Use of 16S rRNA Gene Sequencing for Rapid Confirmatory Identification of Brucella Isolates

Abstract

Members of the genus Brucella are categorized as biothreat agents and pose a hazard for both humans and animals. Current identification methods rely on biochemical tests that may require up to 7 days for results. We sequenced the 16S rRNA genes of 65 Brucella strains along with 17 related strains likely to present a differential diagnostic challenge. All Brucella 16S rRNA gene sequences were determined to be identical and were clearly different from the 17 related strains, suggesting that 16S rRNA gene sequencing is a reliable tool for rapid genus-level identification of Brucella spp. and their differentiation from closely related organisms.

Brucella spp., the causative agents of brucellosis, are pathogenic to a variety of domesticated and wild animals. The genus Brucella is comprised of gram-negative, facultative, intracellular pathogens (1). Phenotypic characteristics, antigenic variation, and prevalence of infection in different animal hosts have resulted in the initial recognition of six species: B. abortus (cattle), B. melitensis (goats/sheep), B. suis (swine), B. canis (dogs), B. ovis (rams), and B. neotomae (desert rats) (13, 27). Recently two Brucella strains from different marine mammals have been reported (7, 11, 22), and the names B. pinnipediae (seal/otter) and B. cetaceae (porpoise/whale) have been proposed (12). DNA hybridization analyses indicate a high level of homology among the brucellae, suggesting that the genus Brucella may comprise only one species with several biovars (40).

Brucellosis impacts public health and agricultural economies worldwide because of its high infectivity rate (13). Brucella spp. have also long been considered a potential biological weapon (20), and currently, with the renewed threat of biological warfare and agricultural terrorism, B. melitensis, B. suis, and B. abortus are listed as category B biothreat agents by the Centers for Disease Control and Prevention Strategic Planning Group (34).

Infections in humans generally result from (i) transmission via the gastrointestinal route by the consumption of unpasteurized diary products (18, 32) and contaminated meat (25), (ii) airborne transmission in animal husbandry by inhaling dust contaminated by aborted tissues (23), and (iii) transmission caused by laboratory-associated exposure to aerosols (26). In the event of a bioterror attack, the preferred method of dissemination would most likely be via aerosol (3, 20, 34).

Because of the limited availability of animal vaccines, cost of animal inoculations, a lack of vaccines for human use (13, 37), and its low infectious dose for humans, human brucellosis is endemic in many parts of the world, including the Mediterranean region, Latin America, Asia, and Africa. The reported incidence varies from <0.01 to >200 per 100,000 population (13). Human brucellosis is rare in the United States, with approximately 100 human cases reported per year, mostly caused by consumption of unpasteurized diary products and, to a lesser degree, occupational exposure (10, 38).

The rapidity with which a laboratory diagnosis can be obtained is an important component of every outbreak investigation, because knowledge about the causative agent plays a pivotal role in implementing appropriate public health decisions in a timely manner (21, 34). At present, diagnosis of human brucellosis depends primarily on isolation of Brucella spp. from blood (43), followed by performing a set of bacteriological tests that allows for reliable identification to the subspecies (biotype) level but that requires up to 7 days for completion (1, 42). Another diagnostic option is the use of serologic assays; however, the specificity of these tests is inconsistent because of cross-reactivity with bacteria closely related to the genus Brucella (29).

Extensive efforts have been expended on the development of molecular diagnostic assays based on amplification of different genomic targets by the PCR for the identification of Brucella spp., as recently reviewed by Bricker (6). Although these molecular approaches appeared to be faster and more sensitive than traditional bacteriological tests, they are not considered to be confirmatory tests due to limited specificity (8, 9, 14). Recently a multiplex system has been developed that is sensitive for Brucella spp. and is able to differentiate between B. melitensis and B. abortus (33). However, discrepant results were observed with some B. abortus isolates. So far, none of these assays has been accepted for common use in diagnostic laboratories.

In collaboration with the Association of Public Health Laboratories, the Centers for Disease Control and Prevention Laboratory Response Network has developed standard algorithms based on traditional bacteriological tests (1) for the identification of Brucella spp. by Laboratory Response Network laboratories. Our current efforts are aimed at developing and validating more rapid methods that will allow confirmatory identification of Brucella spp. from cultures and, subsequently, directly in clinical specimen samples.

We chose to evaluate 16S rRNA gene sequencing to determine its utility for confirmatory identification of Brucella spp. The advantage of this method is that results can be obtained within 1 day as compared to 7 days by traditional microbiological testing. Previous work on other bacteria has indicated that differences in 16S rRNA gene sequences may be useful for subtyping or for the differentiation of virulent subtypes from nonvirulent subtypes (30, 36). Low variability in the 16S rRNA locus has been noted as an impediment in using 16S rRNA gene sequencing to discriminate at the species level (41); recent studies of other biothreat select agents have indicated that even subtle differences in the 16S rRNA gene sequence may be used for differentiating and identifying closely related species, which are often cross-reactive in biochemical identification systems commonly used in diagnostic laboratories (19, 35). In this study, we have primarily focused on the applicability of 16S rRNA gene sequencing as a rapid confirmatory identification tool and consequently have sequenced the near-full-length 16S rRNA genes of a reference set of Brucella isolates that represent the species commonly encountered in animal and human infections. In addition, we sequenced 17 closely related diagnostically challenging strains.

MATERIALS AND METHODS

Bacterial strains.

In this study, 65 Brucella strains (representing six species) from a collection of over 600 were selected to represent temporal, geographic, and source diversity (Table 1). A panel of 17 related strains likely to present a differential diagnostic challenge was also selected for comparison (Table 2). All strains were stored at −70°C in defibrinated rabbit blood until testing. Identification of all strains was carried out by standard microbiological procedures as described previously (42).

TABLE 1.

Designations of 65 Brucella isolates analyzed in this study

Identifier (n)	Other identifier	GenBank accession no.	Origina
B. abortus biovar 1 (1)
1995009620	G9400	AY513568	Human, United States, 1995
B. abortus biovar 4 (9)
2000031282	U.S. strain 19	AY513567	United States
2000031287	G8108	AY513566	Human, United States, 2000
2000032001	ATCC 27565	AY513565	Cow, United States, 1989
2000032850	A8898	AY513564	Human, United States, 1967
2002013039	E7789	AY513563	Human, United States, 1980
2002013040	F6650	AY513562	Human, United States, 1985
2002013041	G4321	AY513561	Human, United States, 1990
2002013043	F7752	AY513560	Human, United States, 1986
2002034574		AY513559	Human, United States, 2002
B. canis (16)
1995011487	G9398	AY513521	Human, United States, 1995
1997019052	H0263	AY513519	Human, United States, 1995
2000032857	C6328	AY513516	Human, United States, 1973
2000032858	C6744	AY513515	Dog, United States, 1973
2000032864	B3634	AY513514	Dog, United States, 1969
2000032900	D1408	AY513513	Human, United States, 1974
1982065260	F3081	AY513523	Human, United States, 1982
1982065472	F3106	AY513522	Human, United States, 1982
1996034585	H0078	AY513520	Human, United States, 1996
1998034421	H0801	AY513518	Human, United States, 1998
2000031290	ATCC 23365	AY513517	Dog, United States, 1967
2000032901	D2773	AY513512	Dog, United States, 1995
2000032906	D8615	AY513511	United States, 1997
2000032907	D9880	AY513510	Dog, United States, 1997
2002009962		AY513509	Human, United States, 2002
2002721534	A6806	AY513508	Dog, United States, 1967
B. melitensis biovar 1 (7)
1995032533	G9653	AY513532	Human, United States, 1995
1996013690	G9807	AY513531	Human, United States, 1996
1996034727	H0088	AY513530	Human, United States, 1996
1997030040	H0429	AY513529	Human, United States, 1997
1998034324	H0794	AY513528	Human, United States, 1998
1999012120	H0966	AY513527	Human, United States, 1990
2000031283	ATCC 23456	AY594215	Type strain, United States
B. melitensis biovar 2 (5)
2000031298	H0227	AY513535	Human, United States, 1997
1995013840	G9423	AY513537	Human, United States, 1995
1998034639	H0810	AY513536	Human, United States, 1998
2000042326	H1696	AY513533	Human, United States, 2000
2001025901	H1982	AY513534	Human, United States, 2001
B. melitensis biovar 3 (13)
2000027708	H1481	AY513540	Human, United States, 2000
2000032851	A8915	AY513539	Human, United States, 1967
1976069495	D5927	AY513550	Human, United States, 1976
1979030625	E4026	AY513549	Human, United States, 1979
1994041749		AY513548	Human, United States, 1994
1994042277	G9226	AY513547	Human, United States, 1994
1995019300	G9541	AY513546	Human, United States, 1995
1995019540	G9559	AY513545	Human, United States, 1995
1995019574	G9560	AY513544	Human, United States, 1995
1995030467	G9604	AY513543	Human, United States, 1995
1996001118	G9715	AY513542	Human, United States, 1996
1998015411	H0620	AY513541	Human, United States, 1998
2002020524		AY513538	Human, United States, 2002
B. neotomae (1)
2002721533	ATCC 23459	AY594216	Type strain, United States
B. ovis (3)
2002013048	ATCC 25840	AY513526	Type strain, United States, 1953
2002013049	G7864	AY513525	Ram, New Zealand, 1992
2003007064	KC354	AY513524	Ram, United States, 1957
B. suis biovar 1 (8)
1981096256	H0129	AY513557	Human, United States, 1981
1973034528	C4588	AY513558	Human, United States, 1972
1994037280	G9129	AY513556	Human, United States, 1994
1996028458	G9992	AY513555	Human, United States, 1996
1996028782	H0019	AY513554	Human, United States, 1996
1997000956	H0129	AY513553	Human, United States, 1997
1999043423	H1246	AY513552	Human, United States, 1999
2000027958	H1498	AY513551	Human, United States, 2000
B. suis biovar 4 (2)
1997003632	H0134	AY513506	Human, United States, 1997
1996028684	H0012	AY513507	Human, United States, 1996

^aThe source and temporal origin of the isolate is given when available.

TABLE 2.

Designations of strains closely related to Brucella species

Species (n)	% Sequence similarity	Identifier	Other identifier	GenBank accession no.
Agrobacterium spp. (4)	100
Agrobacterium tumefaciens		2002000903	H2110	AY513489
Agrobacterium radiobacter		2003015367	H2549	AY513491
		2003018195	H2587	AY513490
		2001025242	H1961	AY513492
Ochrobactrum anthropi 5D (3)	100
		1999045026	H1269	AY513495
		2000037232	H1635	AY513494
		2002030421	H2323	AY513493
Oligella urethralis (4)	99.9
		2001034128	H2074	AY513496
		2002020954	H2223	AY513497
		2001034309	H2079	AY513498
		2003015904	H2576	AY513499
Vibrio cholerae (2)	100
		2002734018	VC13-Inaba	AY513501
		2002734019	VC12-Ogawa	AY513500
Others
Afipia broomeaa		1998029722	H0781	AY513505
Afipia felis		2002033830	H2373	AY513503
Bartonella henselaeb			882-ANT5	AY513504
E. coli		2002734020	13535	AY513502

^aThe level of similarity is based on BESTFIT analysis of the two most divergent isolates based on DISTANCES results in the GCG Wisconsin package.

^bLee et al (24).

Amplification of 16S rRNA genes.

Bacteria were grown by plating one loop (1 μl) of stock cell suspension on Trypticase soy agar with 5% defibrinated sheep blood agar (BBL Microbiology Systems, Cockeysville, Md.) and incubating the bacteria aerobically for 1 to 2 days at 37°C with 5% CO₂. To yield a DNA template, a single colony was suspended in 200 μl of 10 mM Tris (pH 8.0) in a 1.5-ml Millipore 0.22-μm-pore-diameter filter unit (Millipore, Bedford, Mass.) and heated at 95°C for 30 min. The filter unit was then centrifuged at 6,000 × g for 5 min to recover the filtrate containing the DNA template. Each final PCR (100 μl) contained 5 U of Expand DNA polymerase (Roche, Indianapolis, Ind.), 2 μl of DNA template, 10 mM Tris-HCl (pH 8.0), 50 mM KCl, 1.5 mM MgCl₂, 200 μM (each) dATP, dCTP, dGTP, and dTTP, and 0.4 μM each of eubacterial primers F8 (5′ AGTTTGATCCTGGCTCAG 3′) and R1492 (5′ ACCTTGTTACGACTT 3′) (17). Reaction mixtures were first incubated for 5 min at 95°C. Then 35 cycles were performed as follows: 15 s at 94°C, 15 s at 50°C, and 1 min 30 s at 72°C. Reaction mixtures were then incubated at 72°C for an additional 5 min. PCR products were purified with a Qiaquick PCR purification kit (Qiagen, Valencia, Calif.).

16S rRNA gene sequencing.

Sequencing primers were chosen from a panel of eubacterial primers: F8, R1492 (described above), F357 (5′ TACGGGAGGCAGCAG 3′), R357 (5′ CTGCTGCCTCCCGTA 3′), F530 (5′ CAGCAGCCGCGGTAATAC 3′), R530 (5′ GTATTACCGCGGCTGCTG 3′), F790 (5′ ATTAGATACCCTGGTAG 3′), R790 (5′ CTACCAGGGTATCTAAT 3′), F1068 (5′ GTCGTCAGCTCGTGTCGTGAG 3′), F1083 (5′ CGTGACATGTTGGGTTAAGTC 3′), F981 (5′ CCCGCAACGAGCGCAACCC 3′), and R981 (5′ GGGTTGCGCTCGTTGCGGG 3′) (17, 28, 35). For non-Brucella strains sequenced for comparison, four additional eubacterial primers were used: R1333 (5′ CTAGCGATTCCGACTTCATGC 3′), R180 (5′ TCTCTCAAGACGTATGCGGTA 3′), R591 (5′ CATCCTGCTTAAGTAACCGTC 3′), and F1127 (5′ ATTAGTTGCCATCATTCAGTT 3′). Sequencing was performed with an Applied Biosystems (ABI, Foster City, Calif.) BigDye terminator cycle sequencing kit (version 2.0, 1.1, or 1.0) per the manufacturer's instructions, with the exception of using 6 μl of BigDye instead of 8 μl. Sequencing products were purified by using Centri-Sep spin columns (Princeton Separations, Adelphia, N.J.), and these sequencing products were resolved with an ABI model 3100 automated DNA sequencing system (ABI).

Computer analysis of 16S rRNA gene sequences. (i) Evaluation of the 16S rRNA gene sequences of Brucella spp.

The 16S rRNA gene sequencing of the amplified products obtained with PCR primers F8 and R1492 generated 1,412-bp-long sequences that closely matched the size of previously published Brucella 16S rRNA gene sequences. The length of these sequences was close to the full length of the gene, which is 1,485 bp (5, 7, 16), and thus allowed direct comparisons.

The software package used for all data analysis was the Accelrys, Inc. (San Diego, Calif.), GCG Wisconsin package, version 10.2. The raw trace files from the ABI 3100 sequencer were visually examined, aligned, and edited (SEQMERGE).

We compared the Brucella sequences by using DISTANCES to determine the level of similarity among them. The sequences generated in this study were compared with previously published Brucella spp. sequences by using BESTFIT to determine levels of similarity between sequences.

(ii) Comparisons of 16S rRNA gene sequences of Brucella spp. to sequences of closely related strains.

The eubacterial primers F8 and R1492 were also used for the panel of related strains likely to present a differential diagnostic challenge. The 16S rRNA gene sequences from the amplification products that resulted from F8 and R1492 varied in length (1,377 to 1,491 bp). For the Agrobacterium, Ochrobactrum anthropi, Oligella urethralis, and Vibrio cholerae strains, the sequences were organized into sets by genus (Table 2) and aligned (PILEUP). A core segment of 1.4 to 1.5 kb was selected for each that was common to all the sequences in a given set by trimming sequences at the 5′ and 3′ ends. The level of similarity of the sequences in a set was then determined by using Jukes-Cantor correction (DISTANCES). The sequences within the Agrobacterium set matched each other 100%, as did the sequences within the O. anthropi and V. cholerae sets (Table 2). The most divergent sequences within the O. urethralis set matched at 99.9% (Table 2). A representative sequence from each set, as well as the 16S rRNA gene sequences from Escherichia coli O157:H7, Bartonella henselae, Afipia broomeae, and Afipia felis, were then compared to the Brucella consensus sequence to determine the level of similarity by BESTFIT (Table 3).

TABLE 3.

Similarities of Brucella 16S rRNA consensus gene sequence to 16S rRNA gene sequences of strains related to Brucella spp. and species important for diagnostic differentiation

Isolate	% Similarity of 16S rRNA gene sequence to Brucella consensus sequencea	Length of sequence used for analysis (bp)
Ochrobactrum anthropi 5D (1999045026)	98.8	1,427
Bartonella henselae (882-ANT5)	94.9	1,425
Agrobacterium tumefaciens (2002000903)	94.2	1,430
Afipia felis (2002033830)	91.0	1,377
Afipia broomeae (1998029722)	89.9	1,377
E. coli O157:H7 (2002734020)	81.8	1,452
Vibrio cholerae (2002734018)	81.5	1,491
Oligella urethralis (2001034128)	80.3	1,457

^aSequence similarity is based on BESTFIT analysis in the GCG Wisconsin package.

Nucleotide sequence accession number.

A total of 82 16S rRNA gene sequences were determined in this study (Tables 1 and 2). They have been deposited in GenBank under accession no. AY513489 for Agrobacterium tumefaciens; AY513490 to AY513492 for Agrobacterium radiobacter; AY513493 to AY513495 for O. anthropi 5D; AY513496 to AY513499 for O. urethralis; AY513500 to AY5135501 for V. cholerae; AY513502 for E. coli; AY513503 for A. felis; AY513504 for B. henselae; AY513505 for A. broomeae; and AY513506 to AY513568, AY594215, and AY594216 for Brucella spp.

RESULTS

Near-full-length 1,412-bp nucleotide sequences of the 16S rRNA genes from 65 Brucella strains (Table 1) comprising six Brucella spp. were generated. Upon alignment and comparison, all 16S rRNA gene sequences were determined to be identical. This result yielded a Brucella consensus sequence that was used for all comparisons.

A BLAST search (10 October 2003) on GenBank indicated that there were only 17 Brucella 16S rRNA gene sequences (from 10 strains) near 1.4 kb available in this public database (2). Eleven of these gene sequences were a 100% match to our Brucella consensus sequence. Six of these 11 sequences were from B. melitensis strain 16 M and B. suis strain 1330, for which full genome sequences were published in 2002 (15, 31). Both strains have two chromosomes, and each strain has three copies of the 16S rRNA gene with two copies on chromosome I and one copy on chromosome II (15, 31). A BLAST analysis (8 October 2003) of both genome sequences on The Institute for Genomic Research website (http://tigrblast.tigr.org/cmr-BLAST/) (2) revealed that all three copies from each strain were a 100% match to our Brucella 16S rRNA gene consensus sequence. Two additional 16S rRNA sequences that were a 100% match to our Brucella consensus sequence are those of the recently published marine Brucella strain M2357/93 and B. abortus strain 544 (GenBank accession no. AF091353 and AF091354) (7). Finally, the last three 16S rRNA gene sequences of the 11 are from a study published in 2000 that also used the B. melitensis strain 16 M (accession no. AF220149, AF220148, and AF220147) (5).

There were six near-full-length Brucella 16S rRNA gene sequences that were not 100% matches to the consensus sequence. These sequences were entered into GenBank between 1989 and 1995. The first one is that of B. ovis strain ATCC 25840 (accession no. L26168) (K. H. Wilson, unpublished data), which differs from the consensus sequence by 3 bp. Compared to the consensus sequence, this B. ovis sequence has an A instead of a G at position 600, as well as Ns at positions 426 and 700. We sequenced three B. ovis isolates, including the ATCC 25840 strain, and all three sequences were a 100% match to the Brucella consensus sequence. The Brucella consensus sequence also differs substantially from the 16S rRNA gene sequence generated in 1989 for B. abortus strain 11/19 (accession no. X13695) (16). A BESTFIT analysis indicates that there is a 19-bp difference, including an N as well as a 3-bp deletion compared to our Brucella consensus sequence. A BLAST query on GenBank indicates that the sequence of accession no. X13695 is unique and does not perfectly match any other 16S rRNA gene sequence in the public database. The remaining four Brucella 16S rRNA gene sequences in GenBank are from B. suis strain ATCC 23444, B. melitensis strain ATCC 23456, B. neotomae strain ATCC 23459 (accession no. L26169, L26166, and L26167, respectively) (Wilson, unpublished), and B. canis strain ATCC 23365 (accession no. L37584) (K. H. Wilson and H. G. Hills, unpublished data), which differed from the consensus sequence by 1 to 6 bp. Three of these four sequences had Ns in the sequences, indicating an inability to identify bases.

B. abortus strain 11/19 was not available for resequencing: however, this strain is the same as U.S. strain 19 (E. Moreno, personal communication). We sequenced the 16S rRNA genes from B. abortus U.S. strain 19, B. melitensis strain ATCC 23456, B. neotomae strain ATCC 23459, and B. canis strain ATCC 23365. All four sequences were a 100% match to the Brucella consensus sequence.

Comparisons with the panel of 17 related strains likely to present a differential diagnostic challenge indicated that the sequence most similar to the Brucella consensus sequence was that from an O. anthropi strain at 98.8% similarity (Table 3), which is consistent with the relationship seen in previous phylogenetic studies (39). There is a difference of 16 bp, as well as a deletion and an insertion, which easily differentiates the Brucella consensus sequence from this sequence. The other 16S rRNA gene sequences from this panel of strains had even more pronounced differences (Table 3).

DISCUSSION

Current confirmatory identification of Brucella isolates relies upon a set of traditional microbiological tests that requires up to 7 days for completion. At the same time, safety concerns have limited the wide application of automated identification systems, which could potentially shorten the time from collection of an unknown specimen to obtaining the laboratory diagnosis. Few evaluations of such systems for Brucella identification have been made, but all suggest that the performance of automated systems has generally been poor (D. E. Roe, J. M. Janda, D. Lindquist, N. Caton, C. Crandall, J. Wong, and B. Zimmer, Abstr. 101st Gen. Meet. Am. Soc. Microbiol., abstr. C-340, 2001; J. C. David, W. L. Thomas, R. J. Burgess, and T. L. Hadfield, Abstr. 101st Gen. Meet. Am. Soc. Microbiol., abstr. C-335, 2001). Consequently, because Brucella spp. typically require several days of incubation before attaining a level of growth sufficient for the interpretation of various phenotypic tests (1), a more rapid approach for identification is needed, especially for a timely response to a potential biothreat event.

The critical advantage of 16S rRNA gene sequencing is based on its ability to be used for identification of unknown isolates. By using the broad-range eubacterial primers, as demonstrated in this study, a 16S rRNA gene sequence can be readily obtained, allowing for rapid confirmatory identification of an unknown isolate as a Brucella species by simple comparison to the consensus sequence. Confirmatory identification can then serve as a guide to continue all subsequent work with biosafety level 3 practices, while the rapidity with which it is obtained may be the critical factor needed for quickly implementing control and preventative measures in an emergency biothreat situation (6, 34). The PCR tests previously proposed limit the laboratorian to either positively identifying gene regions associated with a Brucella species or ruling out Brucella spp., which would then require subsequent diagnostic tests to identify the etiologic agent. Sequencing the 16S rRNA gene enables querying a database such as GenBank to yield an alternative identification without the need for any additional laboratory work. Consequently, given that the clinical presentation of brucellosis can resemble diseases caused by other select agents, this global diagnostic approach will be invaluable in a potential biothreat setting. We have added to the robustness of this approach by submitting to GenBank 82 high-quality 16S rRNA gene sequences from Brucella spp. and close relatives, which allows for a more accurate comparison of sequences.

One additional advantage of using 16S rRNA gene sequencing is that this method may be able to identify the causative organism from clinical specimens of patients who have received antimicrobial agents prior to the collection of specimens, as has been demonstrated recently during the 2001 bioterrorism-associated anthrax investigation (35). Because antimicrobial and other supportive treatment will not differ regardless of the Brucella species causing the illness (4), detection of the Brucella-specific 16S rRNA gene sequence will be advantageous for timely treatment of patients even without species identification.

Our result of 100% identity of the 16S rRNA gene sequence among the large number of Brucella strains tested is consistent with previous studies using a limited number of strains (5, 7) and could also be used to support proposals that Brucella may be a monospecific genus, based on the DNA hybridization studies that indicate a >77% level of homogeneity for the Brucella spp. (40). Vizcaíno et al. observed that there was some variability in the 16S rRNA locus, but not enough to separate the Brucella species (41). Similarly, there are some Brucella 16S rRNA gene sequences in GenBank that differ from the consensus sequence. We resequenced all but one of these discrepant GenBank submissions. Each was a 100% match to the Brucella consensus sequence. We do not believe the differences in the sequences from the original GenBank entries represent microheterogeneity in the 16S rRNA genes of these Brucella strains. Rather, these discrepant sequences were submitted from 1989 to 1995 before the advent of automated sequencing technologies with higher fidelity. This is supported by the presence of unresolved bases in some of these discrepant sequences.

A comparison of the Brucella 16S rRNA consensus sequence with those of close relatives indicates that sequencing of the first 500 bp is sufficient to differentiate them. The closest match, O. anthropi, was different by 2 bp in this segment, which is a sufficient level of divergence to be detected by current automated sequencing technology. For laboratories that may wish to conserve resources, this may be a viable option compared to sequencing 1.4 kb.

This work demonstrates that 16S rRNA gene sequencing can provide rapid confirmatory identification of an isolate as a Brucella species, allowing prompt and appropriate public health responses to be implemented. This approach will be of even greater value if shown to be effective when used directly on clinical specimens; studies are under way in our laboratory to evaluate its sensitivity and specificity on clinical specimens of patients with culture-confirmed brucellosis.