Fourteen B. ovis field isolates obtained from sheep between 1982 and 1995 were used in this study. They were provided from the collection of Instituto de Pesquisas Veterinárias Desidério Finamor and were isolated (by FPP and MGD) from semen samples collected by electroejaculation from rams in Rio Grande do Sul, Brazil (Santana do Livramento - 10; Uruguaiana - 2; and undefined municipalities - 2). All isolates from Santana do Livramento were from animals of the same herd, whereas the others four B. ovis isolates had not information about herd of origin. All isolates were confirmed to be B. ovis by biochemical and molecular tests 18 19 20 . Approval to use the B. ovis isolates in this study was formally given by the director of IPVDF.

Brucella ovis colonies were inactivated at 85°C for 2 hours and subjected to genomic DNA extraction 21 22 . DNA from each strain was genotyped by MLVA16, which was divided in: panel 1 (Bruce06, Bruce08, Bruce11, Bruce12, Bruce42, Bruce43, Bruce45, Bruce55); panel 2A (Bruce18, Bruce19, Bruce21); and panel 2B (Bruce04, Bruce07, Bruce09, Bruce16, Bruce30) 11 15 .

From digitalized image of each gel, the band size was estimated and then converted into number of repeat units for each locus by using the software BioNumerics 6.1 (Applied Maths, Belgium) 15 . Brucella melitensis 16M (ATCC 23456^T) was used as control for band size estimation of all MLVA16 loci. The genotypes obtained were compared to those deposited in the MLVAbank 2014 (http://mlva.u-psud.fr/brucella/). Clustering analysis was performed using the category coefficient and UPGMA (BioNumerics 6.1) 15 . The Hunter-Gaston diversity index (HGDI) was used 23 . The minimum-spanning tree (MST) was generated using Prim’s algorithm associated with priority rule (eBURST algorithm) and bootstrap resampling 24 25 (BioNumerics 6.1). The MST presented is the top score tree, the tree with the highest overall reliability score.

Analysis of the MLVA16 loci revealed thirteen distinct genotypes among the fourteen B. ovis strains evaluated (Figure 1) and a HGDI of 0.989. All these MLVA16 patterns represented new genotypes, since no correspondence with those deposited on MLVAbank 2014 was found. However, the comparison of results observed in the eight conserved loci included in the panel 1 (MLVA8) with those available in the MLVAbank 2014 (http://mlva.u-psud.fr/brucella/) revealed that nine among the fourteen isolates had MLVA8 profile identical to profile 1 (Bruce06: 3; Bruce08: 5; Bruce11: 2; Bruce12: 10; Bruce42: 1; Bruce43: 1; Bruce45: 5; Bruce55: 2). The other five B. ovis isolates exhibited two different MLVA8 patterns, which were different of the MLVA8 1 and 2 genotypes (genotype 2 = Bruce06: 2; Bruce08: 5; Bruce11: 2; Bruce12: 10; Bruce42: 1; Bruce43: 1; Bruce45: 5; Bruce55: 2) (the only ones already described for B. ovis) due to polymorphisms in loci Bruce06, 08 and 12. The MST created based on MLVA16 genotypes is shown in Figure 2. Besides the B. ovis strains tested in the present study, all nineteen MLVA16 genotypes of B. ovis available in the MLVAbank 2014 were included in clustering and MST analyses. Analysis of geographical origin in the MST showed that B. ovis strain BCCN 98–46 from Argentina was closely related to a Brazilian B. ovis isolate, strain 241E (Figures 1 and 2). Moreover, MST analysis also revealed the existence of two clearly distinct major clonal complexes (clonal complexes A and B).

Genotyping of microorganism of great veterinary importance, such as B. ovis, is a valuable tool for the control of disease, since it allows the characterization of outbreaks and, the determination of the source of infection and transmission routes 26 . In the present study, molecular characterization of fourteen B. ovis field isolates revealed a high genetic diversity among strains (Figure 1). Interestingly, among ten B. ovis isolates from same herd only two strains had identical patterns (Figure 1). The existence of many different genetic profiles within the same herd has two possible explanations: first, the existence of an intense animal traffic led the introduction of the agent from different origins and second, all B. ovis strains isolated from outbreak were originated from the same B. ovis strain that undergone some changes in loci of MLVA16. Although there are no epidemiological data that can confirm or refute the first explanation, the second hypothesis seems less likely, since the differences observed among the ten B. ovis strains from same herd were not the result of one-repeat unit increase or decrease and were also not restricted to only one MLVA16 locus or panel. Moreover, even though some data had suggested short term evolution particularly among panel 2B loci 27 28 , there was also polymorphism at locus Bruce08 from the most conserved panel (panel 1) (Figure 1). On the other hand, in contrast to smooth strains such as B. abortus, B. melitensis and B. suis that have demonstrated a high stability of all MLVA16 loci under in vivo and in vitro conditions 12 13 14 29 , MLVA16 performed on B. canis, a rough strain, suggesting a hypervariability particularly in some panel 2B loci 30 . Whole genome sequencing of these B. ovis strains from the same herd would be the better way to understand the biological significance of the high genetic diversity observed without any concerns, however it is less practical and much more expensive.

Clustering analysis also showed a large distance between the two isolates from Uruguaiana (Bruce09, 04, 07 and 16), and between the two B. ovis strains from undefined municipalities (Bruce08, 09, 07 and 16), likewise in comparison among all four isolates (Figure 1). These major differences in the MLVA16 genotypic profile and the large difference in the years of isolation of the strains (1982, 1985 and 1995) (Figure 2), together, strongly suggest that no epidemiological relationship exist among these four B. ovis isolates.

Minimal spanning tree analysis revealed the existence of two clearly distinct major clonal complexes (clonal complexes A and B) (Figure 2), one composed by most of Brazilian B. ovis isolates plus French strains and a single strain from Argentina, Australia, Spain and USA (clonal complex A), and a second one with fewer representatives and composed by two strains from France and a single strain from Argentina and Brazil (clonal complex B) (Figure 2). The establishment of these relationships is central to develop a model for evolutionary steps in the difference of the B. ovis MLVA16 genotypes. Nevertheless, more representative sampling is needed for inclusion into this model for a more robust comparison. Therefore, data of present study are especially important, because it expands the universe of B. ovis strains genotyped by MLVA16 in both, amount and origin of strains.

Moreover, since Rio Grande do Sul State is bordered by Argentina, the close relationship between B. ovis strain BCCN 98–46 from Argentina and the Brazilian B. ovis isolate 241E suggests that B. ovis strains were circulating in the Brazilian – Argentinean border. In this context, animal importation could also explain the very close localization of B. ovis isolates from Brazil and B. ovis strains from France and Spain in MST analysis. Although there are no recent records about importation of animals from these countries to Rio Grande do Sul, historical records show that the formation of the sheep flock of this State was mainly achieved through the importation of animals from various countries of Europe and Oceania 31 32 . Furthermore, the main activity of the flock from Santana do Livramento, RS, from where most B. ovis strains were isolated, was the rearing of Texel breeders, a breed whose origin is in France and the Germany.

In conclusion, the results of the present study showed a high genetic diversity among B. ovis field isolates from Rio Grande do Sul State, Brazil by MLVA16.

The authors declare that they have no competing interests.

EMSD and GNF participated in design of the study, data acquisition and analysis. EMSD and wrote the paper. MGD, FPP and APL conceived and participated in design of the study, and critically reviewed the manuscript. All authors read and approved the final manuscript.