Putative ATP-Binding Cassette Transporter Is Essential for Brucella ovis Pathogenesis in Mice

Abstract

Brucella ovis is a major cause of reproductive failure in sheep, which is associated with epididymitis and infertility in rams. Importantly, B. ovis is one of the few Brucella species that is not zoonotic. Due to the scarcity of studies on B. ovis infection, a murine model of infection was developed. The roles of B. ovis genes encoding a putative hemagglutinin and an ABC transporter were investigated in the mouse model. The kinetics of B. ovis infection were similar in BALB/c and C57BL/6 mice, and both strains of mice developed multifocal microgranulomas in the liver and spleen, but only minimal colonization and histopathological changes were observed in the genital tract. Therefore, the mouse was considered a suitable infection model for B. ovis but not for B. ovis-induced genital disease. Two mutant strains were generated in this study (the ΔabcAB and Δhmg strains). The B. ovis ΔabcAB strain was attenuated in the spleens and livers of BALB/c mice compared to the wild-type (WT) strain (P < 0.001). Conversely, the Δhmg strain infected mice at the same level as WT B. ovis, suggesting that a putative hemagglutinin is not required for B. ovis pathogenesis. Additionally, the ΔabcAB strain did not survive in peritoneal macrophages, extracellularly in the peritoneal cavity, or in RAW 264.7 macrophages. Moreover, infection with the ΔabcAB strain was not lethal for male regulatory factor 1-knockout mice, whereas infection with the B. ovis WT strain was 100% lethal within 14 days postinfection. These results confirm that the predicted ABC transporter is required for the full virulence and survival of B. ovis in vivo.

Brucella ovis is one of the main causes of reproductive failure in sheep (6). The disease is characterized by chronic epididymitis, orchitis, and infertility in sexually mature rams and occasional abortion and stillbirth in ewes (4, 15, 46). B. ovis has a worldwide distribution in economically important sheep-raising areas, with the exception of Great Britain (6). This organism may affect as much as 46% of a herd (41), leading to significant economic losses for the sheep industry due to infertility and early culling (8). B. ovis is a stable, rough, Gram-negative coccobacilli that belongs to the Alpha-2-Proteobacteriacea family (4, 17), and it is one of the few classical Brucella species that are not pathogenic to humans (4, 47).

Brucella spp. are facultative intracellular bacteria that are able to survive and replicate in phagocytic and nonphagocytic cells and to establish chronic infections in animals and humans (18, 46). Virulence factors are required for the invasion of host cells by Brucella spp. and for their survival and replication in host cells. Although classical virulence factors, such as capsules and fimbriae, are absent in Brucella species (18), pathogenic mechanisms specific to Brucella spp. have been identified (20, 29, 39, 42). The mouse is often used as an animal model to investigate the pathogenesis of animal and human brucellosis (2, 13, 28). The murine model also allows studies that may reveal mechanisms underlying the species-specific pathogenicity of Brucella spp. However, few studies have characterized B. ovis infection in the mouse model, as studies generally investigate the mouse immune response against potential vaccines for B. ovis (9, 23, 24).

Recently, the genome of B. ovis strain ATCC 25840, originally isolated in Australia, was completely sequenced and analyzed, resulting in the identification of a 26.5-kb-specific island in chromosome II (GenBank accession number NC_009504) that comprises 28 open reading frames (ORFs) (48). This island, named B. ovis pathogenicity island 1 (BOPI-1), is absent in other classical Brucella species that were previously sequenced, including Brucella melitensis (12), Brucella suis (35), and Brucella abortus strains 9-941 and 2308 (10, 19). However, this genomic island was detected in three marine isolates of Brucella from bottlenose dolphins (48). PCR amplification targeting 12 of the previously described ORFs showed that the sequence was conserved in 18 B. ovis field isolates from various geographical locations (48). Additionally, all 12 analyzed sequences were absent in other bacterial species that potentially cause epididymitis in rams, as well as in species phylogenetically related to B. ovis, including Ochrobactrum intermedium and Ochrobactrum anthropi (51).

Characterization of BOPI-1 resulted in identification of genes encoding B. ovis proteins potentially involved in pathogenesis, including ORFs carrying an ATP-binding cassette (ABC) transporter (abcABCDE) and a putative hemagglutinin gene (hmg) (48). Bacterial ABC transporter systems are associated with nutrient uptake and the export of toxins and antibiotics, and they may play an important role in gene expression (27, 44). In Brucella spp., ABC transporter proteins have a potential pathogenic role during host infection (39). Hemagglutinins are responsible for adhesion and may define the specificity of bacterial adhesins during host infection (38). However, a previous study reported that a putative hemagglutinin does not have an important role in the pathogenesis of Brucella melitensis (36), although it is not the same hemagglutinin as that evaluated in this study. No previous studies have shown the role of these proteins in B. ovis pathogenesis during chronic infection of the animal. Importantly, study of these genes may elucidate the genital tract tropism of B. ovis in rams or its lack of virulence in humans (48). In this study, a murine model of B. ovis infection was developed and applied to investigate the requirements for a predicted ABC transporter and a hemagglutinin in the pathogenesis and survival of B. ovis in vivo.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The virulent B. ovis strain ATCC 25840 and two mutant strains were used in this study. All bacterial inocula were cultured on tryptic soy agar (TSA) plates with 1% hemoglobin (Becton Dickinson) for 3 days at 37°C in 5% CO₂ and then suspended in sterile phosphate-buffered saline (PBS). Inoculum concentrations were estimated by spectrophotometry in a SmartSpec instrument (Bio-Rad) at an optical density of 600 nm (OD₆₀₀), and the inocula were plated after serial 10-fold dilutions for counting CFUs. For bacteriology, mouse tissues and macrophages were plated on TSA medium with 1% hemoglobin for counting CFUs after 4 to 7 days of growth at 37°C in 5% CO₂. For counting the CFUs of mutant strains in tissues and macrophages, 100 μg/ml of kanamycin (Gibco; Invitrogen, Brazil) was added to TSA medium with 1% hemoglobin. To select mutant candidates, ampicillin (200 μg/ml) or kanamycin (100 μg/ml) was added to the medium. Although B. ovis is not zoonotic, all experiments were performed in a biosafety level 3 laboratory.

Generation of mutant strains.

Two mutant strains were generated in this study by the deletion of specific genes from chromosome II of B. ovis (GenBank accession number NC_009504) as previously described (48). One mutant strain (TMS2) lacks the putative hemagglutinin gene hmg (BOV_A0512; GenBank accession number 5203266), and the other mutant strain (TMS3) lacks two ORFs, abcA (BOVA_0500; GenBank accession number 5204038) and abcB (BOV_A0501; GenBank accession number 5203285), that encode a predicted ABC transporter.

To generate the TMS2 mutant strain (Δhmg), a 1,971-bp fragment was amplified by PCR using primers BOVA0512FW and BOVA0512RV (Table 1). The product was cloned into a pCR2.1TOPO-TA vector using a Topo cloning kit (Invitrogen, CA). The PCR product was excised by double digestion with BamHI and XhoI and cloned into pBluescript KS (Stratagene, CA). Then the kanamycin cassette (1.6 kb) from pUC4-KIXX (Amersham Pharmacia Biotech, United States) was digested by SmaI and cloned into pBluescript KS by digestion with EcoRV. The final plasmid, which has a KIXX cassette interrupting hmg, was named pBO6 (Table 1).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Description	Source
Strains
WT	B. ovis WT strain	ATCC 25840
TMS2 (ΔabcAB)	B. ovis ΔBOV2_500-501::kanr	This study
TMS3 (Δhmg)	B. ovis ΔBOV2_512::kanr	This study
XL1 Blue	E. coli strain used for cloning	Phoneutria
DH5α	E. coli strain used for cloning	Phoneutria
Plasmids
pUC4-KIXX	Plasmid with kanamycin resistance cassette (1.6 kb)	Amersham Pharmacia Biotech
pBluescript KS	Cloning vector	Stratagene
pCR1.1TOPO-TA	Cloning vector	Invitrogen
pBO4	PCR products of BOVII 518658-519618 and 521570-522550 separated by KIXX in pBluescript KS	This study
pBO6	PCR product of BOVII 530959-532701 separated by KIXX in pBluescript KS	This study

In order to generate the TMS3 mutant strain (ΔabcAB), upstream and downstream fragments (961-bp and 993-bp, respectively) of abcA and abcB were amplified by PCR using primer pairs BOVA0500FW/BOVA0500RV and BOVA0501FW/BOVA0501RV, respectively (Table 1). Both products were cloned into pCR2.1TOPO-TA. The upstream fragment was excised by double digestion with XbaI and HindIII and then cloned into pBluescript KS. The downstream fragment was excised by double digestion with HindIII and XhoI and cloned into the same plasmid. A kanamycin resistance cassette from pUC4-KIXX was digested with HindIII and cloned between up- and downstream fragments. The plasmid with ABC transporter fragments interrupted by the kanamycin cassette was named pBO4 (Table 1).

The correct sequences and directions of cloned fragments in pBO4 and pBO6 were verified by sequence analysis. Plasmids were transformed into Escherichia coli, extracted by a commercial kit (plasmid midi-kit; Qiagen), and transformed into electrocompetent B. ovis ATCC 25840 cells by electroporation as previously described (45). After electroporation of B. ovis cells, bacteria were suspended in 1 ml of SOCB medium (2% tryptone, 0.5% yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl₂, 10 mM MgSO₄, and 20 mM glucose), and 100 μl of the solution was plated in TSA medium without antibiotic and incubated for 16 h at 37°C. Colonies that grew on TSA medium were recovered, dissolved in 100 μl of sterile PBS solution, and plated once again in TSA medium with kanamycin. In addition, the rest of the bacterial solution was incubated in a shaker for 16 h at 37°C and then plated in TSA medium with kanamycin. Colonies that were kanamycin resistant and ampicillin sensitive were selected as mutant candidates.

Confirmation of mutant strains by PCR.

Deletion of the ABC transporter (ΔabcAB) or putative hemagglutinin gene (Δhmg) was confirmed by targeting the specific deleted region, using primer pairs BO4FW and BO4RV or BO6FW and BO6RV, respectively (Table 2). Mutant candidates were negative only for their specific PCRs. Both mutant strains were verified by using primers BrucellaP31-FW and BrucellaP31-RV to target the bcsp31 gene, which is highly conserved in Brucella spp. (Table 2). B. ovis strain ATCC 25840 was used as a positive control for all reactions.

TABLE 2.

Primers used in this study

Primer name	Sequence (5′-3′)a	Restriction enzyme	Product size (bp)
BOVA0500FW	TCTAGACATATGTGCCAGCATCAC	XbaI	961
BOVA0500RV	AAGCTTGATTTCGAGTAATAGACC	HindIII
BOVA0501FW	AAGCTTCACTCATATCTTGCGGGT	HindIII	993
BOVA0501RV	CTCGAGGGCTTGGCTCTTTGCTGA	XhoI
BOVA0512FW	GGATCCTCATAGCTCAATGCCGTT	BamHI	1,971
BOVA0512RV	CTCGAGTCGACTGGAGGCTGTACA	Xho
BO4-FW	TGGTATCTTCAGCCGTTCCAAG		Negb
BO4-RV	ATCTTTGCCCGTTCCAGTCG		135c
BO6-FW	TTCAGGCGACTGCTAATGGCAC		Negd
BO6-RV	AAACCGATACCTCATCCCCGAG		225c
BrucellaP31-FW	TGGCTCGGTTGCCAATATCAA		223
BrucellaP31-RV	CGCGCTTGCCTTTCAAGGTCTG

^aUnderlined sequences are restriction enzymes.

^bThe size of the ΔabcAB mutant was negative per the specific PCR used, which targeted the deleted region.

^cWT B. ovis strain ATCC 25840.

^dThe size of the Δhmg mutant was negative per the specific PCR used, which targeted the deleted region.

For DNA extraction, selected colonies of WT B. ovis or the mutant strains were suspended in 100 μl of sterile water, inactivated at 100°C for 10 min, and centrifuged for 2 min at 12,000 × g. For PCR, 2 μl of the supernatant with DNA was added to a solution containing 23 μl of a commercial PCR mix (PCR SuperMix; Invitrogen, Brazil), 0.5 μl of each primer (25 μM), and 0.25 μl of Taq polymerase (Invitrogen, São Paulo, Brazil). Cycling parameters were as follows: denaturation at 94°C for 5 min; 30 cycles of denaturation at 94°C for 30 s, annealing at 60°C for 30 s, and an extension at 72°C for 1.5 min; and a final extension at 72°C for 10 min. PCR products were resolved by 1% agarose gel electrophoresis. PCRs were considered positive when their products were 135 bp, 225 bp, and 223 bp when primers BO4, BO6, and BrucellaP31, respectively, were used (Table 2).

In vitro growth of B. ovis WT and mutant strains.

Because B. ovis does not grow adequately in liquid medium without hemoglobin, the in vitro growth of B. ovis mutant strains was evaluated in solid media. Inocula with 10³ CFU/ml of the B. ovis ΔabcAB or Δhmg mutant strain or the WT strain were prepared with PBS solution, and 100 μl of each solution was plated on TSA medium with 1% hemoglobin without antibiotics. All plates were incubated at 37°C in 5% CO₂, and colonies were harvested from the plates at 0, 12, 24, 48, 72, 96, and 120 h, using 1 ml of sterile PBS solution at 0 to 48 h and 2 ml at 72 to 120 h. For counting CFUs, 100 μl of the total PBS solution recovered from each plate was submitted to serial 10-fold dilutions and plated in duplicate on TSA medium without antibiotics. The total number of CFU per milliliter was determined for each time point. This experiment was performed in triplicate.

Mouse infection.

Animal experiments were approved by the Animal Ethics Committee (CETEA, UFMG, protocol no. 136/08). Male BALB/c and C57BL/6 mice 7 to 9 weeks old were obtained from the Instituto de Ciências Biológicas (ICB, UFMG, Brazil).

Interferon regulatory factor type 1-knockout mice (IRF-1 KO) were used to compare the virulence of the B. ovis WT strain and the attenuated mutant strains. IRF-1 KO male mice 7 to 12 weeks old were divided into groups of five IRF-1 KO mice and infected intraperitoneally (i.p.) with approximately 2.3 ×10⁶ CFU/animal of the B. ovis WT or ΔabcAB strain. All infected mice were monitored daily for survival until 21 days postinfection (dpi).

Bacteriology.

Fragments of tissue were aseptically collected from infected mice, weighed in falcon tubes, and homogenized in 2 ml of a sterile PBS solution by using a mixer (Hamilton Beach, United States). Then each sample was submitted to serial 10-fold dilutions and plated for counting CFUs. After 4 to 7 days of growth at 37°C in 5% CO₂, the number of CFUs per gram of organ (spleen and liver) or the total number of CFUs per organ (testis, epididymis, and seminal vesicle) was estimated.

Histopathology.

Fragments of spleen, liver, testis, epididymis, and seminal vesicle from infected mice were evaluated for histopathological changes. Tissue fragments were fixed in 10% formalin for 24 h, paraffin embedded, sectioned, and stained with hematoxylin and eosin. Inflammatory lesions in the spleen and liver were scored from 0 (absent) to 3 (severe). Tissue fragments from mice inoculated with PBS were used as controls.

Immunohistochemistry.

In order to associate histopathologic lesions with intralesional B. ovis colonies, immunohistochemistry was performed as previously described (50) for all tissue fragments. Briefly, tissue sections were hydrated and incubated with 10% hydrogen peroxide in PBS for 30 min. After being washed with PBS, slides were transferred to a humid chamber at room temperature, incubated with 25 mg/ml of skim milk for 45 min, and then incubated with a primary antibody for 30 min. For immunolabeling, diluted (1:5,000) serum from a rabbit experimentally inoculated twice (at a 1-month interval) with 1 × 10⁹ CFUs of B. ovis (strain ATCC 25840) was used as the primary antibody. Then tissue sections were washed with PBS, incubated with secondary antibody for 20 min, washed again with PBS, and incubated for 20 min with streptavidin-peroxidase from a commercial kit (LSAB⁺ kit; Dako Corporation, Carpinteria, CA). The reaction was revealed using 0.024% diaminobenzidine (DAB; Sigma), and sections were counterstained with Mayer's hematoxylin.

In vivo infection of peritoneal macrophages.

To evaluate the kinetics of early B. ovis infection in the mouse peritoneal cavity, four groups of male BALB/c mice (n = 4) were infected with approximately 1 × 10⁶ CFUs per animal of the B. ovis WT strain or the ΔabcAB mutant strain by peritoneal injection. At 6, 12, and 24 h postinfection (hpi), mice were euthanized by cervical dislocation and peritoneal macrophages were harvested as previously described (52). Briefly, 10 ml of cold Dulbecco PBS was injected i.p., immediately aspirated, and transferred to a 15-ml falcon tube on ice. To estimate the total levels of infection, i.e., the levels of extracellular and intracellular infection, 100 μl of the total PBS solution harvested from the peritoneal cavity was submitted to serial 10-fold dilutions in sterile water and plated for counting CFUs. To estimate the levels of intracellular infection in peritoneal macrophages, the same PBS solution as that recovered from the peritoneal cavity was centrifuged at 400 × g for 10 min at 4°C. Then the supernatant was discarded and macrophages were suspended in 500 μl of RPMI solution (Gibco; Invitrogen, Carlsbad, CA) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 50 μg/ml of gentamicin. In order to kill extracellular bacteria, all samples were incubated for 1 h at 37°C in 5% CO₂. After another centrifugation and washing of macrophages with RPMI solution without antibiotics, 300 μl of 0.01% Triton X solution was added, followed by vigorous vortexing, washing with PBS, serial 10-fold dilutions, and plating for counting CFUs.

In addition, to evaluate early systemic B. ovis infection in mice, spleens were collected for bacteriology at 6, 12, and 24 hpi and compared to those from mice infected with the ΔabcAB mutant.

In vitro infection of RAW 264.7 murine macrophages.

The murine macrophage cell line RAW 264.7 was cultured in RPMI medium (Gibco; Invitrogen) supplemented with 10% FBS. Cells were distributed in 96-well culture plates (5 × 10⁵ macrophages/well) and incubated at 37°C with 5% CO₂. Macrophages were infected with the B. ovis ΔabcAB or Δhmg mutant strain or the WT strain at a multiplicity of infection (MOI) of 1:100. Then culture plates were centrifuged at 1,000 × g for 5 min at 15°C and incubated at 37°C for 30 min. Macrophages were washed once with a sterile PBS solution and then incubated with RPMI solution supplemented with 10% FBS and 50 μg/ml of gentamicin (Invitrogen, São Paulo, Brazil) at 37°C for 1 h to kill the extracellular bacteria. Afterwards, each well was washed once with sterile PBS, and macrophages were lysed with sterile distilled water for 20 min at 0, 4, 24, and 48 hpi. Intracellular bacteria recovered from lysed macrophages were submitted to serial 10-fold dilutions in PBS and were plated in duplicate on TSA medium with or without kanamycin. The numbers of intracellular CFUs were determined after 4 days of incubation at 37°C with 5% CO₂. This experiment was performed in triplicate and repeated three times.

Sensitivity assays.

The sensitivities of the B. ovis WT and ΔabcAB mutant strains to bactericidal peptides were evaluated using polymyxin B as previously described (7). Briefly, bacterial suspensions with approximately 10³ CFUs of bacteria were incubated with 0.5 mg/ml or 1 mg/ml of polymyxin B (Invitrogen) or with 100 μl of PBS as a negative control. After being incubated at 37°C with 10% CO₂ for 1 h, the content of each well was diluted and plated for counting CFUs. The percentages of bacteria surviving were determined in relation to the numbers of CFUs recovered from wells with PBS (100% survival). Additionally, resistance to complement killing was evaluated as previously described (5). Serum was obtained from a nonimmune rabbit, and half of the serum was inactivated at 56°C for 60 min (serum negative control). Briefly, bacterial suspensions (100 μl/well) with approximately 1 × 10⁶ CFU/ml of bacteria were placed into wells with 100 μl of either inactivated serum or fresh serum at different concentrations (i.e., 12.5%, 25%, and 50%). After being incubated at 37°C for 1 h, the solution in each well was diluted and plated on TSA plates for counting CFUs. The percentage of bacteria surviving was determined in relation to the number of CFUs recovered after treatment with heat-inactivated serum (100% survival). Both experiments were performed in triplicate and repeated twice.

Statistical analyses.

All CFU data were logarithmically transformed, expressed as means and standard deviations, and submitted to analysis of variance (ANOVA). Means were compared with Tukey's test (GraphPad InStat 3). The lethality rates in the two groups of IRF-1 KO mice were compared using Fisher's exact test (GraphPad InStat 3). The spleen weights and histopathological lesion scores of the two groups were compared using the Mann-Whitney nonparametric test, and the Kruskal-Wallis nonparametric test was used for comparing more than three groups (GraphPad InStat 3). For sensitivity assays, CFU data were logarithmically transformed and submitted to ANOVA, and the means were compared with the Student Newman-Keuls test (GraphPad InStat 3). P values of less than 0.05 were considered significant.

RESULTS

Kinetics of B. ovis infection in male mice.

Because there have been few studies of B. ovis infection in the murine model, in this study an infection model in male mice was developed and characterized. Initially, an experiment was performed to determine the minimal dose of B. ovis that results in systemic infection in the mouse. Male BALB/c mice (n = 4) were inoculated with 4 × 10⁴, 4 × 10⁵, 4 × 10⁶, or 4 × 10⁷ CFU/animal of the virulent B. ovis strain ATCC 25840 by intraperitoneal injection. At 7 dpi, all mice were euthanized and fragments of spleen and liver were collected for bacteriology. CFUs per gram of organ were estimated for different doses and compared. No statistically significant differences were observed for these doses, and doses higher than 4 × 10⁵ CFUs resulted in similar bacterial loads in the spleen and liver (Fig. 1). Mice infected with 4 × 10⁶ CFUs of B. ovis showed the smallest variation in CFUs, and this dose was selected for all further experiments.

FIG. 1.

Evaluation of infectious doses of B. ovis in male BALB/c mice. Mice were infected with 4 × 10⁴, 4 × 10⁵, 4 × 10⁶, or 4 × 10⁷ CFU/animal of the B. ovis WT strain by the peritoneal route. Each data point represents the number of CFUs recovered from the spleen (A) or liver (B) of each animal at 7 dpi. A circle represents each animal, and a line represents the standard deviation for 4 mice at each time point. No statistical significance was observed. Raw data were logarithmically transformed prior to ANOVA, and means were compared by Tukey′s test.

After a suitable inoculum was selected, the kinetics of B. ovis infection were characterized in male BALB/c and C57BL/6 mice. BALB/c and C57BL/6 mouse strains are known to be susceptible and resistant, respectively, to B. abortus infection (14, 33). Four groups of each mouse strain (n = 4) were infected i.p. with 1.3 × 10⁶ CFUs per animal of B. ovis strain ATCC 25840. Control groups were inoculated with 100 μl of sterile PBS. Mice were euthanized at 0 (control group), 1, 7, 30, and 90 dpi, and fragments of spleen, liver, and the genital tract, including both testes, epididymides, and seminal vesicles, were collected for bacteriology.

As shown in Fig. 2, higher numbers of CFUs were recovered from the spleens of both mouse strains at 1 and 7 dpi, followed by a significant decrease in CFUs at 30 dpi. In addition, BALB/c mice showed higher numbers of CFUs than C57BL/6 mice in the spleen at 7 (P < 0.05) and 30 (P < 0.01) dpi, which suggests that initially BALB/c mice are more susceptible to B. ovis infection (Fig. 2A). In both mouse strains, the bacterial load in the liver was lower than in the spleen at all time points during the course of infection. However, the kinetics of B. ovis infection in the liver was similar to that described in the spleen, with decreased CFUs at 30 dpi. Interestingly, the numbers of CFUs recovered from the livers of BALB/c mice and C57BL/6 mice were similar at 1, 7, and 30 dpi, whereas BALB/c mice showed higher bacterial loads only at 90 dpi (P < 0.05). In the genital tract, similar numbers of CFUs of B. ovis were recovered from testes, epididymides, and seminal vesicles during the early stages of infection in male BALB/c and C57BL/6 mice (Fig. 2C to E). In C57BL/6 mice, these numbers decreased at 30 dpi and minimal bacterial loads were detected in testes and seminal vesicles. In BALB/c mice, epididymides and seminal vesicles showed high levels of variation in numbers of CFUs during the course of infection, which explains the lack of statistical significance for CFUs in BALB/c seminal vesicles (Fig. 2E). Therefore, these data demonstrate that B. ovis establishes a systemic infection in the murine model but not a persistent infection in the male genital tract.

FIG. 2.

Kinetics of B. ovis infection in male BALB/c and C57BL/6 mice. Mice were infected with 1 × 10⁶ CFUs of the B. ovis WT strain by peritoneal injection. Fragments of spleens (A), livers (B), testes (C), epididymides (D), and seminal vesicles (E) were collected for counting of CFUs at 1, 7, 30, and 90 dpi. Each bar represents the mean for 4 mice at each time point, and standard deviations are shown. Raw data were logarithmically transformed prior to ANOVA, and means were compared by Tukey's test. Significant differences between BALB/c and C57BL/6 mice are indicated by asterisks (*, P < 0.05; **, P < 0.01). For each mouse strain, significant differences between time points are indicated by different letters. A horizontal line indicates the detection limit for CFUs for each organ.

Splenomegaly in BALB/c and C57BL/6 mice during B. ovis infection was evaluated at 7, 30, and 90 dpi. Percentages of total spleen weight in relation to mouse body weight were determined, and the results for the control and infected groups of both mouse strains were compared, as shown in Fig. 3. The spleen weight in both mouse strains was significantly higher than in the control group at 7 dpi, although the splenomegaly was more severe in BALB/c mice until 30 dpi.

FIG. 3.

Splenomegaly in male BALB/c and C57BL/6 mice during B. ovis infection. Mice were infected with 1 × 10⁶ CFUs of the B. ovis WT strain, and the weight of the spleen was compared to the body weight of the mouse at 7, 30, and 90 dpi and given as a percentage. The control group was inoculated with a sterile PBS solution. Data were analyzed by the Kruskal-Wallis nonparametric test. Significant differences between the control group and the infected group are indicated by asterisks (*, P < 0.05).

B. ovis infection leads to histopathological lesions in mice, although it does not have a clear tropism for the male genital tract.

To characterize the inflammatory lesions caused by B. ovis infection in BALB/c and C57BL/6 mice, fragments of spleens, livers, testes, epididymides, and seminal vesicles were collected at 1, 7, 30, and 90 dpi and evaluated by histopathology and immunohistochemistry (Fig. 4). In the spleen, the inflammatory changes were characterized by mild to moderate multifocal microgranulomas next to the white pulp in both mouse strains, especially after 7 dpi (Fig. 4A and B). In the liver, all mice showed multifocal microgranulomas at 7 and 30 dpi characterized by histiocytic infiltrates with epithelioid macrophages (Fig. 4C and D). Average scores were given to histological lesions in the spleens and livers of BALB/c and C57BL/6 mice (Table 3). The spleen and liver scores were similar for the two mouse strains during the course of infection. In the early stages of B. ovis infection (1 dpi), the inflammatory changes were absent from or rarely seen in either mouse strain, even though all mice showed high numbers of CFUs in the spleen and liver. The average scores were higher at 7 or 30 dpi for both mouse strains, which shows that histopathological lesions in these organs are more evident during the chronic stages of B. ovis infection and are associated with a high bacterial load. In addition, mild B. ovis immunolabeling was observed in the cytoplasms of macrophages associated with hepatic microgranulomata in BALB/c and C57BL/6 mice, mainly at 7 dpi (Fig. 4G). The detection of intralesional bacteria confirms that the inflammatory lesions previously described are due to systemic B. ovis infection in the murine model.

FIG. 4.

Histopathology and immunohistochemistry of male BALB/c mice infected with B. ovis. (A and C) Spleen (A) and liver (C) from the BALB/c control group. (B and D) Multifocal microgranulomas in the spleen (B) of a B. ovis-infected BALB/c mouse at 30 dpi and in the liver (D) at 7 dpi. Arrows indicate microgranulomas. Bars, 100 μm. (E) Epididymis from the BALB/c control group. (F) Mild neutrophilic and histiocytic periepididymitis in a B. ovis-infected BALB/c mouse at 7 dpi. The arrow indicates a focal inflammatory infiltrate. Bars, 400 μm. (G) Intralesional B. ovis immunolabeling in a hepatic microgranuloma from a BALB/c mouse at 7 dpi. Arrows indicate immunolabeled organisms. Bar, 70 μm (H) Mild B. ovis immunolabeling next to the epididymis and associated with neutrophilic and histiocytic infiltrate in a BALB/c mouse at 30 dpi. Arrows indicate immunolabeled organisms. Bar, 100 μm. The same lesions were seen in male C57BL/6 mice.

TABLE 3.

Average score for inflammatory lesions in spleen and liver of BALB/c and C57BL/6 male mice infected with Brucella ovis ATCC 25840

Mouse	Score for spleen on indicated day postinfectiona				Score for liver on indicated day postinfectiona
	1	7	30	90	1	7	30	90
BALB/c	0B	1.25A	1.13AB	0.63AB	0B	1.63A*	1.13AB	0.88AB
C57BL/6	0B	0.38AB	1.05A	0.13AB	0.25B	1.88A	1.50AB	0.50AB

^aDifferent uppercase letters (A and B) in the same line indicate significant differences (P < 0.05) among time points in each mouse strain. No significant differences were observed between BALB/c and C57BL/6 mice at any time point. *, P < 0.01.

In the male genital tract, histopathological lesions were characterized by mild neutrophilic and histiocytic infiltrates adjacent to the testes and epididymides and, less frequently, in the seminal vesicles and adjacent adipose tissue (Fig. 4E and F). Only one BALB/c mouse had a moderate histiocytic and neutrophilic interstitial vesiculitis at 30 dpi. Periorchitis and periepididymitis were observed bilaterally in all infected mice. Additionally, these lesions were associated with mild B. ovis immunolabeling in the male genital tract (Fig. 4H). Given that the bacterial load in the mouse genital tract decreased significantly during systemic B. ovis infection and that nearly all histopathological lesions were restricted to the adjacent tissues, these data show that B. ovis does not have a clear tropism for the genital tract in the mouse.

Characterization of the B. ovis ΔabcAB and Δhmg mutant strains.

The whole-genome sequence of B. ovis was recently published and resulted in the identification in chromosome II of BOPI-1, which potentially encodes specific pathogenic proteins (48). In order to investigate the roles of a predicted ABC transporter and a hemagglutinin during B. ovis infection in the mouse, mutant strains were generated by deleting the putative hemagglutinin gene hmg (TMS2) or two ORFs (abcA and abcB) that putatively encode an ABC system (TMS3). Both mutant strains were confirmed by PCR (data not shown). The B. ovis Δhmg strain did not amplify the BOV0512 region (225 bp), and the B. ovis ΔabcAB strain did not amplify a 135-bp region of BOVA0500/501. Other B. ovis genes analyzed in this study were conserved in these mutant strains, including bcsp31 (223 bp), which is highly conserved in the Brucella genus (1). All genes were detected in B. ovis strain ATCC 25840.

The two mutant strains grew similarly in vitro in TSA medium with 1% hemoglobin, as did the B. ovis WT strain. As shown in Fig. 5, the B. ovis Δhmg and ΔabcAB mutants and the WT strain were in the exponential growth phase between 24 and 72 h of incubation at 37°C with 5% CO₂ and then entered the stationary growth phase.

FIG. 5.

In vitro growth curves of B. ovis mutant strains. The B. ovis ΔabcAB, Δhmg, and WT strains were grown on TSA plates with 1% hemoglobin at 37°C in 5% CO₂. Data points represent averages of triplicate experiments at each time point. Standard deviations are shown.

abcAB, but not hmg, is required for WT levels of B. ovis infection in mice.

To evaluate infection with the B. ovis mutant strains in the murine model, male BALB/c mice were infected with approximately 1 × 10⁶ CFUs per animal of the B. ovis Δhmg mutant strain (n = 4), the B. ovis ΔabcAB mutant strain (n = 4), or the B. ovis wild-type WT strain (n = 8). Mice were euthanized at 1, 7, 30, and 90 dpi, and fragments of spleen and liver were collected for bacteriology, histopathology, and immunohistochemistry.

At all time points, significantly fewer CFUs (P < 0.001) were recovered from the spleens and livers of mice infected with the B. ovis ΔabcAB mutant than from spleens and livers of mice infected with the B. ovis WT strain (Fig. 6). At 30 dpi, the B. ovis ΔabcAB mutant was recovered from the spleen of only one mouse, and very few CFUs of this mutant strain were recovered from the liver after 7 dpi. Consistent with this finding, no significant histopathological lesions or B. ovis immunolabeling were observed in these tissues from mice infected with the ΔabcAB mutant (Table 4 and Fig. 7 A and B). These results indicate that the mutant lacking abcAB is attenuated in the spleen and liver as early as 1 dpi, and therefore, this predicted ABC transporter may be required for B. ovis infection and pathogenesis in the mouse model.

FIG. 6.

Infection of male BALB/c mice with B. ovis mutant strains. Mice were infected with 1 × 10⁶ CFUs of the B. ovis ΔabcAB, Δhmg, or WT strain by peritoneal injection. Fragments of spleen (A) and liver (B) were collected for counting CFUs at 1, 7, 30, and 90 dpi. Bars represent the means for 4 mice (mutant strains) or 8 mice (WT strain) at each time point, and standard deviations are shown. Raw data were logarithmically transformed prior to ANOVA, and means were compared by Tukey's test. Significant differences in the numbers of CFUs of the B. ovis WT strain and the mutant strains are indicated by asterisks (*, P < 0.001). For each mutant strain, significant differences (P < 0.05) between time points are indicated by different letters. A horizontal line indicates the detection limit for CFUs for each organ.

TABLE 4.

Average score for inflammatory lesions in spleen and liver of BALB/c male mice infected with Brucella ovis mutant strains

Strain	Score for spleen on indicated day postinfectiona				Score for liver on indicated day postinfectiona
	1	7	30	90	1	7	30	90
WT	0B	0.9B	0.71A	0.17B	0.16B	2.17A	1.44A	0.75A
Δhmg mutant	0B	0.87B	0.33AB	0B	0B	2.13A	1.33A	0.50AB
ΔabcAB mutant	0B	0B	0B	0B	0B	0B	0B	0B

^aDifferent uppercase letters (A and B) in the same column indicate significant differences (P < 0.05) among groups at each time point.

FIG. 7.

Histopathology and immunohistochemistry of male BALB/c mice infected with B. ovis mutant strains. (A and B) No histopathological lesions were observed at any time point in the spleen (A) or liver (B) of BALB/c mice infected with the B. ovis ΔabcAB mutant strain. Bars, 200 μm. (C) Moderate multifocal microgranulomas in the liver of a mouse infected with the B. ovis Δhmg mutant strain at 7 dpi. (D) Mild immunolabeling of intralesional B. ovis Δhmg strain in hepatic microgranuloma (arrows). Bars, 100 μm.

Conversely, mice infected with the Δhmg mutant (TMS2) showed wild-type levels of B. ovis infection at all time points. High numbers of CFUs were recovered from spleen and liver at 1 and 7 dpi, followed by a significant decrease in bacterial load in these tissues at 30 dpi (Fig. 6). Mice infected with the B. ovis Δhmg strain showed larger bacterial loads in the spleen during the course of infection than in the liver, as described for B. ovis-infected mice. Also, histopathological lesions in spleen and liver were characterized by mild to moderate multifocal microgranulomas (Fig. 7C), which were associated with high scores for lesions, mainly at 7 and 30 dpi (Table 4). Mild B. ovis immunolabeling was observed in the cytoplasms of macrophages associated with hepatic microgranulomas at 7 dpi (Fig. 7D). Taken together, these findings show that the B. ovis Δhmg strain has a pathogenic level similar to that of the B. ovis WT strain, suggesting that the putative hemagglutinin gene is not required for B. ovis pathogenesis during mouse infection.

A predicted ABC transporter is essential for intracellular survival of B. ovis and persistent infection with B. ovis in the mouse.

In order to study the role of the ABC transporter during early B. ovis infection in the mouse, BALB/c male mice were infected i.p. with the WT or ΔabcAB mutant strain, and the extracellular and intracellular survival of the bacteria in peritoneal macrophages was evaluated at 6, 12, and 24 hpi. Early systemic distribution of the B. ovis mutant strain was also evaluated by estimating the bacterial load in the spleen at the same time points (Fig. 8).

FIG. 8.

In vivo infection of peritoneal macrophages with B. ovis or an attenuated mutant strain in male BALB/c mice. (A and B) Extracellular (A) and intracellular (B) survival of the B. ovis WT (•) and ΔabcAB (□) strains was evaluated at 6, 12, and 24 h post-peritoneal infection. (C) Early colonization of spleen was also evaluated at 6, 12, and 24 hpi. Each data point represents the means for 4 mice at each time point, and standard deviations are shown. Raw data were logarithmically transformed prior to ANOVA, and means were compared by Tukey's test. Significant differences in the numbers of CFUs of the B. ovis WT strain and the mutant strain are indicated by asterisks (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

At an early time point in infection (6 hpi), the number of extracellular CFUs recovered from the peritoneal cavities of mice infected with the ΔabcAB mutant was similar to that recovered from B. ovis-infected mice. However, the number of extracellular CFUs of the ΔabcAB mutant quickly decreased during the course of infection, whereas the B. ovis WT strain was able to survive extracellularly for up to 24 hpi (Fig. 8A). In peritoneal macrophages, neither the numbers of intracellular CFUs of the ΔabcAB mutant nor those of the B. ovis WT strain were statistically significant at early time points. However, the number of intracellular CFUs of the B. ovis ΔabcAB mutant strain was significantly decreased at 12 hpi compared to that of the WT strain (P < 0.01), which demonstrates that this mutant strain does not survive in peritoneal macrophages. The B. ovis WT strain was able to survive and replicate inside peritoneal macrophages, with the numbers of intracellular CFUs increasing for up to 24 hpi (Fig. 8B). Additionally, mice infected with the B. ovis WT strain had high bacterial loads in the spleen at early stages of infection (Fig. 8C). Conversely, the numbers of CFUs recovered from the spleen during early infection with the ΔabcAB mutant were significantly lower than those for the WT strain at all time points (6 to 24 hpi). In close agreement with the bacteriology results, these data demonstrate that the B. ovis ΔabcAB strain is attenuated in mice. Overall, this predicted ABC transporter is required for extracellular and intracellular survival of B. ovis in peritoneal macrophages and, consequently, for establishing a persistent in vivo infection. Since the B. ovis ΔabcAB mutant had an impaired ability to survive extracellularly in vivo during the early stages of infection, the effects of complement and antimicrobial peptides (i.e., polymyxin B) on the survival of the WT and ΔabcAB strains were assessed. No significant differences were observed in the WT and ΔabcAB strains incubated with various concentrations of serum complement (Fig. 9 A), but the ΔabcAB mutant strain was slightly but significantly more sensitive to 1 mg/ml of polymyxin B than the WT B. ovis strain (Fig. 9B). The rough lipopolysaccharide (LPS) phenotype of the ΔabcAB mutant strain was confirmed by crystal violet staining (data not shown).

FIG. 9.

Serum complement and polymyxin B sensitivities of WT B. ovis and the attenuated ΔabcAB mutant strain. (A and B) The B. ovis ΔabcAB mutant or the WT strain was incubated for 1 h with various concentrations of serum complement (A) or polymyxin B (B). There were no statistically significant differences in the rates (%) of survival of these two strains with various concentrations of serum complement, but the ΔabcAB strain was significantly more susceptible to 1 mg/ml of polymyxin B than the WT strain (*, P < 0.001; the Student Newman-Keuls test). These data are the means and standard errors of two independent experiments performed in triplicate.

Intracellular survival of the B. ovis ΔabcAB and WT strains was evaluated during in vitro infections of the murine macrophage cell line RAW 264.7 (Fig. 10). RAW 264.7 macrophages were infected with the B. ovis ΔabcAB or WT strain at an MOI of 1:100, and the numbers of intracellular CFUs were determined at 0, 4, 24, and 48 hpi. At an early time point in infection (0 hpi), the level of infection of the ΔabcAB mutant was similar to that of the WT. However, the levels of infection of the B. ovis ΔabcAB mutant were significantly decreased at 24 (P < 0.01) and 48 (P < 0.001) hpi compared to those of the WT strain, which confirms that the mutant strain is not able to survive intracellularly during the course of infection. In contrast, the B. ovis WT strain was able to survive and replicate in RAW 264.7 murine macrophages, showing increasing numbers of CFUs (∼1 log) for up to 48 hpi. These results support the notion that the putative ABC transporter is required for the intracellular survival and replication of B. ovis in phagocytic cells, although this protein may not be essential for the internalization of B. ovis in murine macrophages.

FIG. 10.

In vitro infection of RAW 264.7 murine macrophages with B. ovis or an attenuated mutant strain. RAW 264.7 murine macrophages in a 96-well plate were infected with the B. ovis ΔabcAB mutant or the WT strain at an MOI of 1:100. Intracellular bacterial loads were recovered at 0, 4, 24, and 48 hpi. Time zero represents the number of intracellular CFUs after 1 h of incubation with gentamicin. Data points represent the means for triplicate experiments at each time point, and standard deviations are shown. Raw data were logarithmically transformed prior to ANOVA, and means were compared by Tukey's test. Significant differences in the numbers of CFUs of the B. ovis WT strain and the mutant strain are indicated by asterisks (*, P < 0.01; **, P < 0.001). Data are from an individual experiment that is representative of three independent experiments.

Infection of male IRF-1 KO mice determined the virulence of B. ovis mutant strains.

Previous studies demonstrated that IRF-1 KO mice have a defective immune response via gamma interferon (IFN-γ) signaling and as a result are highly susceptible to lethal Brucella spp. infections (25). These mice are currently used to verify the virulences of mutant strains of Brucella spp. (25, 26, 37). In order to investigate the virulences of B. ovis and the mutant strain in a lethal-infection model, IRF-1 KO male mice were infected with 2.3 × 10⁶ CFU/animal of the WT or ΔabcAB strain and monitored daily, as shown in Fig. 11.

FIG. 11.

Susceptibilities of male IRF-1 KO mice to infections with the B. ovis WT and mutant strains. Mice (n = 5) were infected with 2.3 × 10⁶ CFUs of the ΔabcAB mutant or WT strain and were monitored daily for survival until 21 dpi. The lethality rates for the two groups at the end of the experiment were compared using Fisher's exact test. IRF-1 KO mice infected with the B. ovis ΔabcAB mutant strain survived longer than those infected with the B. ovis WT strain (P = 0.0079).

Mice infected with the B. ovis WT strain started to die at 11 dpi, and this group reached 100% lethality at 14 dpi. These data show that male IRF-1 KO mice are not able to control B. ovis infection, as was previously described for IRF-1 KO mice infected with other pathogenic Brucella spp. Conversely, all mice infected with the ΔabcAB strain survived for up to 21 dpi, which was significantly different from the mouse group infected with the B. ovis WT strain (P = 0.0079). Thus, male IRF-1 KO mice may be used as a lethality model to study the virulence of B. ovis mutant strains. In this study, the attenuation of the B. ovis ΔabcAB strain in the mouse model was confirmed in immunocompromised mice.

DISCUSSION

Given that B. ovis mainly affects sexually mature rams, causing epididymitis and orchitis (4, 6), the development of animal models in male hosts is essential to expand our knowledge of B. ovis pathogenesis. In addition, the mouse has frequently been used as an infection model to study chronic infections with Brucella spp. (2, 13, 21, 34, 43). In this study, we developed and characterized a murine model for B. ovis infection for up to 90 dpi in male BALB/c and C57BL/6 mice. All mouse strains were infected i.p. with 10⁶ CFUs of the B. ovis WT strain, which established a systemic distribution and chronic infection in the mouse model. Our study is in agreement with a previous study by Jiménez de Bagüés and colleagues (24), who also demonstrated that doses of B. ovis higher than 10⁷ CFUs are not required to establish a chronic infection in male mice. Previous studies have described hepatic and splenic microgranulomas during systemic infections with pathogenic Brucella spp. in the mouse (13, 25, 34). In this study, these same histopathological changes were observed mainly at 7 and 30 dpi, i.e., in chronic stages of B. ovis infection. Overall, the inflammatory lesions associated with high numbers of CFUs in the spleen and liver characterized a suitable murine model for B. ovis infection.

Interestingly, the kinetics of infection and the histopathological lesions observed during infection were the same for both mouse strains. Although lower numbers of CFUs were recovered from the spleens of C57BL/6 mice at 7 and 30 dpi than from those of BALB/c mice, C57BL/6 mice were not able to clear the infection until 90 dpi. Previous reports have indicated that BALB/c mice are more susceptible to B. abortus infection than C57BL/6 mice, resulting in higher bacterial loads (∼1 log) during the course of infection (2, 32, 33).

In male mice, early B. ovis infection resulted in the colonization of the genital tract, but the numbers of CFUs decreased quickly up to 30 dpi. Additionally, histopathological lesions were restricted to tissues adjacent to the genital tract, which may be due to the intraperitoneal route of inoculation. Therefore, B. ovis did not show a genital tropism in the mouse. Previous studies have shown that pathogenic species of Brucella may colonize the genital tract of male mice (21, 37). Recently, Izadjoo and colleagues (21) reported that B. melitensis infection in BALB/c mice leads to perivascular inflammation of the testes. However, the histological findings were not correlated with intermittent colonization of this organ. Conversely, in natural hosts B. ovis has a clear tropism for the male genital tract (15), resulting in the colonization of the reproductive organs around 30 days postinfection (3). Chronic B. ovis infection in sexually mature rams causes epididymitis, orchitis, and vesiculitis, which are characterized by an interstitial neutrophilic and histiocytic infiltrate (4, 40). This study demonstrated that the mouse is not a good model for B. ovis-induced genital disease. However, the mouse may be used as an infection model for studies comparing the B. ovis WT strain and mutant strains.

Sequencing of the complete genome of B. ovis and the identification of BOPI-1 allowed for study of the roles of predicted proteins, including an ABC transporter and a hemagglutinin, during B. ovis infection in mice (48). ABC transporter systems from classical species of Brucella have been evaluated in the mouse model (11, 27, 39). The B. ovis ΔabcAB mutant was attenuated in the spleen and liver of BALB/c male mice as early as 1 dpi, which suggests that this protein has an important role in B. ovis pathogenesis and infection in the mouse model. A previous study (39) reported that a polysaccharide ABC transporter is required for B. abortus pathogenesis in the murine model and that an ABC transporter mutant strain can potentially be used as a vaccine candidate (16, 39). Conversely, ABC transporter proteins related to iron transport and toxin excretion were not essential for the intracellular survival of B. abortus and chronic B. abortus infection in mice (11, 27).

The B. ovis ΔabcAB mutant did not survive intracellularly in peritoneal macrophages, in RAW 264.7 murine macrophages, or extracellularly in the peritoneal cavity for 24 hpi. These results indicate that this ABC transporter system is required for the intracellular and extracellular survival of B. ovis during the early stages of infection. This phenotype may be partially explained by the increased susceptibility of the ΔabcAB mutant strain to antimicrobial peptides (polymyxin B) demonstrated in this study, particularly because rough Brucella spp. tend to be more resistant to polymyxin B than smooth Brucella spp. (31).

To gain insights into the degree of attenuation of the B. ovis ΔabcAB mutant, IRF-1 KO mice were used in this study to evaluate the virulence of the B. ovis WT strain and the attenuated ΔabcAB strain. IRF-1 is a transcriptional factor induced by IFN-γ, which is an important cytokine that relates to the efficient control of Brucella sp. infection in mice (2, 33). Interestingly, all male IRF-1 KO mice infected with the B. ovis WT strain died within 14 dpi, but the B. ovis ΔabcAB mutant strain was not lethal for IRF-1 KO mice, confirming the strong attenuation phenotype observed in WT mice. Infection of IRF-1 KO mice with the virulent phenotype of WT B. ovis was similar to those of B. melitensis and B. abortus, which led to fatal infection by 15 dpi, whereas attenuated mutant strains, including B. abortus RB51, are not lethal for these immunocompromised mice (25, 26).

Recent genomic analyses have shown that B. ovis encodes fewer ABC transporter proteins than other classical species of Brucella, due to high numbers of pseudogenes in genomic regions that encode predicted ABC systems (22, 48). As a result, B. ovis apparently lacks the ability to import several nutrients, including polyamines, thiamine, erythritol, and xylose (22), via ABC systems, which may determine the low pathogenicity of this Brucella species during animal and human infections. Using the recently sequenced B. ovis genome (48) and the classification of ABC systems in classical Brucella spp. (22), we determined that the specific ABC transporter evaluated in this study is likely related to a dipeptide import in B. ovis. Intriguingly, in B. ovis, pseudogenes and deleted genes were described in regions of the genome that in Brucella spp. potentially encode other peptide ABC transporters. As there is no alternative pathway for peptide import in B. ovis, the deletion of a single transporter protein may certainly result in a lack of full virulence during animal infection. Thus, these data elucidate the pathogenic role of a B. ovis-specific ABC transporter.

Previous studies support the notion that BOPI-1 is preserved in marine isolates of Brucella spp. from bottlenose dolphins (48). We decided to determine whether the two ORFs deleted from the B. ovis ΔabcAB mutant are preserved in any sequenced Brucella species available at the Broad Institute data bank (http://www.broadinstitute.org/). Both sequences are conserved (∼99% identity) in three different marine-isolate strains, Brucella pinnipedialis M292, B. pinnipedialis B2/94, and Brucella sp. F5/99. B. pinnipedialis M292 and B2/94 are marine strains of a Brucella species originally isolated from seals, and Brucella sp. F5/99 was isolated from aborted fetuses of bottlenose dolphins. It is noteworthy that although Brucella sp. F5/99 was isolated from a dolphin, it is currently not classified as a Brucella ceti strain. It is an isolate from the Pacific, and it has a genotype distinct from those of cetacean and pinniped isolates from the Atlantic classified as B. ceti and B. pinnipedialis, respectively. Interestingly, other marine strains of Brucella isolated from dolphins and seals were also classified as B. pinnipedialis and B. ceti, respectively, although they did not conserve the sequences that encode the pathogenic ABC transporter protein. These findings are in close agreement with those of previous studies that demonstrated distinct genotypes of B. ceti and B. pinnipedialis isolated from different host species (30). Moreover, the B. ovis pathogenic transporter protein is likely involved in the pathogenesis of other Brucella species that have zoonotic potential, including Brucella sp. F5/99 (49). For that reason, the specific role of this peptide ABC transporter in marine isolates of Brucella should be further evaluated.

In conclusion, the results reported in this study demonstrate that the predicted peptide ABC transporter is required for B. ovis pathogenesis in the mouse and for intracellular survival. In addition, male BALB/c and C57BL/6 mice were shown to be suitable models for B. ovis infection, and male IRF-1 KO mice are an efficient model for evaluation of the virulences of B. ovis mutant strains. These murine models may enhance the identification of other pathogenic proteins in B. ovis, including those that are potentially encoded by BOPI-1.