Examination of Taxonomic Uncertainties Surrounding Brucella abortus bv. 7 by Phenotypic and Molecular Approaches

Abstract

Brucella taxonomy is perpetually being reshuffled, at both the species and intraspecies levels. Biovar 7 of Brucella abortus was suspended from the Approved Lists of Bacterial Names Brucella classification in 1988, because of unpublished evidence that the reference strain 63/75 was a mixture of B. abortus biovars 3 and 5. To formally clarify the situation, all isolates previously identified as B. abortus bv. 7 in the AHVLA and ANSES strain collections were characterized by classical microbiological and multiple molecular approaches. Among the 14 investigated strains, including strain 63/75, only four strains, isolated in Kenya, Turkey, and Mongolia, were pure and showed a phenotypic profile in agreement with the former biovar 7, particularly agglutination with both anti-A/anti-M monospecific sera. These results were strengthened by molecular strategies. Indeed, genus- and species-specific methods allowed confirmation that the four pure strains belonged to the B. abortus species. The combination of most approaches excluded their affiliation with the recognized biovars (biovars 1 to 6 and 9), while some suggested that they were close to biovar 3.These assays were complemented by phylogenetic and/or epidemiological methods, such as multilocus sequence analysis (MLSA) and variable-number tandem repeat (VNTR) analysis. The results of this polyphasic investigation allow us to propose the reintroduction of biovar 7 into the Brucella classification, with at least three representative strains. Interestingly, the Kenyan strain, sharing the same biovar 7 phenotype, was genetically divergent from other three isolates. These discrepancies illustrate the complexity of Brucella taxonomy. This study suggests that worldwide collections could include strains misidentified as B. abortus bv. 7, and it highlights the need to verify their real taxonomic position.

INTRODUCTION

Brucellosis is a major worldwide zoonosis. This disease affects domestic and wild mammals, causing abortion and reduced fertility. The infection is transmitted to humans by animals through direct contact with contaminated animal fluids or indirectly through ingestion of unpasteurized milk products. Despite surveillance and eradication programs recommended by the World Health Organization (WHO), the Food and Agricultural Organization (FAO), and the World Organization for Animal Health (OIE), the disease remains endemic in many regions of the world (1). Accordingly, brucellosis is of serious public health importance and causes substantial losses to livestock producers and international trade for herds in areas where it is enzootic.

Brucella, the causal agent of brucellosis, is a genus of Gram-negative, nonmotile, mostly oxidase- and urease-positive, nonencapsulated, and facultative intracellular bacteria. On the basis of genetic criteria (DNA-DNA hybridization and 16S rRNA sequence comparison), the genus Brucella belongs to the family Rhizobacteriaceae, class Alphaproteobacteria, within the phylum Proteobacteria (2). Classification of these bacteria has been based primarily on phenotypic and biochemical methods and host preference (Table 1). On this basis, the Brucella genus currently contains 10 species, most with a preferential host: B. abortus infects cattle, B. suis is normally associated with swine, B. melitensis infects mainly sheep and goats, B. ovis seems to be responsible for a specific infection of sheep, B. canis is associated with dogs, B. neotomae is from desert wood rats, B. ceti and B. pinnipedialis infect marine mammals (3), B. microti has been isolated from the common vole (Microtus arvalis) (4, 5), and while B. inopinata was isolated from a human infection, its reservoir remains unknown (6). Additional novel strains, such as unnamed strains isolated from baboons (7, 8), foxes (9), frogs (10), and rodents (11), have been described recently, and ongoing updates on the Brucella taxonomy are expected in the near future.

TABLE 1.

Differential characteristics of the Brucella species and biovars^a

^aData are from references 28 and 37 and http://www.the-icsp.org/taxa/Brucellalist.htm. R and S, rough and smooth colonial morphology, respectively; CO₂, CO₂ requirement; H₂S, H₂S production; agglutination, agglutination with monospecific A, M, and R (rough) antisera. +, growth or lysis; −, no growth or no lysis; +/−, partial lysis; (+), most strains positive; (−), most strains negative.

^bDye concentration of 20 μg · ml⁻¹ in serum dextrose medium (1/50,000).

^cWithdrawn (ICSP).

^dND, not yet designed in the Brucella systematics.

^eStrong intensity.

^fLow intensity.

Some species are subdivided into biovars, i.e., B. melitensis bv. 1 to 3, B. abortus bv. 1 to 6 and 9, and B. suis bv. 1 to 5. Prior to 1986, the B. abortus species included 8 biovars (1 to 7 and 9), biovar 8 having been deleted from Brucella nomenclature in 1978 by the International Committee on Systematics of Prokaryotes (ICSP) (http://www.the-icsp.org/taxa/Brucellalist.htm) because no authentic isolate of this biovar had been reported for many years and no reference strain was available (12). From 1977, strain 63/75, also designated type strain NCTC 10506 or ATCC 23454, was considered the B. abortus bv. 7 reference strain (13) (Table 1). Between 1986 and 1988, following successful cloning, Verger et al. (INRA, Nouzilly, France) suggested that strain 63/75 consisted of a mixture of B. abortus bv. 3 and bv. 5 strains (14), so biovar 7 was suspended from the Approved Lists of Bacterial Names Brucella classification in 1986 by the International Subcommittee on the Taxonomy of Brucella (ISTB) (14) until the situation could be clarified.

The genetic homogeneity of the Brucella genus (DNA homology of >90%) initially strongly hindered the development of molecular tools for species and biovar identification. The rrs polymorphism is applicable for genus identification but does not allow differentiation between Brucella species (there is almost 100% identity in the 16S rRNA sequences) (15). To date, the most relevant genus identification technique is a real-time PCR (RT-PCR), which targets bcsp31, IS711, and per (16). For species identification, molecular approaches must target other loci. Polymorphism of omp (outer membrane protein) genes, mainly the deletion of omp31 in B. abortus, is useful for identifying genetic variants (17). Other multiplex PCR assays are available, i.e., AMOS PCR (18), which discriminates B. abortus (biovars 1, 2, and 4), B. melitensis (biovars 1 to 3), B. ovis, and B. suis (only biovar 1), and Bruce-ladder (8 target genes), which differentiates between the classical, vaccine and marine Brucella species (19, 20). In addition, molecular typing methods with greater discriminatory power, such as multilocus variable-number tandem repeat analysis 16 (MLVA-16) (21, 22), variable-number tandem repeat 21 (VNTR-21) (23), and extended multilocus sequence analysis 21 (MLSA-21) (24; A. M. Whatmore, unpublished data) can be used to further subdivide species, giving insight into phylogenetic, taxonomic, and/or epidemiological links between different terrestrial and marine strains.

The withdrawal of B. abortus bv. 7 from Brucella systematics remains poorly understood or simply ignored within the scientific community. Indeed, the literature still abounds in typing studies that include the former biovar 7 reference (presumably mixed) strain (5, 25–27). Moreover, several laboratories still count in their strain collections some isolates probably misidentified as B. abortus bv. 7. Therefore, we proposed to determine if, as for the reference strain 63/75, these isolates were a mixture of various biovars or whether they constituted a new biovar with its own characteristics. The aim of our study was to clarify the situation concerning B. abortus bv. 7 by both conventional and molecular approaches, including RT-PCR, omp polymorphism, Bruce-ladder, AMOS-ERY PCR, IS711 fingerprinting, MLSA-21, VNTR-21, and MLVA-16, and thus provide recommendations on the future taxonomic status of this biovar.

MATERIALS AND METHODS

Bacterial strains.

A total of 14 strains, including strains historically labeled B. abortus bv. 7 from the AHVLA and ANSES Brucella collections and the reference strain 63/75, were analyzed. The Brucella strains used in this study are listed in Table 2. In order to exclude the possibility of a mixed culture of different biovars, the strains were subjected to three successive cloning isolations (probability of a mixed colony = 10⁻⁹) (J.-M. Verger, personal communication).

TABLE 2.

Strains identified as Brucella abortus bv. 7 in the AHVLA and ANSES collections and their phenotypic characteristics^a

Isolate	AHVLA no.	Yr	Host	Origin	Pure strain no.	Genus	Species	Biovar	Morphology	CO2	H2S	Oxidase	Urease	Agglutination			Dye sensitivityb				Phage lysis
														A	M	R	T1	T2	F1	F2	Tb RTD	Wb RTD	Iz RTD	R/C RTD
07-994-2401	84/35	1984	ND	Poland	07-994-2401	Brucella	abortus	4	S	+	+	+	+	−	+	−	−	−	+	+	+	+	+	−
07-994-2402	62/05	1962	ND	ND	07-994-2402	Brucella	abortus	3	S	+	+	+	+	+	−	−	−	+	+	+	+	+	+	−
07-994-2403	84/31	1984	ND	ND	07-994-2403a	Brucella	abortus	1	S	−	+	+	+	+	−	−	−	−	+	+	+	+	+	−
					07-994-2403b	Brucella	melitensis	1	S	−	−	+	+	−	+	−	+	+	+	+	−	−	+	−
07-994-2404	84/37	1984	ND	ND	07-994-2404a	Brucella	abortus	1	S	−	+	+	+	+	−	−	−	−	−	+	+	+	+	−
					07-994-2404b	Brucella	abortus	4	S	+	+	+	+	−	+	−	−	−	+	+	+	+	+	−
07-994-2405	84/40	1984	ND	ND	07-994-2405a	Brucella	abortus	4	S	+	+	+	+	−	+	−	−	−	+	+	+	+	+	−
					07-994-2405b	Brucella	abortus	1	S	−	+	+	+	+	−	−	−	−	−	+	+	+	+	−
07-994-2406	F6/5/05	2005	Bovine	ND	07-994-2406	Brucella	abortus	3	S	+	+	+	+	+	−	−	+	+	+	+	+	+	+	−
07-994-2407	84/32	1984	Human	ND	07-994-2407a	Brucella	abortus	1	S	−	+	+	+	+	−	−	−	−	−	+	+	+	+	−
					07-994-2407b	Brucella	abortus	4	S	+	+	+	+	−	+	−	−	−	+	+	+	+	+	−
07-994-2408	63/59	1963	ND	Poland	07-994-2408	Brucella	abortus	1	S	−	+	+	+	+	−		−	−	−	+	+	+	+	−
07-994-2409	63/66	1963	ND	Poland	07-994-2409	Brucella	abortus	5	S	−	−	+	+	−	+	−	−	−	+	+	+	+	+	−
07-994-2410	63/75c	1963	ND	Poland	07-994-2410a	Brucella	abortus	3	S	−	+	+	+	+	−	−	+	+	+	+	+	+	+	−
					07-994-2410b	Brucella	abortus	5	S	−	−	+	+	−	+	−	−	+	+	+	+	+	+	−
07-994-2411	63/294	1963	Bovine	Kenya	07-994-2411	Brucella	abortus	ND	S	−	+	+	+	+	+	−	+	+	+	+	+	+	+	−
03-4923-239-D		2003	Bovine	Turkey	03-4923-239-D	Brucella	abortus	ND	S	+/−	+	+	+	+	+	−	+	+	+	+	+	+	+	−
99-9971-135d		1988	Bovine	Mongolia	99-9971-135	Brucella	abortus	ND	S	−	+	+	+	+	+	−	+	+	+	+	+	+	+	−
99-9971-159		1993	Bovine	Mongolia	99-9971-159	Brucella	abortus	ND	S	−	+	+	+	+	+	−	+	+	+	+	+	+	+	−

^aR and S, rough and smooth colonial morphology, respectively; CO₂, CO₂ requirement; H₂S, H₂S production; agglutination, agglutination with monospecific A, M, and R (rough) antisera. +, growth or lysis; −, no growth or no lysis; +/−, partial lysis. ND, not determined.

^bT, thionin; F, basic fuchsin; T1 and F1: dye concentration of 10 μg · ml⁻¹ in serum dextrose medium (1/25,000); T2 and F2, dye concentration of 20 μg · ml⁻¹ in serum dextrose medium (1/50,000).

^cFormer B. abortus bv. 7 reference strain.

^dProposed reference strain for B. abortus bv. 7.

The reference strains B. abortus bv. 1 strain 544, B. abortus bv. 2 strain 86/8/59, B. abortus bv. 3 strain Tulya, B. abortus bv. 4 strain 292, B. abortus bv. 5 strain B3196, B. abortus bv. 6 strain 870, and B. abortus bv. 9 strain C68 and the B. abortus bv. 1 vaccine strains S19, S99, and RB51, as well other reference strains, such as B. melitensis bv. 1 strain 16M, B. melitensis bv. 3 strain Ether, B. suis bv. 1 strain 1330, B. suis bv. 4 strain 40, and B. ovis strain 63/290, were included in this study as controls for phenotypic typing and/or for molecular analysis.

Analysis of phenotypic characteristics.

Pure cloned strains were characterized using the conventional Brucella typing methods, as previously described, i.e., CO₂ requirement, H₂S production, urea hydrolysis, oxidase test, agglutination with monospecific sera (anti-A, anti-M, and anti-R), dye sensitivity (basic fuchsin and thionin), and phage typing (Tbilisi [Tb], 10⁴ Tb, Weybridge [Wb], Izatnagar₁ [Iz₁], and R/C) (28).

Molecular analysis. (i) DNA preparation.

Molecular tools were applied only on pure clones according to classic biotyping description of biovar 7. Genomic DNA was extracted using the High Pure PCR template preparation kit (Roche Diagnostics, France) according to the manufacturer's instructions.

(ii) PCR analysis and typing methods.

(a) Real-time PCR. To confirm the genus Brucella, the RT-PCR assay which targets bcsp31, IS711, and per was performed on the genomic DNA as previously described (16).

(b) omp fingerprinting. Polymorphism of the omp31, omp2a, and omp2b genes was studied by PCR assays as previously described (29). Furthermore, restriction fragment length polymorphism (RFLP)-PCR was applied to various outer membrane protein-encoding genes (omp31/AvaII and/HaeIII, omp2a/NcoI and/StyI, and omp2b/KpnI and EcoRI) as previously described (30, 31).

(c) Bruce-ladder multiplex PCR. Bruce-ladder is species specific, and all the strains and biovars from the same Brucella species give the same profile. Pure clones obtained in our study were characterized by the Bruce-ladder multiplex PCR as previously described (19).

(d) AMOS-ERY PCR assay. For the identification and discrimination of the B. abortus strains of biovars 1, 2, and 4 from other B. abortus biovars, the S19 vaccine strain, and other species, the B. abortus species-specific (BaSS) PCR assay, also designated AMOS-ERY PCR, was performed following previously described approaches (18, 32).

(e) IS711 fingerprinting. Restriction profiles of the insertion sequence IS711 (EcoRI and EcoRI plus DdeI) were investigated as previously described (33, 34).

(f) MLSA-21. Analysis by extended multilocus sequence analysis (MLSA) of the 21 distinct sequence fragments covering more than 10.2 kb of genome was performed according to previously described procedures (5, 24). Each allele at each locus gives an arbitrary numerical designation (sequence type [ST]). A representative strain of each genotype was used for phylogenetic analysis. Phylogenetic trees were constructed in MEGA (35) with the concatenated sequence data of the 21 loci using the neighbor-joining algorithm and the Jukes-Cantor model.

(g) VNTR-21 assay. The diversity of the isolates highlighted in this study was analyzed by the VNTR-21 method, based on the examination of 21 loci, including some of those described in the HOOF-Prints assay (36), as previously described (23).

(h) MLVA-16 assay. The selected pure strains, as well the B. abortus reference strains, were characterized by MLVA-16, using 16 genetic markers, as previously described (22). Fragment sizes converted to repeat unit numbers were imported into BioNumerics v6.6 as a character data set. The obtained MLVA patterns were compared with the Brucella2012 MLVA database, hosted by University Paris-Sud (Orsay, France) (http://mlva.u-psud.fr/mlvav4/genotyping/). According to the speed of molecular evolution, weights were assigned to the distinct panels (weights of 2, 1, and 0.1 per locus for panel 1, panel 2A, and panel 2B, respectively).

A minimum spanning tree (MST) was constructed to compare the pure strain genomes within the network comprising 714 Brucella isolates of distinct species (MLVA-16 patterns are available in the Brucella2010 database, http://mlva.u-psud.fr/mlvav4/genotyping/). The MST results were shown by using a logarithmic scaling. To compare the strains of interest among other B. abortus isolates, a cluster analysis was performed using the unweighted-pair group method (UPGMA) algorithm with categorical coefficient.

RESULTS

Cultures and biotyping.

Fourteen strains, including the reference strain 63/75, previously identified as B. abortus bv. 7 (in particular due to their A+M+ serotype) were subjected to 3 successive cloning isolations. The characteristics of the Brucella cultures obtained are presented in Table 2. Five strains were found to consist of mixed cultures of different B. abortus biovars or a mixture of different Brucella species (Table 2). Three field strains, isolated in 1984, were found to represent a mixture of B. abortus bv. 1 (A+M−) and bv. 4 (A−M+), while, as expected, the former reference strain 63/75 was not a pure strain but was confirmed as a combination of two B. abortus biovars, biovar 3 (A+M−) and biovar 5 (A−M+), as described previously (14). Furthermore, one strain was characterized as a mixture of different species, B. abortus bv. 1 (A+M−) and B. melitensis bv. 1 (A−M+). In addition, five other strains that had historically been deposited in strain collections as biovar 7 were recharacterized as pure cultures of B. abortus bv. 1, 3, 4, or 5, with characteristics different from those expected for the former biovar 7.

Finally, only four strains, all isolated from cattle, were found to represent B. abortus pure clones with characteristics which did not conform to the B. abortus profiles of any recognized biovar (biovars 1 to 6 and 9): one strain isolated in 1963 in Kenya (07-994-2411, designated K), two strains isolated in Mongolia in 1988 and 1993 (99-9971-135 and 99-9971-159, respectively, designated M1 and M2, respectively), and one strain isolated in Turkey in 2003 (03-4923-239D, designated T). Their particular profile conformed to the differential characteristics of the former biovar 7, particularly agglutination with both anti-A and anti-M monospecific sera (Tables 1 and 2) (37).

Molecular analysis.

Accurate species-specific PCR methods (RT-PCR, Bruce-ladder, AMOS-ERY PCR, omp polymorphism, and IS711 fingerprinting) were performed separately to identify the species to which the four pure strains belonged and to exclude their affiliation with the recognized biovars (biovars 1 to 6 and biovar 9). These assays were complemented by phylogenetic and/or epidemiological methods, such as extended MLST, VNTR-21, and MLVA-16.

The pure strains M1, M2, T, and K, conforming to the former biovar 7, were analyzed by RT-PCR. The cycle threshold (C_T) values indicated strong positive reactions (data not shown). The RT-PCR assay confirmed that these clones belonged to the Brucella genus. Likewise, the obtained Bruce-ladder pattern designated these pure strains M1, M2, T, and K as B. abortus (data not shown). This result was confirmed by omp31 PCR (Fig. 1a), which evidenced the deletion of this gene for the species B. abortus. The omp2a PCR polymorphism (Fig. 1b) confirmed the thionin resistance of the four strains, in agreement with the results of the biotyping (Table 2), showing that these B. abortus strains were not biovar 1, 2, or 4. In parallel, in AMOS-ERY PCR (Fig. 1c), the pattern for the four strains showed uniquely the ery bands (1.2 kb), confirming the genus Brucella and showing that the biovar was different from B. abortus bv. 1, 2, and 4, which show an additional specific 0.5-kb band (Fig. 1c).

FIG 1.

PCR analysis of Brucella reference strains and the 4 field B. abortus bv. 7 strains. Lanes: 1, B. melitensis bv. 1 strain 16M; 2, B. abortus bv. 1 strain 544; 3, B. ovis 63/290; 4, B. suis bv. 1 strain 1330; 5, B. abortus bv. 1 vaccine strain S19; 6, B. abortus bv. 7 field strain T; 7, B. abortus bv. 7 field strain M1; 8, B. abortus bv. 7 field strain M2; 9, B. abortus bv. 7 field strain K; 10, negative control. (a and b) omp31 and omp2a polymorphism, respectively, for strains M1, M2, T, and K investigated in this study and the reference strains (29). (c) Discrimination between the investigated strains and the reference strains as determined by AMOS-ERY PCR assay (18).

With regard to IS711 fingerprinting (data not shown), the isolates M1, M2, and T shared an identical profile by both EcoRI and EcoRI-plus-DdeI digestion. The EcoRI profile was identical to that seen with some biovar 5, 6, and 9 isolates, but the EcoRI-plus-DdeI profile was unique to these three isolates. The restriction profiles of strain K were distinct from the profiles of the other three isolates using both the EcoRI and EcoRI-plus-DdeI approaches. Furthermore, the omp RFLP showed that the same three isolates (M1, M2, and T) shared an identical profile, mirroring the relationship determined by IS711 fingerprinting (data not shown).

The MLSA-21 scheme indicated that the four field isolates belonged to three distinct sequence types (STs) (Fig. 2). The two Mongolian isolates M1 and M2 share an identical ST (ST2), along with many non-African B. abortus bv. 3 strains. Strain T was a member of a distinct ST, ST3. However, genotype ST3 is very closely related to ST2, possessing only one discriminating SNP in gpd in over 10.2 kb of sequence. This ST has not been described in any other isolate to date. In contrast, strain K, which represented the sole member to date of ST37, is distantly related to the above isolates and indeed is rather divergent from most other B. abortus strains. The only other closely related isolate is the sole representative of ST36, also a field isolate from Africa.

FIG 2.

Unrooted phylogenetic reconstruction of relationships between 440 Brucella isolates representing distinct species and biovars by extended MLSA. The tree was constructed with the concatenated sequence data of the 21 loci (>10.2 kb) using the neighbor-joining algorithm with the Jukes-Cantor model. Brucella species are distinguished by different colors. Strains M1, M2, T, and K investigated in this study are designated by arrows (24; Whatmore, unpublished data).

The VNTR-21 genotypes of the M1, M2, T, and K isolates were compared with those of B. abortus reference and vaccine strains (Table 3). Isolates M1 and M2 from Mongolia were closely related and shared an identical profile with isolate T at the six loci used for taxonomic resolution (VNTR14, VNTR21, VNTR27, VNTR24, VNTR7, and VNTR26) (23). In contrast, strain K appeared to be distantly related to these three isolates and possessed a unique profile, not observed in any other isolate to date, at these six loci.

TABLE 3.

VNTR-21 genotypes for the four B. abortus pure strains phenotypically identified with a biovar distinct from biovars 1 to 6 and 9 in comparison with the reference strains

Strain	B. abortus biovar or designation	Genotypea
		Hoof 1	Hoof 2	Hoof 3	Hoof 4	Hoof 5	Hoof 6	Hoof 7	Hoof 8	VNTR2	VNTR16	VNTR17	VNTR5A	VNTR5B	VNTR12A	VNTR12B	VNTR14b	VNTR21b	VNTR27b	VNTR24b	VNTR7b	VNTR26b
544	1	3	4	6	5	2	2	4	2	3	4	3	5	4	2	13	1	2	1	5	2	4
86/8/59	2	7	4	6	2	2	2	11	2	2	3	4	4	5	3	2	1	2	1	5	2	4
Tulya	3	9	4	1	5	7	5	9	2	10	3	6	5	4	2	9	1	2	2	5	2	4
292	4	7	4	5	3	2	2	7	2	2	3	4	4	2	3	5	1	2	1	3	2	4
B3196	5	6	2	8	6	6	5	10, 11	2	2	4	5	7	10	8	5	1	2	1	5	1	4
870	6	9	2	4	10	4	2	6	2	2	4	5	6	9	4	13	1	2	1	5	2	4
C68	9	7	2	8	5	2	5	3	2	2	4	5	6	13	4	9	1	2	1	5	1	4
S19	Vaccine strain	5	4	4	2	2	2	8	2	2	4	4	5	6	5, 6, 7	5	1	2	1	5	2	4
RB51	Vaccine strain	5	4	4	2	2	2	4	2	2	4	4	7	4	3	5	1	2	1	5	2	4
07-994-2411c	K	5	3	4	5	6	4	5	5	6	2	3	2	5	4	2	1	2	1	5	2	3
03-4923-239c	T	9	2	8	4	8	5	8	5	2	5	4	7	4	2	13	1	2	1	1	2	4
99-9971-135c	M1	15	2	1	4	6	4	10	3	2	5	4	6	7	4	5	1	2	1	1	2	4
99-9971-159c	M2	13	2	1	4	6	4	9	3	2	5	4	6	6	4	5	1	2	1	1	2	4

^aHoof data are from reference 36, and VNTR data are from reference 28.

^bLocus useful for taxonomic resolution.

^cStrain investigated in this study.

The MLVA patterns obtained for the M1, M2, T, and K isolates (Table 4) were compared with those of 714 Brucella isolates of distinct species (Fig. 3) and with those of the B. abortus reference and vaccine strains (Fig. 4). The MLVA-16 scheme divided the organisms into clusters corresponding to their taxonomic designations. Isolates M1, M2, and T were closely related to the B. abortus members. Interestingly, isolate K, the MLVA-8 profile of which did not correspond to any described pattern in the Brucella2012 database, seemed to be closer to the B. melitensis bv. 1 cluster (only a 1-U difference in bruce42 with MLVA-8 profile of BCCN87-92, a B. melitensis bv. 1 strain isolated in the United States in 1997). Consequently, the MLVA-16 profile of K was very distant from those for other B. abortus genotypes, while strains M1, M2, and T were closer to the B. abortus bv. 3 strains, as also observed with the MLST assay.

TABLE 4.

MLVA-16 genotypes for the four B. abortus pure strains phenotypically identified with a biovar distinct from biovars 1 to 6 and 9 in comparison with the reference strains

Strain	B. abortus biovar or designation	Host	Yr	Genotypea
				Panel 1								Panel 2A			Panel 2B
				bruce06	bruce08	bruce11	bruce12	bruce42	bruce43	bruce45	bruce55	bruce18	bruce19	bruce21	bruce04	bruce07	bruce09	bruce16	bruce30
544	1	Bovine	1942	4	5	4	12	2	2	3	3	5	21	8	3	5	3	4	5
86/8/59	2	Bovine	1959	4	5	4	12	2	1	3	3	6	21	8	3	4	3	3	5
Tulya	3	Human	1958	3	5	4	11	2	2	3	3	8	20	8	6	5	3	11	5
292	4	Bovine	1961	4	5	4	12	2	2	3	2	6	21	8	3	4	3	3	5
B3196	5	Bovine	1959	3	5	3	12	2	2	2	3	7	21	8	6	7	3	3	3
870	6	Bovine	1959	3	5	3	12	2	2	3	3	7	21	8	3	6	3	3	3
C68	9	Bovine	1958	3	5	3	12	2	2	2	3	7	21	8	6	6	3	3	3
S19	1 (vaccine)	Bovine	1943	4	5	4	12	2	2	3	3	6	21	8	3	5	3	3	5
S99	1		1957	4	5	4	12	2	2	3	3	6	21	8	3	5	3	4	6
07-994-2411b	K	Bovine	1963	2	4	2	12	3	2	3	3	5	22	6	5	2	6	7	5
03-4923-239b	T	Bovine	2003	4	5	3	12	2	2	3	1	6	21	8	6	7	6	3	3
99-9971-135b	M1	Bovine	1988	4	5	3	12	2	2	3	1	6	21	8	5	6	4	3	3
99-9971-159b	M2	Bovine	1993	4	5	3	12	2	2	3	1	6	21	8	5	6	4	3	3

^aThe 16 investigated loci are organized in agreement with their molecular evolution rate in two panels; panel 1 is useful for species identification, and panel 2 shows a higher discriminatory power (22).

^bStrain investigated in this study.

FIG 3.

Minimum spanning tree of 718 Brucella isolates, representing the majority of known species/biovars, by MLVA-16. An MST of 718 Brucella isolates of distinct species and biovars was constructed, based on MLVA-16 patterns obtained in this study or available in the Brucella2010 database (http://mlva.u-psud.fr/mlvav4/genotyping/). The MST results are shown by using a logarithmic scaling. Brucella species and B. melitensis biovars are distinguished by different colors. Strains M1, M2, T, and K investigated in this study are designated by arrows.

FIG 4.

Analysis of relationships between 56 B. abortus isolates, representing all biovars, by MLVA-16. Clustering analysis was performed using the UPGMA algorithm with categorical coefficient; each data set (panel 1, panel 2A, and panel 2B) was weighted according to the molecular evolution rates (22). Reference strain genotypes were obtained both from this study and from the Brucella2012 database. B. abortus biovars are distinguished by different colors. Strains M1, M2, T, and K investigated in this study are designated in red. The strain identity, the host, the geographical origin, the biovar, and the corresponding genotype reference are listed in the dendrogram. The bars reflect the percentage of weighted divergence.

DISCUSSION

In the past decades Brucella taxonomy has been hotly debated and has undergone many reorganizations at both the species and intraspecies levels. Indeed, from 1986, on the basis of results of DNA-DNA hybridization (high similarity values up to 99%) and in agreement with Bergey's Manual of Systematic Bacteriology, some scientists proposed that all the Brucella species should be regarded as belonging to a single species, B. melitensis (14). In 2003, ISTB recommendations allowed a return to the pre-1986 taxonomy of the Brucella genus (classical Brucella species with their recognized biovars) (38), and a number of molecular typing and phylogenetic studies illustrating that the classical species correspond to genetically distinct, if closely related, entities (23, 24, 39) have supported this decision.

In the same way, the presence of B. abortus bv. 7, represented by the reference strain 63/75, in Brucella systematics has been the subject of controversy. Indeed, this biovar was suspended from the Approved Lists of Bacterial Names Brucella classification in 1988 because of unpublished evidence that the reference strain 63/75 was a mixture of B. abortus bv. 3 and bv. 5 (14). To formally confirm and clarify the situation with regard to B. abortus bv. 7, all isolates phenotypically identified in the past as B. abortus bv. 7 in the AHVLA and ANSES strain collections were characterized by classical microbiological and multiple molecular approaches.

Among the 14 investigated collection strains, including the reference strain 63/75, 10 conformed to the previous B. abortus bv. 7 characteristics. However, they either were a mixed culture (different B. abortus biovars or a mixture of distinct Brucella species), conferring a typical A+M+ serological profile or were another pure B. abortus biovar, possibly reflecting the loss of another strain that resulted in the originally described A+M+ serological profile. Only four cattle strains, isolated in Kenya (K) in 1963, in Turkey (T) in 2003, and in Mongolia (M1 and M2) in 1988 and 1993, showed a particular profile, previously described as B. abortus bv. 7, characterized mainly by an A+M+ serological pattern.

The molecular techniques (RT-PCR and IS711 fingerprinting) identified these four field strains as members of the Brucella genus. Furthermore, AMOS PCR, Bruce-ladder multiplex PCR, and the polymorphism of the outer membrane genes omp2a, omp2b, and omp31 confirmed that these strains belonged to the B. abortus species but were distinct from the designated biovars (biovars 1 to 6 and biovar 9) and vaccine strains. The phylogenetic and/or epidemiological schemes supported these results.

In agreement with the above data, strains M1, M2, and T were found to represent a genetic cluster, with the allelic profiles, obtained by VNTR-21 and MLVA-16 assays, allowing discrimination of these strains with a characteristic MLVA-11 pattern, closely related to other B. abortus strains, especially with the B. abortus bv. 3 strains. In addition, phylogenetic multilocus sequence analysis placed the isolates as two closely related sequence types (ST2 and ST3) in the same genetically conserved group of B. abortus strains.

Interestingly, strain K, phenotypically identical to isolates M1, M2, and T, is divergent from them on molecular characterization grounds. Indeed, strain K possessed a distinct and novel profile using the VNTR approaches, adjacent to a B. melitensis bv. 1 member, and a unique MLST genotype, related to an African field isolate. While several molecular strategies (Bruce-ladder, omp31 deletion, and AMOS PCR) allowed placement of strain K among B. abortus strains, this strain is unique among isolates examined to date, and genotyping data revealed that this isolate is rather distant from other B. abortus strains. Phenotypic and molecular discrepancies concerning strain K illustrate the complexities of Brucella taxonomy.

The definition of Brucella species is based only on characteristics of lysotyping and urease and oxidase hydrolysis (40). In agreement with these phenotypic criteria, the four field isolates M1, M2, T, and K might be classified in the species B. abortus. Nevertheless, their profile does not coincide with any designated B. abortus biovars (biovars 1 to 6 and biovar 9), but it matches the criteria of the former biovar 7. In addition, these data are partially congruent with molecular approaches. On the basis of polyphasic taxonomy results, taking into account available phenotypic and genotypic data, we propose that the B. abortus biovar 7 could be reintroduced into the Brucella classification, with as a reference strain the oldest isolate (99-9971-139), isolated in Mongolia in 1988. However, as with the existing B. abortus bv. 3 scenario, where a number of studies have shown that this biovar includes genetically distinct clusters (27, 41), the results here highlight the limitations of a classification system based on phenotype alone. The “clustering” of genetically unrelated strains clearly limits the value of biotyping as a typing or epidemiological tool, as equally does the finding that the ST2 type is shared by many biovar 3 isolates. Such observations strengthen the argument for moving toward a molecularly based classification, which, in the scenario described here, would readily separate the genetically distinct but phenotypically identical strain K.

Conclusion.

The aim of our study was to clarify the situation concerning B. abortus biovar 7 by both conventional and molecular typing approaches. This consensus approach confirmed that worldwide collections could possess some strains misidentified as B. abortus bv. 7, a biovar suspended from the Approved Lists of Bacterial Names Brucella classification, and highlighted the need to verify the taxonomic position of these strains.

Our study allowed the identification of four clonal B. abortus strains among 14 strains investigated, with specific phenotypic criteria corresponding to the criteria defining the former biovar 7. The phenotypic data results were strengthened by molecular strategies. This polyphasic investigation allows us to propose the reintroduction of biovar 7 into the Brucella classification, with three representative strains (M1, M2, and T) and strain 99-9971-135 as the potential reference strain. However, the study also highlighted the existence of (i) genetically unrelated strains sharing the same biovar 7 phenotype and (ii) non-biovar 7 isolates that share the same genotype, which could compromise the reintroduction of this biovar.