The Three Flagellar Loci of Brucella ovis PA Are Dispensable for Virulence in Cellular Models and Mice

Brucella ovis is a facultative intracellular bacterium that causes a non-zoonotic ovine brucellosis mainly characterized by male genital lesions and is responsible for important economic losses in sheep farming areas. Studies about the virulence mechanisms of Brucella have been mostly performed with smooth (bearing O-polysaccharide in lipopolysaccharide) zoonotic species, and those performed with B. ovis have revealed similarities but also relevant differences. Except for few strains recently isolated from unconventional hosts, Brucella species are non-motile but contain the genes required to assemble a flagellum, which are organized in three main loci of about 18.5, 6.4, and 7.8 kb. Although these loci contain different pseudogenes depending on the non-motile Brucella species, smooth B. melitensis 16M builds a sheathed flagellum under particular culture conditions and requires flagellar genes for virulence. However, nothing is known in this respect regarding other Brucella strains. In this work, we have constructed a panel of B. ovis PA mutants defective in one, two or the three flagellar loci in order to assess their role in virulence of this rough (lacking O-polysaccharide) Brucella species. No relevant differences in growth, outer membrane-related properties or intracellular behavior in cellular models were observed between flagellar mutants and the parental strain, which is in accordance with previous results with B. melitensis 16M single-gene mutants. However, contrary to these B. melitensis mutants, unable to establish a chronic infection in mice, removal of the three flagellar loci in B. ovis did not affect virulence in the mouse model. These results evidence new relevant differences between B. ovis and B. melitensis, two species highly homologous at the DNA level and that cause ovine brucellosis, but that exhibit differences in the zoonotic potential, pathogenicity and tissue tropism.


INTRODUCTION
The genus Brucella is constituted by six classical species (B. melitensis, B. abortus, B. suis, B. canis, B. ovis, and B. neotomae), that cause brucellosis in terrestrial mammals, and six other species (B. ceti, B. pinnipedialis, B. microti, B. inopinata, B. vulpis, and B. papionis) that have been isolated since the 1990s from other terrestrial mammals or from marine mammals (https://lpsn.dsmz.de/genus/brucella). The Brucella spp. host range has more recently increased to amphibians and fish, with atypical strains isolated from several frog species and a ray (1,2).
Despite the high percentage of DNA-DNA hybridization detected among the classical Brucella species (96 ± 5% when compared to B. melitensis 16M) (3), some differential genetic markers have been found and they differ in several phenotypic characteristics, host preference and pathogenicity. Nevertheless, a common trait is their ability to survive and replicate inside phagocytic cells (4)(5)(6)(7). The Brucella species are smooth (S) or rough (R) depending on the presence or absence, respectively, of O-polysaccharide chains in the lipopolysaccharide (LPS). B. ovis and B. canis are the only rough Brucella species but are virulent for their natural hosts (sheep and dogs, respectively), which contrasts with the other Brucella species that are smooth and require S-LPS for full virulence (8)(9)(10).
Although studies regarding the virulence of R strains has increased in the last years, most work in this respect has been performed with S Brucella species (mainly with zoonotic B. melitensis, B. abortus, and B. suis). Among the genes involved in the virulence of smooth B. melitensis, flagellar genes are required for the establishment of a chronic infection in mice (11), which constitutes an intriguing trait since B. melitensis is a non-motile species (1). In fact, among the brucellae only B. inopinata and the Brucella atypical strains isolated from frogs and a ray are motile (1,2,(12)(13)(14) and at least frog isolates are able to build a polar flagellum in culture medium (1). Despite the presence of several pseudogenes in the three main flagellar loci (1,11) and its non-motile character (1), B. melitensis 16M is able to build a sheathed polar flagellum in the early exponential phase of growth (11) and, as mentioned above, flagellar mutants of B. melitensis 16M are attenuated in virulence (11). Although the three flagellar loci are conserved in the genus Brucella, with a different pattern of pseudogenization in most cases, no additional studies have been performed to evaluate the relevance of flagellar genes in the virulence other Brucella species. According to its rough phenotype, its particular outer membrane (OM)-related and virulence characteristics and its shared preference with B. melitensis by the ovine host (15)(16)(17)(18)(19)(20)(21), we have selected B. ovis to extend the knowledge about the role of flagellar genes in the virulence of the genus Brucella. With this aim, we have constructed a panel of flagellar mutants in rough virulent B. ovis PA (with one, two or the three flagellar loci deleted) that has been characterized regarding growth characteristics, OMrelated properties, intracellular behavior in cellular models of professional and non-professional phagocytes and virulence in the mouse model.

Plasmids, Bacterial Strains, and Culture Conditions
Plasmids pGEM-T Easy (Promega, Madison, WI, United States) and pCVDKan-D (18) were used to construct the recombinant plasmids containing the inactivated flagellar loci. They were maintained in Escherichia coli JM109 and CC118 (λpir), respectively. Recombinant E. coli strains were cultured at 37 • C in Luria Bertani (LB) medium supplemented with 50 µg/ml ampicillin (pGEM-T Easy derived plasmids) or kanamycin (pCVDKan-D derived plasmids). Virulent B. ovis PA was used as parental strain to obtain the panel of flagellar mutants and as reference strain for comparisons in the different assays. It was obtained from the bacterial culture collection maintained at the Institut National de la Recherche Agronomique, Nouzilly, France. B. ovis strains were cultured in tryptic soy agar (TSA) or tryptic soy broth (TSB) (Pronadisa-Laboratorios Conda, Torrejón de Ardoz, Spain), supplemented with 0.3% yeast extract (YE) (Pronadisa-Laboratorios Conda, Torrejón de Ardoz, Spain) and 5% horse serum (HS) (Gibco-Life Technologies, Grand Island, NY, United States). When required for the mutagenesis procedure, TSA-YE-HS was supplemented with kanamycin (Kan) at a final concentration of 50 µg/ml or with 5% sucrose (Sigma-Aldrich, St. Louis, MO, United States). B. ovis strains were cultured at 37 • C under a 5% CO 2 atmosphere.
DNA primers (IDT, Leuven, Belgium) used for gene expression analysis and for the construction and characterization of mutant strains are described in Table 1. PCR amplification was performed with AccuPOL DNA polymerase (VWR, Leuven, Belgium), Red Taq DNA polymerase master mix (VWR, Leuven, Belgium) or Expand TM Long Template PCR System (Roche, Mannheim, Germany), depending on the experiment. For gene expression studies, RNA was extracted with RNeasy mini kit (Qiagen, Hilden, Germany) from 5 × 10 9 CFU of B. ovis that had been cultured in liquid medium for 16 or 49 h (t16 or t49; exponential and stationary growth phase, respectively). Residual DNA was removed with RQ1 DNase (Promega, Madison, WI, United States) and cDNA was synthetized with the first strand cDNA synthesis kit for RT-PCR (Roche, Mannheim, Germany) using the random hexamers provided with the kit as primers for reverse transcriptase (RT). Parallel control reactions were performed in the same conditions but omitting RT. Subsequent PCR reactions were performed (using the cDNA as template and

Mutagenesis Procedure
Mutant strains for the three main flagellar loci ( Table 2) were obtained by in-frame deletion with overlapping PCR as described previously (18 Table 2) were obtained similarly with their specific primers ( Table 1). The single mutants for each flagellar locus served as parental strains for a second round of mutation leading to the deletion of an additional flagellar locus. The double mutants obtained ( Table 2) were subsequently used as parental strains to obtain the panel of triple mutants of B. ovis PA lacking the three main flagellar loci ( Table 2).

Growth, Autoagglutination, and Susceptibility Assays
Growth of mutant strains in solid and liquid medium was analyzed as previously described (26). Briefly, to evaluate growth in solid medium, bacterial suspensions in PBS with values of optical density at 600 nm (OD 600 ) of 0.2 were appropriately diluted and plated on TSA-YE-HS plates to determine the  To evaluate the autoagglutination ability, bacterial suspensions in TSB-YE-HS of OD 600 values of 0.8 (100% OD 600 ) were incubated for 48 h under static conditions to measure the evolution of the OD 600 scores (18,27). Susceptibility to polymyxin B, sodium deoxycholate and H 2 O 2 (all from Sigma-Aldrich) was measured using a disc assay as follows. Bacterial suspensions (100 µl) with OD 600 values of 0.2 were plated on TSA-YE-HS. Discs of 0.9-mm diameter (Los Productos de Aldo, Spain) were then placed in the middle of the plate and soaked with 20 µl of polymyxin B (250 000 UI/ml), sodium deoxycholate (10 mg/ml) or 30% H 2 O 2 . The diameter of the growth inhibition halo was recorded in quadruplicate for each plate after a 5-day incubation period and the results expressed as mean ± SD of three plates.

Virulence Evaluation in Cellular Models and Mice
Intracellular behavior of mutant strains was studied in J774.A1 macrophages and HeLa cells as previously described (19). Briefly, 2 × 10 4 J774.A1 macrophages/well or 1.5 × 10 4 HeLa cells/well were cultured in 96-well plates for 24 h. Bacteria (4 × 10 6 or 8 × 10 6 CFU/well for J774.A1 or HeLa cells, respectively) were allowed to internalize for 2 h in the cell lines. Extracellular bacteria were killed with gentamycin and intracellular bacteria were enumerated in three wells per bacterial strain after lysis of the eukaryotic cells and plating on TSA-YE-HS (t0). The remaining wells were incubated in the presence of gentamycin to evaluate intracellular bacterial numbers at 20 and 44 h (t20 and t44) post-infection (p.i.).
Virulence in mice was evaluated in 6-week old female BALB/c mice (Charles River Laboratories, Chatillon-sur-Chalaronne, France) received 1 week before. They were intraperitoneally inoculated with 10 6 CFU of parental B. ovis PA or the flagellar triple mutants B. ovis flg1 flg2 flg3, B. ovis flg2 flg1 flg3 or B. ovis flg3 flg2 flg1. Bacterial numbers in spleen were determined -as previously described (28)in 5 mice per group at 3, 7, and 11 weeks p.i. (W3, W7, and W11), which in B. ovis PA corresponds to the peak of infection in the acute phase, to the chronic phase and to the decline phase of infection, respectively (26,27).

Statistical Analysis
Statistical comparisons were performed with one-way ANOVA and Fisher's Least Significant Differences test on a GraphPad Prims Software (GraphPad Software Inc., San Diego, CA, United States). Statistically significant differences (P < 0.01) were established with a 99% confidence interval.

Genomic Organization and Transcription of the Flagellar Loci in B. ovis
According to the annotated whole genome sequence of the B. ovis reference strain (29), the three flagellar loci of B. ovis  Table 2. End-point RT-PCR (B) was performed with cDNA obtained by retrotranscription of RNA extracted at t16 from B. ovis PA or the flg1 flg2 flg3 triple mutant (RT+ reactions). Reactions of cDNA synthesis lacking RT were used as controls of DNA absence (RT-reactions). Amplification with fliC primers in the flg1 flg2 flg3 triple mutant (B) was expected, since both primers hybridize adjacent but externally to the deleted fragment (the deletion removes 80% of fliC). qRT-PCR (C) with fliF was performed with RNA extracted at t16 and t49 from B. ovis PA parental strain. Gene expression levels (C) were determined, with the StepOne TM software v2.3, by the 2 − Ct method with the 16S gene as internal reference and t16 results as control condition. Four biological replicates, with three technical replicates each, were analyzed and the results are expressed as means ± SD.
( Figure 1A) present a similar organization to that described for B. melitensis 16M (11) and are also located in chromosome II. A search in the PATRIC genome of motile Brucella sp. B13-0095 revealed the presence of additional flagellar genes motA, motB, fliJ, and fliO, that were also detected in chromosome I of B. melitensis 16M and B. ovis 63/290 ( Table 3). Hypothetical motA and motB genes were previously identified in locus III and locus I, respectively, of Brucella chromosome II (1, 11, 13) ( Table 3) but the four hypothetical flagellar genes detected in chromosome I have not been reported before in studies targeting the Brucella flagellum (1,11,13). According to the flagellum structure described for Gram-negative bacteria (30)(31)(32)(33)(34)(35)(36)(37), FliO would be part of the export gate (Figure 2) that extends from the membrane-supramembrane (MS ring) of the flagellum to the cytoplasm. FliJ, together with FliI and FliH, would constitute the ATPase complex (Figure 2) of the type III export machinery, although no gene potentially encoding FliH have been detected in the Brucella genomes. Similarly, fliD that encodes the filament cap protein in flagellated bacteria (Figure 2), has not been detected in the genus Brucella.
When compared to the genome of motile flagellated Brucella sp. B13-0095 isolated from a Pac-Man frog (13), B. melitensis 16M and B. ovis exhibited a different pattern of defective genes (Table 3, Figure 2) with the characteristics listed in Table 4. In the B. melitensis 16M genome, six flagellar genes with premature stop codons or frameshifts have been detected (Tables 3, 4). Additionally, the hypothetical start codon and ribosome binding site of the B. melitensis 16M flgA gene -coding for a putative chaperone for the FlgI P-ring protein (whose gene is frameshifted when compared to Brucella sp. B13-0095 flgI)-are lost due to a deletion of 18 nt. In B. ovis 63/290, five flagellar genes contain internal in-frame deletions shortening the encoded protein ( Table 3, blue lettering, and Table 4) and six additional genes ( Figure 1A, Table 3, red lettering, and Table 4) contain premature stop codons or frameshifts. However, flagellin fliC gene is not annotated as pseudogene in the whole genome sequence of B. ovis. Since in this work we have used virulent B. ovis PA, the possibility of some differences with B. ovis 63/290 cannot be discarded. However, previous genomic studies with B. ovis PA provided the same results as B. ovis 63/290 (18,19,38,39) and all sequences we have determined in this work for the   Figure 2 according to the general structure described for flagella of Gram-negative bacteria (30)(31)(32)(33)(34)(35)(36)(37) construction of flagellar mutants (including internal sequences not reported here) were identical to those of B. ovis 63/290. To evaluate whether flagellar loci are transcribed in B. ovis PA, end-point RT-PCR was performed with RNA extracted at the exponential phase of growth and primers targeting nine genes distributed in the three main loci. Transcription of all evaluated genes was detected in the parental strain B. ovis PA (Figure 1B), while no amplification was observed with cDNA obtained from the flg1 flg2 flg3 mutant, except for fliC ( Figure 1B). This exception was expected because the selected primers for RT-PCR amplify a 164 nt fragment of the 5 ′ -end of fliC that is externally bordering the deleted DNA fragment of locus I (deletion removes 99% of locus I, which includes 80% of fliC). Studies of relative expression in TSB-YE-HS liquid medium of the locus I fliF gene (performed by qRT-PCR using 16S as internal reference gene) showed that fliF is down regulated (with about 3.5-fold reduction) in the stationary growth phase (t49) when compared to the exponential growth phase (t16) (Figure 1C).

Construction, Growth, and OM-Related Properties of B. ovis PA Flagellar Mutants
Three initial B. ovis PA mutants were constructed (B. ovis flg1, flg2, and flg3), each with one of the three main flagellar loci deleted (locus I, II, and III, respectively). Despite the high size of the deleted fragment, mainly in B. ovis flg1 (about 18 kb deleted), no difficulties were found to obtain the three mutants. Similarly, double and triple mutants combining the deletion of two or the three flagellar loci, were obtained (32.5 kb deleted in triple mutants). All possible combinations of double and triple mutants were obtained in order to set out a panel of mutants that could be analyzed in case of discovering a differential behavior in one mutant and thus minimize the risk of attributing differences caused by other undesired mutations to the absence of flagellar loci.
No differences in growth in solid medium were observed in single, double or triple mutants that showed similar CFU/ml values for bacterial suspensions of OD 600 = 0.2 than the parental strain (data not shown). Equivalent results among strains were also observed in TSB-YE-HS liquid medium with a similar evolution of OD 600 values and CFU/ml with time (Supplementary Figure 1). In the autoagglutination assay no differences were found among strains since, as expected for the parental B. ovis PA strain (26), all of them remained in suspension (Supplementary Figure 2). According to these results, only three triple mutants (B. ovis flg1 flg2 flg3, flg2 flg1 flg3 and flg3 flg2 flg1, which have the three flagellar loci deleted in a different order) were initially selected for the remaining studies. The other mutants would only be analyzed if differences were found with the triple mutants.
Properties related to the OM, and that have also been related to survival in the host, were evaluated in the selected triple mutants in comparison with B. ovis PA. Diameters of growth inhibition halos obtained by exposure to H 2 O 2 , sodium deoxycholate or polymyxin B did not show relevant differences between the three triple flagellar mutants and the parental strain (Supplementary Figure 3).

Virulence of B. ovis PA Flagellar Mutants
Since B. ovis is an intracellular pathogen (6,17,19,26,40), the behavior in J774.A1 and HeLa cells of the triple mutants was evaluated in comparison to that of the parental strain. Removal of the three main flagellar loci in B. ovis PA did not affect the internalization of the bacterium or its intracellular evolution (Figures 3A,B). These results were somehow expected since flagellar mutants of B. melitensis 16M did not show an altered intracellular pattern (11), although it must be taken into account that each one of these mutants was only defective in one single gene.
On the contrary, while single gene mutants of B. melitensis 16M were unable to establish a chronic phase of infection in mice (11), the three triple mutants of B. ovis PA analyzed, lacking the 32.5 kb of the three main flagellar loci, did not show attenuation in BALB/c mice (Figure 3). Thus, both the splenic bacterial counts ( Figure 3C) and the spleen weight (Figure 3D) of the flagellar mutants showed the same temporal evolution than those observed with the parental strain. Flagellar proteins encoded by genes that, when compared to those of motile Brucella sp. B13-0095, contain frameshifts, premature stop codons or mutations affecting the start codon are written in red bold characters and represented by unfilled forms with red borders. This pattern has also been used for FliD and FliH, which have not been found in the Brucella genomes. B. ovis proteins with internal in-frame deletions in the encoding genes are written in blue bold characters. B. ovis motB in chromosome II contains a frameshift; the figure represents a MotB homolog encoded in chromosome I that is shorter than the B. melitensis ortholog due to an internal in-frame deletion in the encoding gene. C-ring (cytoplasmic ring), MS-ring membrane supramembrane ring, P-ring (peptidoglycan ring), L-ring (lipopolysaccharide ring), IM (inner membrane), PG (peptidoglycan), OM (outer membrane).

DISCUSSION
Since the classical Brucella species lack motility (41) and display random patterns of pseudogenization (1), it is tempting to hypothesize that flagellar loci, which are conserved in the genus Brucella, are remnants of an environmental ancestor that no longer required motility after the evolutive adaptation to the animal host and to an intracellular lifestyle. However, the detection of a flagellum in B. melitensis 16M and the involvement of flagellar genes in its virulence in mice (11,42,43) and probably in goats (44) raises a new perspective. Moreover, although the relevance for virulence remains unexplored, the recently reported motility of atypical Brucella strains (1,2,(12)(13)(14) and their ability, at least for amphibian isolates, to build a polar flagellum (1) also encourages additional studies to elucidate the function of flagellar genes in the genus Brucella. In this work, we have selected a virulent B. ovis strain to construct and characterize a panel of mutants in flagellar loci with a main focus in virulence.
Although the profile of defective flagellar genes detected in B. ovis (Figure 2, Tables 3, 4) makes the assembly of a flagellum unlikely, B. melitensis 16M also contains defective genes (Figure 2, Tables 3, 4) that would not be compatible with the synthesis of a flagellum. Therefore, it is probable that either the modified proteins are functional or B. melitensis 16M is able to synthetize at least some whole-length molecules by suppression of the stop codons (i.e., fliF and flhA) (11) and compensation of DNA frameshifts (i.e., flgI, fliG, flgF, and fliI) by transcription slippage or ribosomal frameshift (45). Accordingly, the defects observed in some B. ovis flagellar genes do not necessarily imply the impossibility of assembling a complete or partial flagellar structure that could contribute to virulence. We have detected that B. ovis PA is able to transcribe flagellar genes located in the three main loci and that, as described for B. melitensis 16M (11), the transcription level is higher in the exponential growth phase, at least for fliF (Figures 1B,C). To our knowledge, no other studies have evaluated expression of flagellar genes in B. ovis, either in culture medium or inside phagocytes, and the sole study that analyzed the intracellular transcriptome of B. ovis did not report upregulation or downregulation of flagellar genes (46). However, expression of B. ovis flagellar genes in an intracellular environment, as it has been reported for B. melitensis 16M (11), cannot be discarded. Although this aspect would merit further attention, our results clearly demonstrate that the entire three main flagellar loci of B. ovis PA (accounting for 39 genes) are dispensable for all properties evaluated, including intracellular survival and virulence in the mouse model (Figure 3 and Supplementary Material). Therefore, most likely B. ovis does not build a flagellum or, if it does, it would not be required for the establishment of infection.
Since the mechanism/s responsible for the contribution of flagellum to virulence of B. melitensis 16M has not been elucidated, it is difficult to hypothesize about how the presumed absence of flagellum in B. ovis has influenced the host-pathogen interaction. Flagella may be involved in the four main stages of the infectious process of bacterial pathogens (47): (i) reaching the host or target site (ii) colonization and invasion, (iii) maintenance and replication, and (iv) dispersal to new hosts. The absence of motility (at least in vitro) and of chemotactic systems would exclude the first role in the classical Brucella species. Role in adhesion and invasion of host cells (at least in cell cultures) is also unlikely, since B. melitensis flagellar mutants show no internalization defects in HeLa cells or in bovine peritoneal macrophages (11). The same cellular models also revealed that, even though the fliF promoter is induced intracellularly, the flagellum is not required for intracellular replication of B. melitensis 16M (11). Additionally, attenuation in mice of B. melitensis 16M flagellar mutants was not detected at 1W p.i., but only at later time points (11,43). However, in vivo mouse imaging technology showed that a luminescent B. melitensis 16M flagellar mutant, lacking four genes coding for rod proteins, had a limited ability to disseminate from the point of intraperitoneal inoculation (48). This impaired dissemination might be related to the impossibility of B. melitensis 16M flagellar mutants to establish a chronic infection in mice (11). But also, if this behavior were reproduced in the natural host, the flagellum could be responsible, at least in part, for the tropism of B. melitensis 16M by the placenta. This statement would be in accordance with the fact that B. melitensis infections frequently induces abortions (21) while B. ovis (that share with B. melitensis the preference for the ovine host) exhibits a marked tropism by the male genital tract and is seldom associated to abortions (20) despite its ability to internalize and replicate in trophoblasts (6). Moreover, a contribution of the flagellum to the zoonotic potential of B. melitensis, which is the highest of the genus, should also be considered. More studies involving flagellar genes in other Brucella species associated with abortions in their preferred hosts or able to infect humans would help to clarify these points.
The exacerbated virulence pattern, accompanied by histological damage in spleen, that was observed in BALB/c mice with a non-polar fliC mutant of B. melitensis 16M constitutes an additional remarkable observation regarding flagellar genes (43). It was proposed that FliC flagellin of B. melitensis 16M triggers the innate immune response and that a tight regulation of flagellar expression in this strain is part of the stealthy strategy that allows to maintain a persistent infection without severely damaging host tissues (43). Some other evidences point to the requirement for a finely tuned regulation of flagellar genes in B. melitensis 16M to establish a persistent infection: (i) the large number of reported direct or indirect regulators of flagellar gene expression: FtcR, FlbT, and FlaF, which are encoded in flagellar locus I (Figure 1A), or VjbR, BlxR, RpoE1, BpdA, and YbeY (48)(49)(50)(51)(52)(53)(54), (ii) the results obtained in an in vivo model simulating the onset of B. melitensis 16M infection in cattle (first 4 h) showing that the three main flagellar loci were repressed while transcription of the rpoE1 repressor gene was activated (55), and (iii) the flagellum is sheathed by an extension of the outer membrane ending by a club-like structure that has been suggested to contribute to the assembly of FliC flagellin subunits (42) in the absence of the filament cap protein FliD (Figure 2) that has not been detected in the Brucella genomes; since the Brucella outer membrane is considered as part of its stealthy strategy to establish persistent infections (56), the flagellar sheath could contribute to limit FliC presentation to the immune system.
In the case of B. ovis PA and even if a flagellum is not assembled, flagellin is likely to be synthetized, since this strain is able to transcribe fliC ( Figure 1B) and B. melitensis 16M synthesizes FliC even in the absence of deeper flagellar structural proteins such as FliF (basal body protein) or FlgE (hook protein) (53). Therefore, flagellin might be translocated to the cytoplasm of the host cell through the type-IV secretion system (encoded by the virB operon), as it has been suggested for B. melitensis 16M, and induce an innate immune response mediated by cytosolic NCRC4 receptors (43). However, even if B. ovis PA produces flagellin and is able to translocate it into the host cell cytoplasm, it would not be relevant in the induction of a detrimental immune response for the bacterium because flagellar mutants behave as the parental strain (Figure 3). This fact could be related with the differences detected in the C-terminal residues of FliC in B. ovis when compared to the protein of B. melitensis (Tables 3, 4). Both N-and C-terminal domains are involved in the self-polymerization of flagellin subunits (35,57), but Cend residues have also been proposed as targets for the innate immune response sensed by NLRC4 receptors (43). On the other hand, presentation of assembled flagellin subunits in a surface-exposed flagellar structure could be essential to induce the immune response (and/or to interact with its effectors), and the pattern of defective flagellar genes in B. ovis would not allow this requirement.
Another intriguing observation regarding the B. melitensis 16M flagellum is the contrast between the exacerbated virulence in mice of the fliC mutant (unable to synthetize flagellin and therefore the filament of the flagellum) and the attenuation of the fliF mutant (unable to synthetize the MS-ring proteins) (43). This characteristic suggests that the flagellar export channel of B. melitensis 16M could also be used for the transport of molecules required to maintain a chronic infection in the host. Whether the Type III machinery involved in the specialized export of the flagellar subunits (32,33) contributes to the export of other molecules in B. melitensis 16M that might participate in virulence has not been elucidated. If this were the case and considering the full virulence of flagellar mutants (Figure 3), this possibility does not seem to occur in B. ovis either by a naturally defective channel or by the absence of the hypothetical virulence determinants.
The demonstration that flagellar genes are dispensable for B. ovis virulence in mice (Figure 3) constitutes a new particular characteristic of this rough species to add to the previously reported differences with other brucellae (16)(17)(18)(19). To build a profile of differential characteristics for each Brucella species would contribute to decipher the mechanisms underlaying the differences of pathogenicity and host preference that exist between the classical Brucella species despite their high similarity at the DNA level. However, although the mouse model usually mimics the results obtained in the natural host for attenuated mutants, it has limitations (58,59), and Brucella mutants exhibiting whole virulence in the mouse model but attenuated in the natural host have been reported (60). Therefore, although unlikely, a role of B. ovis flagellar genes in the natural host cannot be completely discarded. More studies in other Brucella species, including abortifacient and zoonotic Brucella species and the recently isolated motile strains, would help to clarify the relevance of flagellar genes in the genus Brucella.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

ETHICS STATEMENT
Mice experiments were designed according to the Spanish and European legislation for research with animals (RD 53/2013 and directive 86/609/EEC). Microbiological procedures and experimentation with mice were approved by the Biosecurity and Bioethics Committees of the University of Salamanca and certified by the competent authority of Junta de Castilla León, Spain.

AUTHOR CONTRIBUTIONS
RS-M and NV conceived the study and wrote the manuscript. RS-M, CT, and NV participated in the experimental work, the discussion of the results, and the revision of the manuscript. All authors read and approved the final version of the manuscript.

FUNDING
This work was financed by Grant SA151G18 that was awarded by Consejería de Educación, Junta de Castilla y León, Spain.