Brucellosis remains one of the world's major zoonotic disease problems and can lead to reproductive problems in a number of large livestock and other animals. In humans, brucellosis is associated with a broad spectrum of symptoms and can occasionally be fatal [1]. Whilst brucellosis is the global term for infection by organisms in the Gram-negative genus Brucella, there are actually a number of causative agents. Differentiation to species level within the genus is based on both different phenotypic characteristics and the host from which it was initially isolated [2]. These divisions are also supported by pulse field gel electrophoresis (PFGE)-based physical mapping, showing that each species has a specific physical organisation [3], and techniques such as multilocus sequencing [4] and variable number tandem repeat (VNTR) based typing [5, 6]. Bovine infection is generally caused by B. abortus, B. melitensis infects predominantly sheep and goats, B. suis infects pigs, hares and reindeer, B. ovis infects sheep only, B. canis infects dogs, whilst B. neotomae has only been isolated from the desert wood rat. In addition to the six classically described species novel members of the genus have more recently been isolated from marine mammals, predominantly dolphins, porpoises and seals. These have very recently been classified as two additional species, B. ceti (cetaceans) and B. pinnipedialis (pinnipeds) [7]. In addition, a novel species of Brucella (named as Brucella microti), associated with voles, has recently been reported [8, 9]. Some Brucella species are further sub-divided into groups known as biovars. Whilst these are of epidemiological interest in B. abortus and B. melitensis, the five biovars that are found in B. suis appear to relate to host preference. In humans, four of the classical species (not B. ovis or B. neotomae) have been shown to cause disease [2]. There is also evidence to suggest that humans can be infected by marine mammal Brucella further highlighting the importance of this group of organisms to human health [10–12].

Identification to species level is traditionally based on biotyping that examines antigenic, phenotypic and phage susceptibility profiles. While the approach has been useful for many years, it has a number of recognised drawbacks. Being culture based, biotyping requires specialist handling facilities for the initial isolation of a pure culture and then needs additional culture time for phenotype and phage susceptibility determination. In addition, some of the differentiating characteristics can be very subtle meaning that biotyping should only be carried out by highly experienced individuals. However, even with this caveat, interpretation of this method is acknowledged to be rather subjective. Furthermore there are frequent reports of atypical organisms that do not conform to the expected characteristics of particular species.

Molecular biology provides tools that can circumvent such problems associated with classical microbiology by using suitably validated genetic markers for objective and unambiguous identification of groups. As DNA is the target of this testing, culturing can be reduced or potentially bypassed, both removing the necessity for specialist handling facilities, and shortening time from sampling to identification. With the explosion of molecular sequence data and an expanding array of technological platforms the use of assays based on single nucleotide polymorphisms (SNPs) in microbial genotyping is becoming increasingly common. For example, point mutations in rpo B and kat G have been used as markers for resistance to rifampicin and isoniazid in Mycobacterium tuberculosis [13] while SNPs have also been used to differentiate vaccine and field strains of canine parvovirus [14]. In terms of microbial diagnostics SNP-based genotyping has been applied to a number of organisms including members of the Mycobacterium tuberculosis complex [15], Bacillus anthracis [16] and Burkholderia species. [17].

We have recently undertaken extensive phylogenetic analysis of the Brucella group involving the sequencing of some 21 distinct gene fragments, from over 400 strains of Brucella, representing the known genetic diversity of the group [4, M.S. Koylass and A.M. Whatmore, unpublished data]. While this extensive analysis confirmed the long-standing view that the genus is highly homogeneous [18–20] it has nevertheless allowed us to confidently identify SNPs that we believe define each of the classically accepted Brucella species. Furthermore SNPs are particularly powerful markers in such a conserved group as they are highly unlikely to have occurred twice independently (i.e. in distinct evolutionary branches) and equally, having occurred, are unlikely to mutate again back to their ancestral state [21]. In addition, these studies revealed strong linkage disequilibrium in the distribution of alleles between Brucella species [4] implying that recombination is rare and giving increased confidence in the robustness and stability of species-specific SNPs selected as markers.

We have thus selected seven SNPs that we have worked into a multiple outcome assay to identify Brucella isolates at the species level. An assay based on these markers should deliver unambiguous results without the subjectivity associated with biotyping or problems associated with strains that are phenotypically aberrant. We recently described the development of a SNP-based assay identifying Brucella to the species level on a primer extension platform [22]. Here, we describe the development of an equivalent tool for the real-time PCR platform that has certain advantages in terms of speed, technical simplicity and potentially wider applicability being based on a cheaper, and more widely available, technological platform. We describe the use of a panel of short Taqman 5' labelled, 3' Minor Groove Binding (MGB), probes to interrogate species defining SNPs. Use of the MGB protein raises the melting temperature of probes meaning that a single base mismatch causes more destabilisation than would be the case with a longer probe [23]. This facilitates accurate SNP detection allowing us to use this approach in an assay to identify Brucella to the species level.

The aim of this work was to use the robust phylogenetic framework provided by existing MLSA studies to develop a rapid, unambiguous assay for the real-time PCR platform capable of identifying Brucella isolates to species level. The approach is based on a series of discrimination assays interrogating SNPs that we have shown to be specific to a particular Brucella species. As each individual SNP assay will always give one of two results, species determination is straightforward and unambiguous. Amplification profiles, representing binding of either the VIC- or FAM-labelled probe, are generated in all seven individual discrimination assays. One of the seven individual assays should give a positive reaction with the VIC-labelled probe (or two in the case of B. canis), with a reaction with the FAM-labelled probe apparent in all the remaining assays. An assay which did not generate a positive PCR result from each of the seven individual probe pairs would be considered to have failed. This approach avoids any danger of 'false-negatives', where lack of a PCR product could be indicative of PCR inhibition rather than absence of target. Furthermore, although any signals generated by non-Brucella organisms in this study were weak, we sought to absolutely ensure the specificity of the assay by identifying three SNPs in 16S rRNA that, when used in conjunction, differentiate Brucella from closely related bacteria. Allele discrimination assays based on these three markers were shown to clearly differentiate Brucella from closely related α-proteobacteria examined.

As a SNP based speciation assay the approach described here has a number of potential advantages, comparable to those of our primer extension based SNP typing assay [22], when compared to classical biotyping approaches. Both methods overcome the subjectivity associated with biotyping with approaches that are relatively technically straightforward, unambiguous, and have substantial advantages in terms of simplicity and speed. In addition, both techniques reduce the potential exposure to live Brucella. This is desirable given the ease of acquisition of Brucella in the absence of stringent bio-containment facilities [35, 36].

Both assays also have substantial advantages over currently used molecular assays. They can be carried out on crude bacterial lysates bypassing the need for lengthy DNA extraction procedures as required by some molecular typing approaches [20, 37] and, as PCR based methods, both techniques should theoretically be applicable directly to field material obviating the need for any culture. In contrast to some extensively used existing assays such as AMOS-PCR [30] these assays are all encompassing identifying all relevant biovars of all species. Of the classically recognised Brucella species and biovars only B. suis bv 5 will not be positively identified as a particular Brucella species by the real-time PCR assay, reflecting its unique phylogenetic position within the group apparently distinct from other B. suis. However, like other potentially novel Brucella groups, such isolates are positively identified as Brucella displaying seven clear non species-specific FAM signals. Both techniques lend themselves to future expansion to include markers for such new groups. For example, after this manuscript was submitted a novel species, B. microti, was described. We have already described a SNP specific to this species based on MLSA [9] and probes interrogating this SNP could readily be incorporated into this assay. This capacity for rapid inclusion of newly identified species is a potential advantage over some other molecular assays such as the multiplex PCR assay proposed by Garcia-Yoldi et al [38]. Both the primer extension assay and the real-time PCR based assay described here have the advantage of potentially increased sensitivity associated with PCR based techniques. We have shown that the real-time PCR assay described here can detect and differentiate as little as 15 genome equivalents suggesting that it could possibly identify the organism in relevant field material such as foetal tissue or vaginal swabs.

There are, however, substantial advantages in the use of a real-time PCR platform, compared to the primer extension format that led us to investigate and report this alternative approach. The primer extension approach requires a DNA sequencer, while the approach described in this manuscript requires only a real-time PCR machine. Indeed, it could feasibly be undertaken in a simpler format with a conventional thermocycler and a basic fluorescence plate reader. Thus, while the primer extension assay might be appropriate in larger reference laboratories, the technique described in this manuscript is likely to prove far more applicable in routine diagnostics. This is particularly so as real-time PCR machines become more widely available in areas of the world where brucellosis remains a major public and animal health issue. In contrast, the uptake of sequencing based technologies, requiring a substantially greater capital input, is likely to be much slower in these areas. Furthermore, the real-time PCR approach is also technically simpler, faster and cheaper than the primer extension based approach. As amplification and detection is carried out concurrently the real-time assay described here can characterise isolates in around two hours. This compares to the day required for the primer extension based approach that also involves a much less straightforward experimental procedure. Although both primer extension and real-time assays could be further expanded the number of markers that can be included in a single primer extension assay is limited by the need to space markers (ideally by at least 5 bp) between the 13 bp and 88 bp size standards. There is no such limitation with the real-time PCR described here where additional markers can be interrogated simply by the addition of an additional probe pair in another well of a 96 well plate. This will make the latter assay our approach of choice for incorporation of additional markers to type at the sub-species level. Finally, the primer extension assay we described did not include a specific marker for B. suis, with this species being recognised by process of elimination (i.e. lack of any of the markers specific for the other classical Brucella species). While this was appropriate at the time, the recent description of a novel species of Brucella (B. microti) indicates the added value of including a specific marker for each species. The inclusion of a marker for B. suis in the real-time assay ensures that all species are identified 'positively' by the recognition of a SNP shown, by the extensive population genetic analysis that supports this work, to be specific for the B. suis/canis group.

We have described a simple and unambiguous multiple outcome SNP assay based on a robust phylogenetic framework that can characterise Brucella isolates to the species level. The approach is based on a sensitive and rapid real-time PCR platform that will be widely applicable in diagnostic laboratories and is readily expandable as knowledge of the genus increases. Further work will focus on the expansion of the assay to include markers to discriminate live vaccine strains and to type at the sub-species level.