogenesis, our study also could have an impact on vaccine development. Bacterial vaccines are usually based on either live or inactivated whole-cell or subunit preparations (6). Live vaccines for intracellular organisms are generally more efficacious and consist of attenuated variants of particular pathogens that have lost the ability to cause clinical disease but are still able to establish self-limiting infections and hence immune responses in the hosts (36). As genes required for intracellular growth and survival of brucellae are identified, attenuated variants harboring defined genetic defects can be evaluated. Rational genetic attenuation of a pathogen could lead to a new generation of live bacterial vaccines that are safer and do not revert to full virulence. As demonstrated here, the B. abortus ΔexsA mutant induced superior protective immunity compared to the protective immunity induced by the Brucella vaccine strains currently available, S19 and RB51. RB51 induced lower resistance to challenge with the pathogenic strain S2308, probably due to rapid clearance of this strain (30) and to the absence of antibodies against LPS (31). As for strain S19, we suggest that a lower level of protection was induced in mice because of more rapid bacterial clearance with this strain than with the ΔexsA mutant. At 6 weeks postinfection, B. abortus S19 was cleared more rapidly than the B. abortus ΔexsA mutant (data not shown). Ideally, a live-vaccine strain conferring solid immunity without host restrictions would be a significant improvement over available vaccines.