The Predicted ABC Transporter AbcEDCBA Is Required for Type IV Secretion System Expression and Lysosomal Evasion by Brucella ovis

Brucella ovis is a major cause of reproductive failure in rams and it is one of the few well-described Brucella species that is not zoonotic. Previous work showed that a B. ovis mutant lacking a species-specific ABC transporter (ΔabcBA) was attenuated in mice and was unable to survive in macrophages. The aim of this study was to evaluate the role of this ABC transporter during intracellular survival of B. ovis. In HeLa cells, B. ovis WT was able to survive and replicate at later time point (48 hpi), whereas an ΔabcBA mutant was attenuated at 24 hpi. The reduced survival of the ΔabcBA mutant was associated with a decreased ability to exclude the lysosomal marker LAMP1 from its vacuolar membrane, suggesting a failure to establish a replicative niche. The ΔabcBA mutant showed a reduced abundance of the Type IV secretion system (T4SS) proteins VirB8 and VirB11 in both rich and acid media, when compared to WT B. ovis. However, mRNA levels of virB1, virB8, hutC, and vjbR were similar in both strains. These results support the notion that the ABC transporter encoded by abcEDCBA or its transported substrate acts at a post-transcriptional level to promote the optimal expression of the B. ovis T4SS within infected host cells.


Introduction
Brucella ovis is one of the main causes of reproductive failure in sheep [1]. In sexually mature rams, the infection causes chronic epididymitis, orchitis, and infertility, whereas in ewes, it is characterized by uncommon abortion and stillbirth [2,3]. B. ovis has a worldwide distribution in main sheep-raising areas, resulting in significant economic losses for the sheep industry [1,4]. This organism is a stably rough Gram-negative coccobacillus that belongs to the alpha-2-Proteobacteria family [2,5]. Unlike most of the well-described Brucella spp., B. ovis does not cause disease in humans [2].
Genomic analyses of B. ovis resulted in the identification a pathogenicity island (BOPI-1) in chromosome II containing 28 open reading frames (ORFs), which are absent in other classical Brucella species [18]. This island comprises genes that potentially encode pathogenesis-associated proteins, including an ATP-binding cassette (ABC) transporter (BOV_A0504-BOV_A0500, designated abcEDCBA) [18,19]. ABC transporters are responsible for nutrient uptake and the export of toxins and antibiotics, and they may play an important role in gene expression [20,21]. In Brucella spp., a polysaccharide ABC transporter is required for B. abortus pathogenesis in the murine model [11], whereas ABC transporter proteins related to iron transport and toxin excretion were not essential for B. abortus chronic infection in mice [21,22].
In B. ovis, a species-specific ABC transporter located at the BOPI-1 was essential for survival and replication in a mouse model and in macrophages [19]. However, it is not known what the specific role of this transporter is and whether it affects other virulence factors necessary for B. ovis survival in host cells. B. ovis is the classical Brucella species with lowest number of ABC transporters predicted to be functional, due to high numbers of pseudogenes in conserved Brucella spp. regions predicted to encode ABC systems [18,23]. This may be one of the determinants of the low pathogenicity of B. ovis during animal and human infections. Hence, studying specific features of B. ovis may explain why it is not virulent in humans [18]. Moreover, high numbers of pseudogenes in ABC systems may allow evaluation of the pathogenic role of conserved transporters in B. ovis by one single gene deletion. This is less feasible in classical Brucella species, like B. melitensis and B. abortus, due to the presence of redundant transporters, which may compensate the function of a deleted protein.
The goal of this study was to evaluate the role of a specific ABC transporter during B. ovis in vitro growth, intracellular survival, and trafficking. Our results show here that the specific locus abcEDCBA, encoding a putative peptide importer, promotes intracellular survival by affecting T4SS protein expression at a post-transcriptional level and, consequently, contributing to B. ovis evasion of phagosome/lysosome fusion.

Bacterial strains, media and culture condition
Bacterial strains used in this study were the virulent strain B. ovis ATCC 25840 (WT); DabcBA mutant strain (TMS2) lacking a putative ABC transporter [19]; B. ovis WT and DabcBA isogenic strains expressing mCherry fluorescence (named TMS8 and TMS9, respectively), with the insertion of pKSoriT-bla-kan-PsojA-mCherry plasmid [24] (Table 1). All inocula were cultured on Trypticase Soy Agar (TSA, BD) plates with 5% sheep blood for three days at 37˚C in 5% CO 2 , as previously described [25]. For proteomic analysis, B. ovis WT and DabcBA were grown in triplicate on TSA plates with 10% hemoglobin for three days. Kanamycin (Kan, 100 mg/mL) and Ampicillin (Amp, 200 mg/mL) were added to media when necessary. For strains TMS8 and TMS9, selected colonies were Amp resistant and fluorescent, as previously described [24].
Considering that B. ovis does not grow adequately in conventional liquid media [26], a rich Trypticase Soy Broth (TSB, BD) was supplemented with 10% of FBS (Gibco). Strains were cultured overnight at 37˚C on rotary shaker. Additionally, B. ovis in vitro growth was measured in TSB media supplemented with different concentrations of FBS (0, 2, 5, or 10%), nickel (NiSO 4 at 0.5, 1 or 2 mM) or after chelation of divalent cations by adding EDTA (10, 25, or 50 mM

Cloning and complementation
To express B. ovis ABC transporter locus in S. Typhimurium TT17573 (oppBC tppB dpp) strain, the entire abcEDCBA locus (5.6 Kb) was amplified by PCR, using genomic DNA from B. ovis ATCC 25840. A histidine tag (6x-his) and a stop codon were engineered into the C-terminal region ( Table 2). PCR reactions were prepared with 23 mL of Supermix High Fidelity (Invitrogen), 0.7 mM of each primer abc-his (Table 2) and 2 mL of the genomic DNA. Cycling parameters, as described by [27], were as follows: denaturation at 94˚C for 3 min; 35 cycles of denaturation at 94˚C for 1 min, annealing at 50˚C for 1 min and extension at 60˚C for 6 min; and final extension at 60˚C for 10 min. The 5.6 Kb product was purified from agarose gel using QIAEXII kit (Qiagen) and, then, inserted into the cloning vector pCR2.1 TOPO, following the manufacturer's instructions (Invitrogen). The insert was excised by double digestion with SpeI and KpnI, and cloned into expression vector pBBR1-MCS4, Amp R (4 Kb). To confirm the sequence and orientation of the insert, the constructed plasmid, named pTSabc, was sequenced using primers M13 (Invitrogen) ( Table 2). Then, the plasmid was introduced into S. Typhimurium TT17573 (oppBC tppB dpp) by electroporation, with previous heating of the bacteria for 30 min at 50˚C [28]. Colonies resistant to Tet, Kan, and Amp were selected and the ABC transporter expression in TMS14 (TT17573:abcEDCBA) was confirmed by Western blot, using anti-Histidine tag mouse monoclonal antibody conjugated with HRP (Lifetech).
The deletion of oppBC and transposon insertion into dppA and tppB in S. Typhimurium TT17573 were confirmed by PCR as described, using pairs of primers shown in Table 2.

S. Typhimurium lethality assay
Considering that STm has three types of peptide ABC transporters, which mainly transport dipeptides (Dpp), tripeptides (Tpp) or oligopeptides (Opp) [29][30][31], we attempted to use this organism to predict the function of B. ovis transporter by evaluating bacterial resistance to toxic peptides [29,32]. For the lethality assay, STm WT, STm TT17573 (oppBC tppB dpp), and TMS14 (TT17573:abcEDCBA) mutant expressing the B. ovis abcE-A transporter were grown overnight in M9 liquid media. Each bacterial strain was adjusted to 3610 5 CFU/mL in fresh M9 with 0.7% noble agar (BD) and layered over M9 agar plates containing antibiotics. After solidifying, 7 mm-filter paper disks containing 0.5 mg and 1 mg of alafosfalin (L-Alanyl-L-1-aminoethylphosphonic acid, Sigma-Aldrich) or 0.2 mg and 0.4 mg of trilysine (Sigma-Aldrich) were placed onto the plate. After drying the disks, plates were incubated for 16 h at 37˚C. The ability of a toxic peptide to inhibit bacterial growth was quantified by determining the diameter (mm) of the inhibitory zone surrounding a filter paper disk with the toxic peptides. Assays were performed three times independently, with triplicate samples.

Proteomic analysis by Differential Gel Electrophoresis (DIGE)
Protein expression of WT and DabcBA B. ovis were compared by DIGE during in vitro growth in rich media. For each bacterial strain, triplicates grown independently were used. Protein was extracted with 2 vol of lysis buffer (8 M urea, 2 M thiourea, 4% w/v CHAPS, 40 mM Tris1M, mix of protease inhibitors) (GE Healthcare), followed by 3 h of agitation and cellular lysis by passing through a 26G needle. Lysates were centrifuged at 20,0006g for 30 min at room temperature, and the supernatants were recovered, quantified by 2D Quant Kit (GE Healthcare), and kept at 280˚C. A pool of protein extracts obtained from triplicate samples was used.
To identify differentially expressed proteins between B. ovis WT and DabcB mutant, the protein mixture (50 mg) of each strain was labeled with 400 pmol of either Cy3 or Cy5 dye (GE Healthcare), according to the manufacturer's instructions. A protein mixture of both strains (50 mg) was labeled with Cy2 dye as internal control. These reactions were carried out on ice for 30 min in the dark and quenched with 1 mL of lysine (10 mM) for 10 min on ice. Four DIGE gels were done and a dye-swap was performed. All labeled proteins (150 mg) were added to 3.4 mL of immobilized pH gradient (IPG) buffer (10 mL/mL), 450 mg of unlabeled protein, and IEF buffer (8 M urea, 2 M thiourea, 4% CHAPS, 0.0025% bromophenol blue, 10 mg/mL dithiothreitol) in a total volume of 340 mL per IPG strip (18 cm, pH 4-7, GE Healthcare). The samples were incubated overnight with the IPG strips, submitted to isoelectric focusing using Ettan IPGphor system (GE Healthcare), followed by incubation in equilibrium solution (50 mM Tris-HCl, pH 8.8, 6 M urea, 30% glycerol, 2% SDS, 0.002% bromophenol blue, and 125 mM DTT) for 15 min and an additional incubation in a new solution containing 13.5 mM iodoacetamide instead of DTT.
Electrophoresis was performed in 12% SDS-PAGE using an Ettan Electrophoresis unit (GE Healthcare) at 10 mA/gel for 1 h, followed by 45 mA/gel until the dye front reached the bottom of the gel. Each gel was scanned using Typhoon FLA 9000 (GE Healthcare) with excitation/emission wavelengths of 488/ 520, 532/580, and 633/670 nm for Cy2, Cy3, and Cy5 dyes, respectively. Gel images were analyzed using DeCyder 2D software, Version 7.0 (GE Healthcare). Spots with p-value ,0.05 and average of volume ratio over 1.5 were selected for mass spectrometry (MS) identification. To extract the spots of interest, all DIGE gels were subsequently stained with colloidal Coomassie Brilliant blue G-250, as previously described [33].

Mass spectrometry (MS) and prediction of protein interactions
Differentially expressed spots between WT and DabcBA B. ovis were excised from gel, treated with trypsin, and desalted using Zip-Tips (Millipore Corporation), as described elsewhere [34]. Each sample was mixed with 0.5 vol of saturated matrix solution (10 mg/mL a-cyano-4-hydroxycinnamic acid in 50% acetonitrile/0.1% trifluoroacetic acid). Then, samples were spotted on MTP AnchorChip 600/384 (Bruker Daltonics) and let it dry at room temperature. For protein identification, raw data were acquired on a MALDI-TOF/TOF AutoFlex III instrument (Bruker Daltonics) in the positive/reflector mode controlled by FlexControl software. Instrument calibration was previously done using peptide calibration standard II (Bruker Daltonics) as a reference.
Data from mass spectrometry were aligned against all non-redundant protein sequence database from NCBI (http://www.ncbi.nlm.nih.gov) using the MASCOT software MS/MS ion search tool (http://www.matrixscience.com). The parameters in this search were as follows: no restriction on protein molecular weight, loss of a trypsin cleavage site, variable modifications of methionine (oxidation), cysteine (carbamidomethylation), and pyroglutamate formation at the N-terminal glutamine. The mass tolerance for searched peptides were 0.8 Da for MS spectra and 0.6 Da for MS/MS spectra. Peptides were identified when the scoring value exceeded the identity or extensive homology threshold value calculated by the MASCOT (p,0.05).
Interactions of identified proteins were predicted by using the STRING software (http://string-db.org), as previously described [35]. The evidence mode was set up at an average confidence level 0.4 and the search parameters included: neighborhood, gene fusion, co-occurrence, co-expression, experiments, databases, and text mining.

HeLa cell culture and infection
HeLa cells were cultured in a 75 cm 2 flasks with Dulbecco's modified Eagle medium (DMEM, Gibco) supplemented with 10% FBS, at 37˚C with 5% of CO 2 . When reaching 80-90% of confluence, cells were treated with Trypsin-EDTA 0.05% for 10 min, and seeded overnight in a 24-well plate at a density of 1610 5 cells per well. The next day, HeLa cells were infected as previously described [36], with some modifications for B. ovis. Briefly, B. ovis WT, DabcBA and mCherryexpressing isogenic strains (TMS8 and TMS9) were grown for 3 days, resuspended in DMEM supplemented media, and added 0.5 mL to each well, to infect HeLa cells with the multiplicity of infection (MOI) of 1000. The plates were centrifuged at 400 6 g for 5 min at room temperature and incubated for 30 min at 37˚C in 5% CO 2 . The cells were washed three times with Dulbecco phosphate-buffered saline (DPBS) to remove free bacteria, followed by the addition of 0.5 mL of DMEM supplemented media with 50 mg/mL of gentamicin into the wells, to kill the extracellular bacteria. This was considered the zero time point.
In order to determine bacterial survival, the medium was aspirated at 0, 8, 24, and 48 h after infection and HeLa cells were lysed with 0.5 ml of 0.5% Tween 20, followed by rising each well with 0.5 mL of PBS. Viable bacteria were quantified by serial 10-fold dilutions in sterile PBS and plating on TSA with 5% sheep blood. This experiment was performed in triplicate and repeated three times.

Brucella ovis confocal microscopy
HeLa cells were seeded overnight on a 12-mm glass coverslips in a 24-well plate at a density of 3610 4 cells per well. The cells were infected with B. ovis mCherryexpressing WT or DabcBA strain with MOI of 1000. At 8, 24, and 48 h post infection, each coverslip was washed three times with PBS, fixed with 3% paraformaldehyde (pH 7.4) at 37˚C for 20 min, followed by three washes with PBS and incubation in 50 mm NH 4 Cl in PBS for at least 10 min, to quench free aldehyde groups. Samples were permeabilized with 10% horse serum and 0.1% saponin in PBS for 30 min at room temperature. After removing the coverslips, each sample was labeled with rabbit anti-human LAMP-1 antibody (Thermo Scientific), by inverting the coverslips onto droplets of primary antibody diluted (1:1000) in 10% horse serum and 0.1% saponin solution in PBS. After incubating for 1 h at room temperature, each sample was washed with PBS and labeled with diluted (1:1000) Alexa Fluor 488 anti-rabbit antibody (Lifetech) for 1 h at room temperature. Then, cells were washed twice with 0.1% saponin in PBS, once in PBS, once in distilled H 2 O and mounted in Mowiol 4-88 mounting medium (Calbiochem). Samples were analyzed on a Carl Zeiss LSM 510 confocal laser scanning microscope for image acquisition. Confocal images of 102461024 pixels were acquired as projections of three consecutive slices with a 0.38-mm step and gathered using Adobe Photoshop CS5. To quantify B. ovis infection in cells and LAMP-1 + compartment colocalization, at least 100 bacteria and 100 cells per sample were counted. All experiments were performed independently three times and in triplicate.

Expression of T4SS proteins by Western blot
The expression of VirB proteins, which constitute the T4SS, were evaluated during in vitro growth of B. ovis WT and DabcBA strains. Each sample was cultured on TSA plates with 5% of sheep blood for three days and subsequently transferred to TSB with 10% FBS, at a starting OD 600 of 0.1. After incubating overnight on a rotary shaker at 37˚C, cells from 1 mL of each culture were pelleted by centrifugation and resuspended in SDS-PAGE buffer. VirB expression was also analyzed after growth of B. ovis on TSA plates containing 5% of sheep blood, by scraping an aliquot straight from the plate and resuspending in SDS buffer.
Considering that T4SS expression in classical Brucella species is induced by acidic pH and nutrient-poor conditions [37], expression of VirB proteins under these conditions was assayed. To this end, 10 9 CFU of each strain was suspended in modified minimal E medium pH 5.0, and incubated for additional 6 h at 37˚C on a rotary shaker, as previously described [38]. VirB protein expression is also promoted by urocanic acid, which induces expression of two T4SS regulators (HutC and VjbR) in an acidic environment [39]. Therefore, B. ovis WT and DabcBA (10 9 CFU) were suspended in modified minimal E medium (pH 5.0) supplemented with urocanic acid 5 mM or glutamic acid 5 mM (control), and incubated for additional 4 h at 37˚C on a rotary shaker. Assays were performed at least three times independently for each strain.
For Western blot, bacterial proteins were extracted by heating the samples for 10 min in 4% SDS buffer. The total protein (10 to 20 mg) was electrophoresed on a 12% SDS-PAGE gel, and transferred to a nitrocellulose membrane. Membranes were blocked in blocking solution (PBS containing 2% non-fat skim milk powder and 0.05% Tween 20) for 1 h and incubated for 1 h with rabbit anti-VirB8, anti-VirB9 or anti-VirB11 polyclonal antibody diluted in blocking solution (1:5000). Then, membranes were washed three times with blocking solution and incubated for another 1 h with diluted (1:5000) goat anti-rabbit IgG antibody (Biorad) conjugated with horseradish peroxidase (HRP). HRP activity was detected with a chemiluminescent substrate (Perkin-Elmer). As a loading control, same sample concentrations used in Western blot were loaded on a separate SDS-PAGE gel and stained with Coomassie Brilliant Blue.

Bacterial RNA extraction and Real Time RT-PCR
To compare gene expression of B. ovis WT and DabcBA strains during in vitro growth, RNA was extracted from 1 mL of bacterial samples grown on TSB media with 10% FBS for 24 h at 37˚C on rotary shaker. RNA extraction was carried out using TRI reagent (Molecular Research Center, Cincinnati) as previously described [13], followed by RNA purification with RNeasy Minelute cleanup kit (QIAGEN) and DNase treatment (Invitrogen) for 1 h at 37˚C. Real-time PCR was performed using TaqMan reverse transcription reagent (Applied Biosystems) with 10 mL of RNA from each sample in a 60((L volume. To assess whether there was genomic DNA contamination in samples, a new 30((L mix volume of TaqMan reagent was performed with 5((L of RNA, without adding reverse transcriptase enzyme. Four (L of cDNA was used as the template for each reverse transcription-PCR (RT-PCR) in a 25 mL volume, with 12.5 mL of SYBR Green (Applied Biosystems) and 0.3 mM of each primer listed in Table 3. Data were analyzed using the comparative Ct method (Applied Biosystems). Transcript levels of virB1, virB8, vjbR, hutC, abcA, and abcC were normalized with mRNA levels of the housekeeping 16S ribosomal gene (Table 3). Ct values of B. ovis WT genes were expressed in relation to B. ovis DabcBA strain.

Statistical analyses
All CFU data were logarithmically transformed and submitted to analysis of variance (ANOVA). For confocal microscopy, all percentage data were submitted to angular transformation prior ANOVA. Means of groups were compared with Tukey's test (GraphPad InStat 3) and considered significant when p,0.05. Confocal microscopy and real time data represent geometric mean and standard error of three independent experiments. For real time PCR, Ct values were compared between groups by Student T test, and considered significant when p,0.05.

Prediction of ABC transporter function during in vitro growth
According to genomic analysis of B. ovis [18,23], the B. ovis abcEDCBA locus was previously predicted to encode a peptide importer [19]. Additional analysis was performed in this study, by aligning the nucleotide sequence to sequences available at the NCBI protein database (BLASTx).
Two proteins encoded by abcA (BOV_A0500) and abcB (BOV_A0501), which were deleted in the abcBA mutant strain, are predicted to be ATPases of ABC systems, with conserved Walker A and B motifs. Therefore, deletion of abcA and abcB would lead to inactivation of the transporter. Both ATPases showed 99% identity only with Brucella pinnipedialis B2/94 and Brucella sp. 63/311, and 91% identity with two phylogenetically related bacteria (Ochrobactrum anthropi and Ochrobactrum intermedium). Proteins encoded by abcE-C (BOV_A0504-502) were identical to a conserved group of ABC systems functioning in uptake of dipeptides, oligopeptides and nickel (Dpp/Opp/Nik). Both abcD and abcC are predicted to be transmembrane proteins, whereas abcE encodes a predicted substrate-binding protein, suggesting a function of the ABC system in substrate uptake.
The role of the B. ovis abcEDCBA transporter was evaluated during in vitro growth, using a liquid medium that allowed for exponential growth of both WT and DabcBA strains. Both strains showed limited growth after 24 h in TSB media ( Figure S1A), which is a standard laboratory media for Brucella spp. [40]. When TSB was supplemented with different concentrations of FBS (2, 5, or 10%), WT and DabcBA showed equally proportional growth, reaching maximal growth with 10% FBS ( Figure S1A). For further B. ovis experiments, TSB with 10% FBS was used as the standard liquid media.
Considering the genomic prediction of B. ovis abcEDCBA-encoded proteins as a nickel transporter (Nik), it was evaluated if in vitro growth was impacted by adding or removing nickel from the growth medium. Addition of NiSO 4 (0.5, 1, or 2 mM) to TSB with 10% FBS did not increase growth of either B. ovis DabcBA mutant, or the WT strain after 48 h ( Figure S1B). Also, chelation of divalent cations, including nickel, by adding EDTA at 10, 25 or 50 mM into the media equally limited the growth of both strains. These results show that deletion of the abcEDCBA-encoded ABC transporter did not restrict nickel uptake or in vitro growth of DabcBA B. ovis. To assess the predicted function of B. ovis ABC transporter as a peptide importer, this protein was constitutively expressed in S. Tm TT17573, which carries spontaneous mutations in the genes oppBC tppB dpp, resulting in non-functional peptide transporters for oligopeptides (Opp), tripeptides (Tpp) and dipeptides (Dpp). Previous studies have characterized peptides that are toxic for STm when taken up by a specific type of transporter [29,32]. Due to easy growth of STm in protein-restricted media, sensitivity of STm WT, STm TT17573 (oppBC tppB dpp), and TT17573 strain complemented with abcEDCBA (TMS14) were evaluated against trilysine and alafosfalin, which are imported by Tpp and Opp transporters, respectively ( Figure S2). For detection of abcEDCBA expression in S. Typhimurium, a 6x-His tag was engineered into the C-terminus of AbcA. STm WT and TT17573 (oppBC tppB dpp) demonstrated similar growth and metabolic activity by tetrazolium reduction in minimal M9 media ( Figure S2). STm WT was susceptible to trilysine and alasfosfalin at two different concentrations ( Figure S3A-B). However, STm TT17573 (oppBC tppB dpp) and TMS14 (TT17573:abcEDCBA) were resistant to toxic peptides ( Figure S3A), suggesting that these strains have non-functional Tpp and Opp transporters. Introduction of B. ovis abcEDCBA into STm TT17573 did not confer peptide uptake, although expression of abcA was confirmed by Western blot ( Figure S3C). These results suggested that B. ovis abcEDCBA does not function in STm as an oligopeptide or tripeptide importer.

Inactivation of Brucella ovis abcEDCBA affects the abundance of metabolic and virulence-associated proteins during in vitro growth
To gain insight into the role of abcEDCBA in the biology of B. ovis, differential expression of proteins between WT and DabcBA B. ovis strains was evaluated by 2D-DIGE. A representative image of the protein profile for each strain is shown in Figure 1. By DeCyder 2D image analysis software (GE Healthcare). Considering a volume ratio higher than 1.5; 100 spots had differential expression between the strains ( Figure S4), whereas 78 spots were visualized in the gel and extracted for mass spectrometry (MS) identification ( Figure 1A). Among these, 55 spots were successfully identified by MS/MS, whereas 40 spots (72,7%) had lower expression (Tables 4 and S1) and 15 spots had higher expression (Tables 5 and S2) in DabcBA B. ovis. Tables S1 and S2 show MS data for each spot, including peptide sequence, score, percentage of coverage, predicted and experimental values of isoelectric point (pI) and molecular weight (MW).
A total of 22 proteins had lower expression in B. ovis DabcBA, which included the following functional groups: outer membrane protein (Omp31); predicted amino acid and sugar ABC transporter binding proteins (ribose, glycerol, xylose, oligopeptide, and glycine binding proteins); protein folding (chaperonin GroES, acid stress chaperone HdeA); stress proteins (DNA starvation protein HdeA, superoxide dismutase Cu/Zn and Fe-Mn); metabolic enzymes (nucleoside diphosphate kinase); and protein and vitamin biosynthesis (Table 4).
Four proteins, which corresponded to nine spots, were not identified in B. ovis protein database from NCBI. However, the peptide sequences were identified as periplasmic proteins of sugar and aminoacid ABC transporters in other classical Brucella spp. species. Therefore, for these spots, the B. melitensis 16M database was used to predict protein function and interaction. Interestingly, by aligning the nucleotide sequence of protein-encoded genes in B. melitensis with B. ovis database (BLASTn), all four genes were identified as pseudogenes in B. ovis. These findings reveal the expression of four distinct ABC systems in B. ovis during in vitro growth, which were previously annotated incorrectly as pseudogenes.  Nine proteins had higher expression levels in DabcBA B. ovis, including: ABC transporters (nickel NikA and amino acid binding proteins); sugar metabolism (succinyl-CoA synthetase and malate dehydrogenase); protein metabolism (zinc protease); and transcriptional regulation (dehydrogenase quinone) ( Table 5).
Additionally, to detect virulence and metabolic differences between WT and DabcBA strains, interaction network predictions were performed among identified proteins with lower and higher expression in DabcBA B. ovis (Figures S5 and S6). The numbers correspond to specific spots as indicated in Figure 5.  The numbers correspond to specific spots as indicated in Figure 5.

B. ovis DabcBA is not able to survive intracellularly in HeLa cells
A previous study showed that B. ovis DabcBA did not survive intracellularly in murine macrophages and was attenuated early during infection in a mouse model [19]. However, we were interested in determining whether the inability of the DabcBA mutant to survive intracellularly reflected increased susceptibility to macrophage-specific bactericidal effects [36,41] Moreover, the kinetics of infection of B. ovis WT and DabcBA isogenic strains constitutively expressing mCherry (named TMS8 and TMS9) were evaluated in HeLa cells, to confirm that the fluorescent protein expression did not interfere with their phenotype. Both WT-mCherry and DabcBA-mCherry strains exhibited identical infection as shown in Figure 2, which allowed us to study B. ovis intracellular trafficking by confocal microscopy.   HeLa cells infected with WT, of which approximately 80% were able to exclude LAMP-1 from their BCV ( Figure 3A-C). Conversely, significantly lower numbers of DabcBA-mCherry were seen in HeLa cells at 48 mhpi and more than 90% of the bacteria colocalized with LAMP-1 ( Figure 3B-C). These results reveal that WT B. ovis is able to avoid phagosome/lysosome fusion and to replicate, whereas the DabcBA mutant remained within lysosomes, explaining its inability to survive in a human epithelial cell line. Therefore, the ABC transporter is necessary for B. ovis intracellular survival and replication at later stages of infection, potentially by promoting exclusion of lysosomal markers.

Lack of the Brucella ovis abcBACDE-encoded transporter reduces the levels of VirB proteins
The T4SS is one of the main virulence mechanisms for Brucella spp., and it is required for intracellular survival and replication by promoting exclusion of lysosomal proteins from the BCV [17,41]. Taking into account that DabcBA mutant lost the ability to exclude LAMP1, we analyzed expression of T4SS components by B. ovis WT and DabcBA during in vitro growth in rich and nutrient-limited media, using Western blotting (Figure 4).
Both strains were grown on rich liquid media (TSB with 10% FBS) and, then, transferred to modified minimal E media (MM) at pH 5.0, which promotes in vitro expression of the T4SS in Brucella spp. [37,42]. Interestingly, unlike what was described for other Brucella spp., WT B. ovis expressed VirB8 (26 kDa) and VirB11 (40 kDa) in both neutral and acidic media. Conversely, the DabcBA mutant had weak expression of VirB proteins, including in the minimal acid media, when compared to WT B. ovis ( Figure 4A). Expression of VirB8 and VirB9 (32 kDa) were also evaluated after standard 3-day growth on TSA plate with 5% sheep blood; however, only WT B. ovis was able to express VirB proteins in this condition ( Figure 4B). Since virB expression in Brucella spp. has been shown to be induced poor nutritional environment with low pH [38,39,43,44], we evaluated virB expression by B. ovis and B. abortus cultured in vitro in rich neutral medium. Indeed, B. ovis expressed both VirB8 and VirB11 under these conditions, whereas B. abortus did not express these two proteins encoded by the virB operon ( Figures  S7).
Previous studies have shown that adding urocanic acid into minimal acid media induces in vitro transcription of T4SS in Brucella sp., by increasing the expression of two virB-regulatory proteins, HutC and VjbR [39,43]. To determine whether expression of virB genes was affected upstream of HutC and VjbR, VirB8 expression was analyzed after transferring WT and DabcBA B. ovis into MM media pH 5.0 supplemented with 5 mM of urocanic acid or 5 mM of glutamic acid (control). As illustrated in Figure 4C, even in the presence of urocanic acid, DabcBA B. ovis was not able to express VirB8, whereas WT B. ovis maintained the expression of T4SS proteins. For consistent Western blot analyses, protein concentrations were confirmed by staining SDS-PAGE gel of the sample lysate with Comassie brilliant blue ( Figure 4A-C).
These findings showed that the expression of T4SS-encoded proteins is independent of low pH in B. ovis, but is dependent on the putative ABC transporter. In close agreement with the bacteriology and confocal microscopy results, these data suggested that DabcBA B. ovis did not survive intracellularly due to a lack of T4SS expression.

The AbcEDCBA transporter regulates the Brucella ovis T4SS at a post transcriptional level
To determine if weak expression of T4SS-encoded proteins in B. ovis DabcBA was due to an effect on virB transcription, mRNA levels of virB1, virB8 and two T4SS regulators (hutC and vjbR) were measured in B. ovis. Both WT and mutant strains were grown overnight in rich media (TSB with 10% FBS), under the same conditions as used for Western blot analysis, and Ct values of WT were compared to those obtained for the DabcBA mutant. Interestingly, real time RT-PCR showed that WT and DabcBA B. ovis strains had similar abundance of transcripts for virB genes as well as hutC and vjbR ( Figure 5). This result suggests that the predicted ABC transporter does not have an impact on T4SS expression at the level of transcription or mRNA stability, but may rather act at a post-transcriptional level.
As controls, mRNA levels of a deleted gene (abcA) and a conserved gene upstream of the deleted region (abcC) were also measured. As shown in Figure 5, no significant difference on abcC transcription was observed between the strains, inducating that deletion of DabcBA did not affect the transcription of the abcEDCBA operon. As expected, WT B. ovis showed significantly higher Ct values of abcA when compared to DabcBA, which confirms the effective deletion of the transporter gene in the mutant strain.

Discussion
Considering previous data [18,23] and the genomic analysis, B. ovis ABC transporter AbcEDCBA encoded by BOPI-1 was predicted to be a peptide importer of the Opp/Tpp/Nik family. Initially, nickel uptake was evaluated during in vitro growth of both WT and DabcBA B. ovis strains. As shown in Figures S1, although B. ovis DabcBA showed limited growth in different nickel conditions, it was similar to WT B. ovis growth, suggesting that nickel was taken up equally by both strains. In Brucella spp., three main nickel importers are demonstrated: an ABC transporter nikA-E and two Energy Coupling Factor (ECF-type) transporters, nikK-O and ure2 [45,46]. Previous genomic study revealed that two genes from ure2 (BOV_1316 and BOV_1319) and nikD (BOV_A0751) are pseudogenes in B. ovis [18], which may compromise the function of two nickel importers. However, ECF nikK-O is conserved and potentially may compensate nickel uptake in DabcBA B. ovis mutant. Therefore, we could not exclude the role of ABC transporter as nickel importer.
Additionally, ABC transporter function as peptide importer was analyzed in Salmonella Typhimurium by expressing B. ovis abcA-E locus and evaluating its sensitivity to known toxic peptides [29,32,47]. The complemented strain STmcomp was not able to take up the peptides tri-lysine and alafosfalin, which are expected to be imported by Tpp and Opp transporters, respectively (Figures S2). These findings imply that B. ovis ABC transporter probably did not work as tripeptide or as oligopeptide importer in the Salmonella assay, or that its specificity may be different from that of the corresponding Salmonella transporters. Moreover, WT and DabcBA B. ovis strains were not susceptible to the highest dose of toxic peptides in nutrient-limited media. B. ovis resistance against tri-lysine might result from a protective effect of outer membrane proteins, as rough strains are highly resistant to antimicrobial cationic peptides [48]. With these findings, it was not possible to conclude which specific substrate AbcEDCBA transports, so additional metabolic and proteomic comparative assays will be necessary to identify its function.
An additional effort in this study to understand the role of the B. ovis-specific ABC transporter was based on the comparison of the proteomic profiles between the WT and DabcBA. Interestingly, proteomic data resulted in the identification of five ABC transporter binding proteins which were considered pseudogenes according to the currently available B. ovis genomic data. This finding clearly indicates that computational annotation of B. ovis genome needs revision, given that predicted pseudogenes are expressed and potentially functional. Another proteomic study also demonstrated annotation errors in B. abortus database, after identifying four proteins encoded by pseudogenes [49].
Both SodC and SodB are conserved in the Brucella genus and essential for intracellular survival by evading the respiratory burst within phagocytes [51]. In B. abortus, a mutant lacking SodC is attenuated in macrophages and in mouse model [52]. Therefore, decreased Sod expression in the DabcBA B. ovis strain may be, at least partially, responsible for the attenuation of this strain [19,58].
B. ovis AbcEDCBA was previously shown to play a role in pathogenesis of B. ovis, as DabcBA B. ovis strain was attenuated in a mouse model as early as one day post infection and in mouse peritoneal macrophages at 12 mhpi [19]. A recent study also characterized the kinetics of DabcBA infection in sexually mature rams. The mutant lacking ABC transporter was not excreted in semen and urine of infected rams, although it induced a similar lymphocytic proliferative response when compared to WT B. ovis [58]. Considering that DabcBA was attenuated in both murine model and in the natural host, bacterial infection and trafficking were characterized in HeLa cells, to understand how ABC transporter contributes to B. ovis intracellular survival and replication. Epithelial cell lines differ from phagocytic cells due to lack of bactericidal activity against Brucella sp. [36,41], which allows for the study of infection and trafficking of severely attenuated mutants, including DabcBA B. ovis.
Even though B. ovis is a naturally rough and non zoonotic species [2], it was able to establish infection and successfully replicate in HeLa cells. Interestingly, the DabcBA mutant strain showed lower colonization in HeLa cells at 24 mhpi and the infection was controlled until 48 hours, although both strains demonstrated similar internalization (Figure 2). The kinetics of DabcBA infection was similar to that previously described in RAW macrophage cell line [19], which supports the notion that predicted ABC importer is crucial for B. ovis intracellular survival, even in the absence of macrophage bactericidal mechanisms.
Trafficking of classical pathogenic species of Brucella spp. is well described in both phagocytic and nonphagocytic cells [17,36,38,41,[59][60][61]. However, few studies have characterized the replication and intracellular trafficking of naturally rough Brucella spp. [62,63]. This is the first work illustrating B. ovis trafficking in HeLa cells by confocal microscopy, which demonstrate bacterial escape from LAMP-1 + compartment and intracellular replication at later time points. As shown in Figure 3, B. ovis intracellular trafficking was very similar to classical smooth Brucella spp. [17,41], with early interaction of BCV with lysosome (LAMP1 + ), followed by exclusion of LAMP1 and bacterial replication. Notably, B. ovis showed a later evasion of lysosome fusion only seen at 48 mhpi ( Figure 4B), and not as early as 24 hours, as described for B. abortus [41].
Compared to WT B. ovis, the DabcBA mutant strain remained within a LAMP-1 + compartment at all time points and, consequently, was not able to survive in HeLa cells (Figure 4). The trafficking defect observed for DabcBA B. ovis was identical to that previously described for Brucella spp. mutants with a nonfunctional T4SS, as they lose the capacity of excluding LAMP-1 after the initial fusion with the phagolysosomal compartment [7,17,41]. Exclusively in B. ovis, T4SS seems to have a critical role not only for persistent infection, but also for establishing early infection in mice and in peritoneal macrophages [16]. In smooth B. abortus, however, lack of T4SS does not interfere with early infection in the mouse, showing phenotype similar as the WT strain until five days post infection [9,10,13,16]. In agreement with our findings, previous studies demonstrated that B. ovis DabcBA and DvirB2 mutants have identical phenotypes in both mouse model and macrophages [16,19]. Interestingly, this study shows that the B. ovis DabcBA mutant was unable to express VirB proteins in either rich media or acid minimal media that mimics the early BCV environment (Figure 4). Therefore, it is likely that inactivation of AbcEDCBA results in decreased expression of T4SS and, consequently, impairment of B. ovis trafficking and replication within cells.
Early lysosome interaction during Brucella sp. intracellular trafficking is necessary for acidification and maturation of BCV, which is essential for bacterial survival, by inducing T4SS expression [17,38,64]. The expression of virB-encoded proteins in classical Brucella sp. is induced by low pH and poor nutritional conditions, which are observed during the initial phase of intracellular trafficking (approximately 5 hours) [38,42,44,65]. Conversely, WT B. ovis differed from all other Brucella species, due to in vitro expression of VirB proteins in both acid and rich neutral media ( Figure 4). This showed an exclusive mechanism for T4SS regulation in B. ovis, which was independent of an acidic environment.
Previous studies identified protein regulators of T4SS in Brucella spp., including histidine pathway HutC and a quorum sensing VjbR. Both regulators have an important role by directly binding the virB promoter and actively inducing virB transcription in a nutrient-poor environment, with low pH, and in presence of urocanic acid [38,39,43,44]. Considering that B. ovis also expressed T4SS in rich neutral media and that lack of VirB expression in DabcBA B. ovis was observed at any in vitro condition (Figure 4), we analyzed whether the transcription of the these two regulators or of the virB genes was affected in the DabcBA mutant. However, the reduced abundance of VirB proteins was not explained by defects in transcription or mRNA stability, since DabcBA and WT B. ovis showed similar mRNA levels of hutC, vjbR, virB1, and virB8 genes ( Figure 5). Post transcriptional regulation of T4SS proteins have been previously described in B. abortus [36]. In this species, the expression of VjbR and VirB7 were induced in vitro by low pH and solely in the presence of urocanic acid, although similar promoter activities of virB and vjbR were noticed in different growth conditions [39]. Taken together, our data supports the notion that T4SS expression in B. ovis is regulated by AbcEDCBA at a post transcriptional level. One possible mechanism by which this might occur is uptake of a substrate that acts directly to affect translation of virB mRNA, however alternatively, uptake of the ABC transporter substrate could have an indirect effect on VirB protein levels via indirect effects on other metabolic or regulatory pathways. Further experiments will be necessary to distinguish between these possibilities and to identify the substrate of the ABC transporter.
In conclusion, the result of this work revealed that the predicted peptide ABC importer AbcEDCBA was required for B. ovis in vitro expression of other ABC systems and virulence proteins (including Omp31 and Sod), as well as its intracellular survival and evasion from phagosome/lysosome fusion, by interfering with the expression of T4SS-encoded proteins through a post transcriptional mechanism.