Brucella is a bacterial genus responsible for brucellosis, a disease in animals causing infertility, pre-term birth, or abortion (Moreno and Moriyón, 2006). It is also one of the most worldwide spread bacterial zoonosis, not only causing human suffering but also representing a significant economic burden on animal industries. Because this severe and debilitating disease has not been adequately addressed at some national and international level, it is considered by WHO (2014) as one of the “forgotten neglected zoonosis,” constituting a major burden for poor rural communities.

The main etiological agent of cattle brucellosis is B. abortus associated with abortion, infertility, reproductive failure, and decreased milk production (Harmon et al., 1989; Xavier et al., 2009; Neta et al., 2010). Humans usually become infected when in contact with infected animals or their derived products, particularly non-pasteurized dairy products (Spink, 1956). Several B. abortus reference strains have been described and used as models to understand Brucella pathogenesis or as challenge strains for vaccine testing (Meyer and Morgan, 1973; Chain et al., 2005; Halling et al., 2005; Singh et al., 2015).

Genetic drift causing loss of virulence or antigenic properties has been reported in several Brucella vaccines strains (Bosseray, 1991; Mukherjee et al., 2005; Jacob et al., 2006; Miranda et al., 2013), but very little is known on the genetic stability of reference strains.

Brucella abortus strain 2308 was originally described as a highly virulent strain recovered in 1940 from an aborted fetus of a cow which had been in contact with cattle experimentally infected with a mixture of B. abortus cultures (Jones et al., 1965). Since then it has been widely used as a reference and challenge strain within the brucellosis research community (Trant et al., 2010; Dabral et al., 2014; Miranda et al., 2015; Olsen et al., 2015; Truong et al., 2016; Zhu et al., 2016). Whole Genome Sequencing (WGS) of this strain was first carried out by Chain et al. (2005), who found conservation in chromosome synteny with other sequenced B. abortus genomes and also some differences that were suggested to be strain specific. In 2009, the results of a WGS analysis from a strain named B. abortus 2308A were publically available under accession number GCA_000182625.1. No additional information regarding the isolates used in these two WGS projects was provided.

We performed WGS, functional annotation and manual curation of reference strain B. abortus 2308 kept at the Tropical Disease Research Program, Veterinary School, National University in Costa Rica and compared it with these two published B. abortus 2308 genomes. Significant differences were found among the genomes, challenging the idea of reference strains as a non-changing entities in time and among laboratories. The use and communication of standardized quality control and experimental design protocols could help interpretation and follow up of reported results.

For comparative analysis, we performed WGS analysis of the 2308 strain that has been used in our laboratory and referred here as B. abortus 2308W, to distinguish its genome from previously reported ones (2308 and 2308A). The genome consists of two chromosomes, chromosome I is 2.10 kb in size and chromosome II is 1.16 kb. Automated annotation and manual curation were performed and summarized in Supplementary Table S1. An editable spreadsheet was also created in order to facilitate updates in the annotation or additional relevant comments (e.g., virulence, function, mutants availability) and is available through a link at https://en.wikipedia.org/wiki/Brucella#Genomics.

A BLAST comparison of this genome with those two previously reported showed major differences summarized in Figure 1 and detailed in Supplementary Table S2. Three major deletions in chromosome I in B. abortus 2308W were detected relative to B. abortus 2308 and 2308A of 1.13, 3.4, and 5.7 kb. It is important to point out that these three deletions are located at the end of contigs and surrounded by repetitive sequences so mayrepresent incompletely assembled contiguous sequences, rather than genuine deletions. Regions containing repetitive sequences are proven difficult to assemble and sequence, regardless the sequencing technology used (Bidmos and Bayliss, 2014). The 1.13 kb deletion contains the 2308 loci BAB1_0934 to BAB1_0937 associated to IS711 elements and transposases. The 3.4 kb deletion includes loci BAB1_1102–BAB1_1104; the first two CDSs are predicted proteins of unknown function and the third one is predicted to be a site-specific recombinase, DNA invertase. These loci fall within a larger 8.1 kb genomic island named GI-1, encoding mainly predicted proteins and phage-related proteins (Rajashekara et al., 2004). Five CDSs are part of the 5.7 kb region absent in chromosome I of 2308W (BAB1_2221–BAB1_2225). They encode tRNAs and rRNAs genes with copies elsewhere in the genome (Supplementary Table S2). These copies are represented by a higher number of reads as compared with average for the rest of the CDS (9700 versus 3200 reads), suggesting that contig breaks are caused by mis-assembly of the repeated copies of the tRNA and rRNA genes during the assembly.

An insertion of a 5.2 kb segment in chromosome II was observed in 2308W and 2308A, relative to 2308 (Chain et al., 2005). This region displayed even read coverage as compared to the rest of the 2308W genome. It includes four genes encoding sugar binding proteins, proteins involved in nitrogen metabolism as well as the transcriptional activator FtrB involved in regulation of carbon and amino acid metabolism (BAW_20862–BAW_20865). The adjacent loci (BAW_20861 and BAW_20866) contain sequence partial deletions in 2308 (corresponding to BAB2_0903) or 2308W (corresponding to BAB2_0904). Also evident is an inversion of the 47 kb GI-5 (Rajashekara et al., 2004) in 2308W as compared to 2308 and 2308A that results in deletion of BAB2_1075, encoding an IS3 transposase. Notably, insertion elements were detected alongside all of the indel regions. These observations show not only how unstable the B. abortus 2308 genome is but also that chromosomal rearrangements are a source of genetic variability, probably as consequence of IS mediated transposition events (Tsoktouridis et al., 2003; Ocampo-Sosa and García-Lobo, 2008; Mancilla et al., 2012).

According to the National Collection of Cultures (NCTC) in England, reference strains are “stipulated in internationally recognized standard methods as definitive control strains for various microbiological testing procedures⁴.” The fact that Brucella is regarded as a genetically homogeneous genus, but significant differences are present in B. abortus 2308 stored in different laboratories raises a broader question and challenges the concept of “reference strain.” In the case of B. abortus 2308 this is even more relevant, since this strain has been regarded as the canonical challenge organism in vaccine trials where different results about its survival in mice models are reported (Montaraz and Winter, 1986; Ko et al., 2002; Yang et al., 2006). There are probably several reasons for these contrasting results. Some, such as differences due to strain handling can be accounted by using standardized quality control and experimental design protocols. Probably more difficult to control are genome changes. Reference strains are usually isolates obtained from clinical cases used in research labs as infection models, and as such, are subject of genetic modification according to the environment (Hopkins et al., 1981; Turse et al., 2011). A detailed description of the strain used, including a publically available WGS, as well as growth, propagation and maintenance conditions included with reported results could help readers to better assess the reported observations.

MS-E performed bioinformatics analysis, manual curation, and prepared tables and figures. NR-V performed molecular typing and manual curation. AC-Z, CJ-R, RR, and DC performed manual curation. CC provided data for manual curation. EB-C and CC-D performed biochemical typing and provided data for manual curation. KB performed bioinformatics analysis and edited the manuscript. EC-O and EM analyzed data. NT performed bioinformatics analysis and manual curation. XD and JL performed manual curation, organized, and analyzed data. CG-V designed the study, analyzed data, performed manual curation, and wrote the manuscript. All authors edited the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.