Black mustard (Brassica nigra (L.) Koch, 2n = 16, BB genome) represents the diploid Brassica B genome. It is an outcrossing oilseed species [1] which can also be used as a condiment [2]. However, it is a relatively minor crop compared with the closely related mustard species Brassica juncea (L.) Czern (AABB genome) and Brassica carinata Braun (BBCC genome), which are allotetraploid species that share the B genome [1]. Most studies within the Brassiceae have focused on the widely cultivated Brassica species, Brassica rapa L. (AA genome), Brassica oleracea L. (CC genome), and Brassica napus L. (AACC genome) [3]. However, the diploid Brassica B genome is considered to be an important source of useful genes in Brassica breeding, including drought tolerance, disease resistance, and oil seed quality [4–9]. In order to exploit the variation found within the B genome for the breeding of other Brassica oilseeds, an understanding of the relationship between the three Brassica genomes is required.

Although there has been extensive evidence of pairing and recombination between the A and C Brassica genomes [10–12] the B genome appears unusual since no or limited recombination between the B genome and its related A and C homologues has been observed [13–17]. The reason for this difference can be hypothesized as either the result of genetic factors regulating homologous pairing in the B genome similar to those suggested for the A and C genomes [18–21] or significant structural divergence of the B genome relative to the A and C genomes. Although it has been suggested through limited sequence and comparative mapping data that the B genome may have diverged to such an extent that the lack of recombination can be explained [22], questions still arise as to accuracy of this assertion. Uncovering the level of homology at the micro level between the three Brassica diploid genomes should provide insights into the true relationship between the genomes.

The cultivated Brassica species are closely related to the fully sequenced model plant Arabidopsis thaliana, [23] since the Brassica and Arabidopsis lineages diverged between 14.5 and 20 million years ago (Mya) [24]. It is believed that B. nigra evolved from the same polyploidy events, and resultant paleohexaploid, which took place around 7.9 to 14.6 Mya within the Brassica lineage. Although there are good estimates placing the separation of the B. rapa (A genome) and B. oleracea (C genome) at 3.7 Mya [25], there is little data to age the separation of the B genome from the A/C lineage [26]. Comparative mapping has identified blocks of conserved gene content and order between the A. thaliana genome and the Brassica A and C genomes of B. napus; where 21 ancestral segments of the A. thaliana genome could be replicated and rearranged to represent the Brassica genomes [27]. This work and further mapping in related species was used to propose an ancestral karyotype, reconstructed from 24 conserved blocks, A-X, for the entire Brassicaceae family [28]. Physical mapping of the regions homologous to a 222 kb region of A. thaliana chromosome 4 in B. oleracea[29], B. rapa[30] and B. napus[31] showed highly conserved collinearity of the orthologous regions in the three genomes and corroborated the triplicated nature of the diploid Brassica genomes. Previous sequence-level studies in B. oleracea[32], B. rapa[33] and B. napus[34] have clarified aspects of genome evolution and organization in Brassica by exploiting the close relationship with the genome of the model plant A. thaliana.

Physical mapping can provide an accurate representation of a genome when sequence is not available. A number of large insert libraries, the foundation for most physical maps, have been developed for the Brassica A and C genome species and have been exploited to facilitate map-based gene cloning for traits of interest and to provide insights into the evolutionary mechanisms that have formed these complex genomes [29–32, 35–39]. In order to provide insights into the relatively uncharted Brassica B genome we have developed a large insert library for B. nigra and have used this tool to describe the physical organisation of six regions of the B genome of B. nigra homologous to the same 222 kb region of A. thaliana, which was exploited previously to uncover aspects of genome organisation and evolution in the Brassica A/C genomes [29, 30, 32, 34]. We compared these orthologous regions of the B genome with the equivalent regions of A. thaliana, and the Brassica A and C genomes at the sequence level and report highly conserved gene content and order across the three genomes. An earlier divergence time separated the B genome from the A and C Brassica lineage by some 3 million years, which is barely reflected in the level of observed genome fractionation.

In the 1950’s when B. juncea replaced B. nigra as the mustard crop of choice in Asia, B. nigra was effectively abandoned with regards to crop improvement [50] and has been the subject of limited breeding, genetics and genomics research compared to the other two diploid species of U’s triangle [51]. However, the Brassica B genome has been recognised as a useful source of novel alleles for various traits of interest, in particular disease resistance and tolerance to abiotic stress, and numerous attempts have been made to transfer these traits to the Brassica A and C genomes, with varying success [9, 13, 14, 17, 52–54]. An improved understanding of the genome structure and gene composition of the Brassica B genome could provide insights into its relationship among the Brassica species and could potentially facilitate exploitation of this important resource. The current study benefited from previous research that had studied extensively the organisation of the Brassica A and C genome regions homologous to a 222 kb region of A. thaliana chromosome four, which was itself duplicated on chromosome five [44].

The Brassica B genome was separated previously from the A/C genome lineage in phylogenetic analyses [23, 55]. Analysis of the six regions within the Brassica B genome compared to both the A and C genomes allowed a more accurate estimate of the timing for divergence of the two lineages, ~ 6.2 Mya (± 2.19), compared to those previously published which ranged widely from 7.9–14.6 Mya [25] to 5–10 Mya [56]. The pattern of genome rearrangements and gene deletions that differentiate the Brassica genomes over this period have led to the extant species. The Brassica B genome as for the other diploid Brassica genomes, retained three genomic copies of each A. thaliana region, reflecting the underlying whole genome triplication (WGT) event that distinguishes the Brassiceae tribe [25]. The same rate of synonymous substitution was observed between the triplicated copies of the three Brassica genomes and A. thaliana (Table 4) and similarly the intragenomic rate, (A: Ks =0.323 ± 0.12, B: Ks =0.353 ± 0.08, C: Ks =0.353 ± 0.11) across the triplicated regions within the genomes was relatively equivalent, suggesting the B genome evolved from the same paleohexaploidy event as the A and C genome. As reported previously in B. oleracea, B. rapa and B. napus[32, 34] and shown in Figure 2 there is a high level of conservation for the composition and order of genes among the three Brassica diploid genomes compared to A. thaliana. Observed breaks in collinearity were due to predominantly minor rearrangements, such as inversions, duplications, small insertions and deletions, although one large insertion (> 200 kb) of a non-collinear segment in the C genome was found. One major chromosomal rearrangement, a large inversion relative to A. thaliana chromosome four, differentiated the A/C lineage from the Brassica B genome for one of the triplicated copies, which was also apparent from the genome sequence of B. rapa[45]. Genetic mapping of the B genome in B. juncea has certainly indicated that a number of large-scale chromosomal rearrangements may differentiate the Brassica A and B genomes [22], although this study also indicated that at least three of the B genome chromosomes were virtually collinear with their A genome homoeologue [22]. In addition, they identified conservation of ancestral Brassicaceae block structures across the three diploid Brassica genomes. Interestingly, as indicated in Figure 2, endpoints delineating ancestral Brassicaceae blocks, in this case between A. thaliana chromosomes two (block K) and five (block V), were found to be conserved between at least two of the triplicated copies in both the Brassica A (on linkage groups A02 – contig L, and A09 – contig B) and B genomes [45, 46]. The identification of such conserved ancestral rearrangements will further assist in defining the structure of the progenitor genome of the Brassiceae tribe, which is proposed to differ from that of A. thaliana[57].

Although the regions studied demonstrated marked conservation relative to A. thaliana, each species and each genomic region were characterized by the presence of gene sequences from non-collinear regions, which were rarely shared. However, there appear to be regions that are more sensitive to such alterations and although they are common to all genomes in lacking collinearity (for example in contig E between At4g17480 and At4g17390) the changes observed were specific to each. In addition, each of the regions across the genomes varied in size (Table 3). The B. nigra B genome has been reported to be 632 Mb (0.65 pg), smaller than B. oleracea at 696 Mb (0.71 pg) but larger than B. rapa at 529 Mb (0.54 pg), (Johnston et al., 2004). It had previously been shown that the C genome was expanded compared to the A genome for the targeted regions, which would be in accordance with current genome estimates; however, our estimates for this region place the B genome closer in size to that of B. rapa.

There has been considerable interest in attempting to transfer genomic regions from the B genome into the A and C genomes of B. napus[4, 13, 14, 58]; however, this has been largely unproductive due to limited or no recombination occurring between introduced B genome chromosomes and their supposed orthologues. Even the use of artificial resynthesis to develop new allopolyploids, which greatly enhances the level of recombination between the Brassica A and C genomes [11], has been ineffectual [59]. Yet at the genetic level [22] and as shown here at the sequence level there is marked conservation across the three Brassica genomes, in fact the level of karyotype and microstructure divergence across the three genomes is comparable. The question is why is aberrant homoeologous recombination so frequently observed in newly resynthesized B. napus (approximately 10% of the total observed recombination) but absent in equivalent resynthesized B. juncea[11, 59]. Although it is possible that there is a genetic barrier to recombination such as that found in wheat [60] or proposed for B. napus[18] the inability to breakdown this control would suggest limited or no variation exists for the trait, or unlike the progenitors of wheat the B genome diploid maintains activity for a trait which has value only in the polyploid state. If the barrier is then physical, then chromosomal rearrangements not captured in previous genetic mapping studies presumably differentiate the genomes, one such example would be the inversion found here that is specific to the A/C lineage. Inversions have been proposed to have a significant impact on plant evolution [61] and could have lead to the reproductive isolation and speciation of the B genome.

Although the current analyses focuses on two triplicated regions of the Brassica B genome, it provides insights into the gene retention pattern across the wider genome, since the A genome regions studied here reflect the pattern found in the B. rapa genome sequence [45]. The publication of the first Brassica genome sequence, that of B. rapa, provided extensive evidence of gene fractionation following the WGT and perhaps most interestingly due to bias in gene retention in the three genomic copies led to the two-step polyploidy hypothesis with fractionation between steps [45]. This hypothesis, in which the paleohexaploid was formed from the hybridisation between a diploid and an established tetraploid, has been rigorously tested since, with the retention of exonic regions being studied in detail for the A genome [62] and indicating that deletions are the major mode of gene inactivation and hence evolutionary drive in Brassica species. For the studied regions, the B genome appears to follow the same gene fractionation pattern as the A genome with one of the three genomic copies showing a higher level of conservation (Figure 4; Additional file 1: Table S3). The observed level of gene loss or fractionation for the triplicated regions at each important branch point in the evolution of the studied genomes is indicated in Figure 4. The highest percentage of fractionation appears to have occurred prior to the separation of the three genomes that is where gene loss is shared across all three genomes. Although the general pattern of fractionation holds true species-specific differences can be created through the types of genes that are maintained. For example in the regions homologous to A. thaliana chromosome five, no copies of two annotated disease resistance genes, At5g47250 and At5g47260, are found in the A and C genomes, yet multiple full or partial copies are maintained for each in the three homologous B genome regions. In addition, local insertion of a further annotated disease resistance gene (At1g59620) in this region of the B genome for contig A indicates a resistance cluster specific to the B genome.

Analysis of homologous regions of three closely related Brassica diploid genomes has identified extensive conservation of gene content and order, which is not reflected in the established phylogenetic relationship of the three species. The genome of B. nigra, based on its inability to establish effective pairing structures with the A and C genomes, might have been expected to have undergone more rearrangements at the micro-level, suggesting perhaps that major chromosomal events such as the observed inversion could have played a significant role in the speciation of the B genome. The B genome is likely to be an excellent source of novel genes for a number of important traits and capturing this diversity through genome sequencing could prove important for future crop improvement.