Construction of an integrated genetic linkage map for the A genome of Brassica napus using SSR markers derived from sequenced BACs in B. rapa

Background The Multinational Brassica rapa Genome Sequencing Project (BrGSP) has developed valuable genomic resources, including BAC libraries, BAC-end sequences, genetic and physical maps, and seed BAC sequences for Brassica rapa. An integrated linkage map between the amphidiploid B. napus and diploid B. rapa will facilitate the rapid transfer of these valuable resources from B. rapa to B. napus (Oilseed rape, Canola). Results In this study, we identified over 23,000 simple sequence repeats (SSRs) from 536 sequenced BACs. 890 SSR markers (designated as BrGMS) were developed and used for the construction of an integrated linkage map for the A genome in B. rapa and B. napus. Two hundred and nineteen BrGMS markers were integrated to an existing B. napus linkage map (BnaNZDH). Among these mapped BrGMS markers, 168 were only distributed on the A genome linkage groups (LGs), 18 distrubuted both on the A and C genome LGs, and 33 only distributed on the C genome LGs. Most of the A genome LGs in B. napus were collinear with the homoeologous LGs in B. rapa, although minor inversions or rearrangements occurred on A2 and A9. The mapping of these BAC-specific SSR markers enabled assignment of 161 sequenced B. rapa BACs, as well as the associated BAC contigs to the A genome LGs of B. napus. Conclusion The genetic mapping of SSR markers derived from sequenced BACs in B. rapa enabled direct links to be established between the B. napus linkage map and a B. rapa physical map, and thus the assignment of B. rapa BACs and the associated BAC contigs to the B. napus linkage map. This integrated genetic linkage map will facilitate exploitation of the B. rapa annotated genomic resources for gene tagging and map-based cloning in B. napus, and for comparative analysis of the A genome within Brassica species.


Background
Brassica species are one of the most important crop groups in terms of cultivated acreage, contribution to human and animal diets, and economic value. Of the six cultivated Brassica species, B. napus, B. rapa, B. juncea, and B. carinata provide about 12% of worldwide edible vegetable oil supplies [1]. They also provide many of the vegetables in our daily diet such as cauliflower, broccoli, cabbage, kohlrabi, kale (B. oleracea) and turnip, Pakchoi and Chinese cabbage (B. rapa) [2]. Brassica species are also a valuable source of dietary fiber, vitamin C and other anticancer compounds such as glucosinolates. In addition, rapeseed oil has been used as a biofuel, a desirable alternative for fossil oil worldwide.
Brassica species are closely related to the model plant Arabidopsis thaliana, and provide an opportunity to study genome rearrangements associated with polyploidization. Comparative mapping using molecular markers has already revealed extensive synteny between Brassica and Arabidopsis [3,4]. The genomes of diploid Brassica species and Arabidopsis diverged 14.5 to 20.4 million years ago (MYA) from a common ancestor [5][6][7]. Subsequent chromosomal rearrangements including segmental duplications or deletions and extensive interspersed gene loss or gain events since divergence from the common ancestor have resulted in the present diploid Brassica species [6,8], B. rapa (2n = 20, AA), B. nigra (2n = 16, BB) and B. oleracea (2n = 18, CC). The three amphidiploid species, B. juncea (2n = 36, AABB), B. napus (2n = 38, AACC), and B. carinata (2n = 34, BBCC) originated from relatively recent interspecific hybridizations among the three diploid species, most likely during human cultivation of the diploid crops [9]. The genome relationships between these Brassica species is commonly known as the triangle of U [10].
The importance of Brassica species to the world economy and human health, and their potential value as models for studying genome changes associated with polyploidization, have promoted an international effort to sequence the complete set of Brassica genomes [7,11,12]. B. rapa ssp. pekinensis, which has the smallest genome [13], was selected as the representative for Brassica A genome sequencing by the Multinational Brassica Genome Project (MBGP) http://www.brassica. info [7], with the original aim of establishing the complete sequence of this genome using a BAC-by-BAC strategy. The BrGSP developed various genomic resources, including mapping populations, DNA libraries and DNA sequences. Three reference linkage maps, derived from the BraCKDH [14], BraJWF3P [15], and BraVCS-DH http://www.brassica-rapa.org populations, were constructed and had been served as backbones for anchoring BACs and BAC contigs for chromosome-based genome sequencing in B. rapa. Three BAC libraries covering approximately 36 genome equivalents of B. rapa were constructed using restriction enzymes HindIII, BamHI and Sau3AI. A total of 200,017 BAC-end sequences (BESs) were then generated from these BAC libraries. As of August 7, 2007, a stringent Build 2 of the physical map was released that contained 1,428 contigs with an average length of 512 kb, covering an estimated 717 Mb equivalent to 1.3 × of the B. rapa genome http://www.brassica-rapa.org [16]. The physical map based on fingerprint analysis was integrated to the genetic map by STS and SSR markers and chromosome fluorescent in situ hybridization (FISH) analysis, which enabled the positioning of 242 gene-rich contigs to specific locations on the 10 LGs [16]. Based on the physical map of B. rapa and the in silico comparative map of BAC-ends onto Arabidopsis chromosomes, 629 'seed' BACs spanning 86 Mb of Arabidopsis euchromatic regions were selected, distributed throughout the B. rapa genome http://www. brassica-rapa.org [7]. These anchored BAC clones were sequenced to provide starting points ('seeds') from which to continue the whole genome sequencing http://www.brassica-rapa.org.
SSRs, or microsatellites, are tandem repeats of 1-6 nucleotide sequence motifs flanked by unique sequences [17]. SSRs have become desirable molecular markers for gene tagging, germplasm evaluation, molecular-assisted selection and comparative mapping [18] because they have several advantages over other DNA markers, including being co-dominant, highly polymorphic, abundant, and distributed throughout the genome [19]. The DNA sequence information generated from BAC-ends and anchored BACs in B. rapa by the BrGSP (e.g. http://www.brassica.bbsrc.ac.uk; http:// www.brassica-rapa.org) provided an opportunity to evaluate the abundance and relative distribution of these SSRs in the whole genome [20,21], and an excellent opportunity to develop a large number of markers with known map positions to establish direct links between genetic, physical, and sequence-based maps of the Brassica crop species.
To utilize the invaluable genomic resources developed in B. rapa for genome analysis and genetic improvement in B. napus and other cultivated Brassica species, it is necessary to integrate the genetic and physical maps to the existing B. napus genetic maps [22,23]. The objectives of this study were to develop SSR markers from publicly available sequenced BACs in B. rapa, to integrate these markers to the existing B. napus linkage map [22], and to anchor these sequenced BACs and their associated BAC contigs to the A genome in B. napus. Here we report the identification and characterization of SSRs derived from sequenced BACs in B. rapa and the construction of an integrated genetic map in the A genomes of B. napus.

Results
Characterization of microsatellites in sequenced BACs of B. rapa As this study started, a total of 536 seed BACs had been sequenced to phase II or phase III by the BrGSP and deposited into NCBI-GenBank http://www.brassica.info/ resource/sequencing/status.php. These had been used as starting points for the selection of BACs to extend contigs in two directions with minimum overlap, based on matches to BAC-end sequences [7]. As indicated at http://www.brassica-rapa.org, 232 of these BACs had been completely sequenced (phase III) and the remainders were sequenced to phase II (i.e. fully oriented and ordered sequence with some small gaps). Of these, 482 had provisionally been assigned to specific chromosomes of B. rapa, and 54 unassigned to any specific chromosome. Considerably more BACs had been sequenced for chromosomes A3 and A9 (153 and 85 respectively), which had been allocated to the Korea Brassica Genome Project (KBGP). Chromosome A4, which is the shortest LGs in many linkage maps [14,22,24], had the least number of sequenced BACs (22). Twenty seven to forty four sequenced BACs were assigned to each of the remaining seven chromosomes. The total length of the sequenced BACs was 63.5 Mb, representing a coverage of 11.5% of the A genome and corresponding to 86 Mb of the euchromatic regions of the Arabidopsis genome http://www.brassica-rapa.org [7].

Development of microsatellite markers from sequenced BACs
A total of 890 primer pairs, designated as BrGMSs, were designed from the sequenced BACs to amplify the simple sequence repeats. At least one SSR marker was developed from each BAC. The priority was given to the di-and tri-nucleotide repeats because they had higher success rates of amplification and higher levels of polymorphism [22]. Among these markers, the largest proportion was (AG) n , followed by (AT) n , (AAG) n and (AC) n motifs ( Table 2). No (CG) n , (CCG) n and (ACT) n motifs were included because they are very rare in the BAC sequences. Table 2 summarized the features of the new developed BrGMS markers.
The 890 BrGMS markers were tested for amplification using a panel of six rapeseed varieties that had been used in previous study [22]. Of these, 794 (89.21%) markers successfully amplified at least one PCR product from the B. napus genome, and 460 (51.7%) were polymorphic among the six varieties. A downloadable file including the marker names, primer sequences, the source BACs and chromosomes, the amplification and scorability of the markers was available as Additional file 1.

Integration of the BrGMS markers to the B. napus linkage map
Of the polymorphic BrGMS markers, 219 that showed polymorphism between the DH line 'No.2127' and the inbred line 'ZY821' were utilized to genotype the BnaNZDH population. Among these markers, 201 detected only one polymorphic locus, 18 detected two polymorphic loci. Of the 201 markers with a single polymorphic locus, 168 were mapped to the A1 to A10 LGs of the A genome and 33 to the C1 to C9 LGs of the C genome. The 18 markers with two polymorphic loci were mapped to both the A and C genomes, which was often observed in genetic maps of B. napus due to the high level of sequence homology between the A and C genomes [4,23]. In total, 186 BrGMS loci were integrated to the A1 to A10 LGs of the existing genetic linkage map constructed using the same population [22]. The A genome LGs covered a total length of 1013.4 cM, with an average distance of 3.34 cM between adjacent loci ( Figure 1, 2, 3, 4, 5, 6, 7, 8, 9 &10). The BrGMS loci on A1 to A10 were evenly distributed. The 186 BrGMS loci were derived from 161 sequenced BACs. Of these loci, 147 were derived from BACs that had been assigned to specific chromosomes on the B. rapa physical map, while 39 loci were derived from 27 previously unassigned BACs. Among these 161 BACs, 21 BACs each had two BrGMS loci and two BACs each had three loci mapped to the A genome in B. napus, and the remaining 138 each had one locus mapped to the A genome. Of the 186 loci, 18 detected null alleles in one parent, which was also observed in previous research in Brassica [26,27]. No.2127 had 13 null alleles, while ZY821 had five null alleles.

Comparative analysis between the B. rapa and B. napus linkage maps
Three reference linkage maps, the BraCKDH, BraJWF3P and BraVCS-DH, were previously developed to assign BACs and BAC contigs to chromosomes for genome sequencing in B. rapa http://www.brassica-rapa.org [14,15,28]. These maps contained many common markers derived from the same set of sequenced BACs as those used in this study for marker development. These sequenced BACs had been provisionally anchored to the B. rapa linkage map through genetic mapping of SSR or STS markers [28]. For ease of comparison, we constructed an integrated B. rapa linkage map based on the common markers that were shared between two or more maps or different markers derived from the same BAC and had similar map positions on two or three reference linkage maps. The integrated linkage map was constructed merely based on the published map positions and marker orders on the three reference linkage maps. The orientation of LGs for the three B. rapa reference maps were not consistent, for the A genome the agreed orientation were consistent with the BraCKDH map [14,28], so we finished the integrated B. rapa linkage map based on the orientation of LGs in BraCKDH map [14,28]. We referred to the SSR markers on the three B. rapa reference maps and the subsequent integrated linkage map as 'KBGP markers', to distinguish them from those developed in this study. The B. napus linkage map constructed in this study was compared with the integrated B. rapa linkage map based on SSR markers developed from the same BACs. All LGs in B. napus were collinear with their corresponding homologues in B. rapa ( Figure 1, 2, 3, 3, 4, 5, 6, 7, 8, 9 &10), apart from A2 and A9, where single minor intra-chromosomal inversions or rearrangements involved one or two markers [4,15].

Assignment of the B. rapa BACs and BAC contigs to the B. napus linkage map
The mapped SSR markers were then used to construct a map to integrate the A genome map for B. napus with that of B. rapa genetic maps. The sequenced BACs were assigned to the A genome LGs based on map locations of the KBGP and/or BrGMS markers on the B. rapa and B. napus linkage maps. Our criterion for assigning a BAC to a map location required at least two markers derived from the same BAC being mapped to the same position or adjacent position. Due to the polyploid nature of the Brassica A genome, it is possible that a BAC could be incorrectly mapped to a paralogous position, if by chance independent evolution of the duplicated  (Figure 1, 2, 3 , 4, 5, 6, 7, 8, 9 &10 and Additional file 2). For example, marker KS50880 derived from KBrB071K22 was mapped to A2 on the BraJWF3P map and its associated BAC contig (contig31) was assigned to chromosome A2 on the physical map [16]. BrGMS622 derived from the same BAC was mapped to the similar position (within 1 cM) on A2 on the B. napus linkage map. Therefore we could unequivocally assign KBrB071K22 and its associated contig31 to A2 on the B. napus genetic map (Figure 1, 2, 3, 4, 5, 6, 7, 8, 9 &10 and Additional file 2). Nine BACs could not be directly assigned to the B. napus linkage map due to conflicts between the B. rapa and B. napus linkage maps or between the genetic and physical maps. Therefore we assigned these BACs to the B. napus linkage map by mapping two or more additional SSR markers derived from each BAC. The B. napus linkage map also contained 39 BrGMS loci derived from 27 previously unassigned BACs. Ten of the BACs each had at least two markers located at similar positions on the B. napus linkage map, and so these 10 BACs were then assigned to the B. napus LGs ( Figure  1 LGs, five to 10 additional SSR markers were developed from each BAC and used to screen polymorphisms between the two parents. Polymorphic SSR markers were integrated to the B. napus linkage map. Sixteen BACs (Additional file 3) each had Figure 4 Integration of A4 in B. rapa and B. napus using SSR markers derived from the same sequenced BACs. Cumulative recombination distances in cM are shown to the left and marker loci to the right of the linkage group. SSR markers developed from sequenced BACs are indicated in boldface. The correspondence between the SSR markers and the BAC clones is given in Additional file 1. The map was constructed using 88 DH plants derived from No. 2127 and ZY821 using BrGMS (boldface) markers and public SSRs as anchors [22]. Markers from the same BAC are connected by lines to indicate the collinearity between B. napus and B. rapa. The B. rapa BraJWF3P and BraVCS-DH linkage maps were redrawn from http://www.brassica-rapa.org. The dotted lines indicated the assignments of previously unassigned BACs. Integration of A5 in B. rapa and B. napus using SSR markers derived from the same sequenced BACs. Cumulative recombination distances in cM are shown to the left and marker loci to the right of the linkage group. SSR markers developed from sequenced BACs are indicated in boldface. The correspondence between the SSR markers and the BAC clones is given in Additional file 1. The map was constructed using 88 DH plants derived from No. 2127 and ZY821 using BrGMS (boldface) markers and public SSRs as anchors [22]. Markers from the same BAC are connected by lines to indicate the collinearity between B. napus and B. rapa. The B. rapa BraJWF3P and BraVCS-DH linkage maps were redrawn from http://www.brassica-rapa.org. The dotted lines indicated the assignments of previously unassigned BACs. Figure 6 Integration of A6 in B. rapa and B. napus using SSR markers derived from the same sequenced BACs. Cumulative recombination distances in cM are shown to the left and marker loci to the right of the linkage group. SSR markers developed from sequenced BACs are indicated in boldface. The correspondence between the SSR markers and the BAC clones is given in Additional file 1. The map was constructed using 88 DH plants derived from No. 2127 and ZY821 using BrGMS (boldface) markers and public SSRs as anchors [22]. Markers from the same BAC are connected by lines to indicate the collinearity between B. napus and B. rapa. The B. rapa BraJWF3P and BraVCS-DH linkage maps were redrawn from http://www.brassica-rapa.org. The dotted lines indicated the assignments of previously unassigned BACs.

Figure 7
Integration of A7 in B. rapa and B. napus using SSR markers derived from the same sequenced BACs. Cumulative recombination distances in cM are shown to the left and marker loci to the right of the linkage group. SSR markers developed from sequenced BACs are indicated in boldface. The correspondence between the SSR markers and the BAC clones is given in Additional file 1. The map was constructed using 88 DH plants derived from No. 2127 and ZY821 using BrGMS (boldface) markers and public SSRs as anchors [22]. Markers from the same BAC are connected by lines to indicate the collinearity between B. napus and B. rapa. The B. rapa BraJWF3P and BraVCS-DH linkage maps were redrawn from http://www.brassica-rapa.org. The dotted lines indicated the assignments of previously unassigned BACs. two or three SSR markers mapped to the same positions on the B. napus linkage map, and thus could be assigned to specific LGs. Another BAC, KBrB019M05, had four mapped SSR markers, with three markers (BrGMS4463, BrGMS4465 and BrGMS4466) being mapped to the same position on A2 and the other one (BrGMS369) to A6, a paralogous segment of A2 [15], suggesting that KBrB019M05 should be assigned to A2. Therefore, the mapping of SSR markers enabled the assignment of these BACs and their associated BAC contigs to specific LGs.
In the above analysis, we unequivocally assigned a total of 161 BACs to the B. napus linkage map, and thus constructed an integrated genetic map for the A genome of B. napus.

Validation of BAC assignment
To test the reliability of the assignment of BACs on the B. rapa physical map to the B. napus genetic map, we compared the map positions of markers derived from the same BACs. Fourteen BACs each had two SSR markers mapped on the B. napus linkage map (Additional file 4). The markers derived from the same BAC were mapped to the same positions on the B. napus linkage map (Figure 1 , 2, 3, 4, 5, 6, 7, 8, 9 &10) and it was confirmed that the assignments of these BACs were correct.
We further checked the map positions of marker loci that were derived from different BACs that were present within the same contigs. There were 16 BAC contigs that each had two or three different constituent BACs assigned to the genetic and physical map (Additional file 4). All SSR markers derived from BACs in the same contig were mapped at adjacent positions on the corresponding LGs (Figure 1, 2 , 3, 4, 5, 6, 7, 8, 9 &10 and Additional file 4). For example, BrGMS672 and BrGMS690 were derived from KBrB047D06 and KBrB042O05 respectively, within contig73 on A3, and were 3 cM apart on the B. napus linkage map (Figure 1, 2 , 3, 4, 5, 6, 7, 8, 9 &10). These results also demonstrated that the assignment of BACs to the A genome of B. napus were reliable.
To further validate the integrated genetic map constructed in this study, we selected five BACs randomly from each of five contigs, to develop and map additional SSR markers (Additional file 4). The additional markers derived from each BAC co-segregate with the previously mapped SSR markers (Figure 1, 2, 3 , 4, 5, 6, 7, 8, 9 &10 and Additional file 4). These results further indicated that the assignments of B. rapa BACs and BAC contigs to the B. napus linkage map were reliable.

Characterization of SSRs in B. rapa
Previously, our view of the abundance of microsatellites in Brassica had largely been inferred from hybridization Figure 8 Integration of A8 in B. rapa and B. napus using SSR markers derived from the same sequenced BACs. Cumulative recombination distances in cM are shown to the left and marker loci to the right of the linkage group. SSR markers developed from sequenced BACs are indicated in boldface. The correspondence between the SSR markers and the BAC clones is given in Additional file 1. The map was constructed using 88 DH plants derived from No. 2127 and ZY821 using BrGMS (boldface) markers and public SSRs as anchors [22]. Markers from the same BAC are connected by lines to indicate the collinearity between B. napus and B. rapa. The B. rapa BraJWF3P and BraVCS-DH linkage maps were redrawn from http://www.brassica-rapa.org. The dotted lines indicated the assignments of previously unassigned BACs.

Figure 9
Integration of A9 in B. rapa and B. napus using SSR markers derived from the same sequenced BACs. Cumulative recombination distances in cM are shown to the left and marker loci to the right of the linkage group. SSR markers developed from sequenced BACs are indicated in boldface. The correspondence between the SSR markers and the BAC clones is given in Additional file 1. The map was constructed using 88 DH plants derived from No. 2127 and ZY821 using BrGMS (boldface) markers and public SSRs as anchors [22]. Markers from the same BAC are connected by lines to indicate the collinearity between B. napus and B. rapa. The B. rapa BraJWF3P and BraVCS-DH linkage maps were redrawn from http://www.brassica-rapa.org. The dotted lines indicated the assignments of previously unassigned BACs.
screening [29][30][31][32] and analysis of genome survey sequences (GSSs) such as BAC end sequences (BES) [21] and whole genome shotgun sequences [33]. In this study, we analyzed the frequency of microsatellites identified from 536 sequenced BACs, which represents a total length of 63.5 Mb of the B. rapa genome and corresponds to a coverage of the gene-rich euchromatic regions of the Arabidopsis genome that are distributed across all 10 LGs of the B. rapa genome [7]. The frequency of SSR occurrence (one every 2.74 kb) in the BAC sequences of B. rapa is similar to that in the completely sequenced PAC and BAC clones in rice (one every 2.7 kb) [34]. This frequency is much higher than that in BAC-end sequences of B. rapa (one per 4.7 kb) [21] and B. napus (one every 4.0 kb) [22], and in the whole genome shotgun sequences of B. oleracea (one every 4.0 kb) [33], indicating that the frequency of SSRs identified from GSSs of B. rapa, B. oleracea and B. napus underestimated the abundance of microsatellites in Brassica genomes. In these BAC sequences, trinucleotide repeats (37.61%) are the most frequent within the B. rapa genome, followed by di-nucleotide repeats (36.21%). Of the tri-nucleotide repeats, the motif (AAG) n is the most frequent, consistent with the pattern in BES of B. rapa [21]. Among di-nucleotide repeats, the ranking of motifs is (AT) n > (AG) n > (AC) n repeats. This distribution is in good agreement with that observed in the complete genomes of Arabidopsis [25] and rice [34]. This compares with the ranking of (AG) n > (AT) n > (AC) n in BAC-end sequences of B. rapa [21], as well as in whole genome shotgun sequences of B. oleracea [33]. This discrepancy might be due to the non-random distribution of these repeat motifs, the biases introduced in the process of library construction, and the difficulty of sequencing long stretches of selfcomplimentary (AT) n repeats which appeared in BAC ends or small-insert genomic libraries using the single pass sequencing strategy. These data led to the conclusion that the BAC sequences, rather than the GSSs, could be more representative of the actual occurrence and composition of SSRs in the complete genomes of Brassica species [34].
Comparative mapping of the A genome in B. rapa and B. napus SSR markers have previously been developed from various sources including small-insert genomic libraries [29], BAC ends [14,26], ESTs [35] and whole-genome shotgun sequences [33], and contributed to many genetic linkage maps using different mapping populations [22][23][24]36]. However, until now these maps have been difficult to align and integrate due to the lack of common markers. The three B. rapa reference maps Figure 10 Integration of A10 in B. rapa and B. napus using SSR markers derived from the same sequenced BACs. Cumulative recombination distances in cM are shown to the left and marker loci to the right of the linkage group. SSR markers developed from sequenced BACs are indicated in boldface. The correspondence between the SSR markers and the BAC clones is given in Additional file 1. The map was constructed using 88 DH plants derived from No. 2127 and ZY821 using BrGMS (boldface) markers and public SSRs as anchors [22]. Markers from the same BAC are connected by lines to indicate the collinearity between B. napus and B. rapa. The B. rapa BraJWF3P and BraVCS-DH linkage maps were redrawn from http://www.brassica-rapa.org. included many common SSR markers developed from hundreds of sequenced BACs, and so an integrated linkage map was able to be constructed for B. rapa based on these common markers [15] http://www.brassicarapa.org. Our 890 BrGMS markers were developed from the same set of sequenced BACs as used for the construction of the B. rapa reference maps. Common BACs were then used as bridges to align the A genome LGs in B. rapa and B. napus. As seen in Figure 1, 2, 3 , 4, 5, 6, 7, 8, 9 &10, all the A genome LGs showed good collinearity between B. rapa and B. napus. These analyses suggested that the A genome is retained relatively intact in B. napus [37] despite extensive genome rearrangements such as insertions and deletions that have occurred following hybridization of the A and C genomes [38,39]. The gene order and content are conserved in the A genome of B. napus and B. rapa as revealed by comparative sequencing [39]. Comparative mapping [40] also showed that these LGs appear intact within B. juncea, the other amphidiploid of the triangle of U [10] that contains the A genome. The markers developed in this study will therefore be valuable for ongoing comparative analysis among Brassica species, especially within the A genomes.

Integrated genetic and physical map in B. napus
Map integration through the use of common markers is a powerful tool for anchoring BACs and BAC contigs to genetic and physical maps [16]. The mapping of a single marker can often result in mis-assignment of BACs due to highly conserved intra-and inter-chromosomal duplications, or recently evolved gene paralogs that have distinct locations in the triplicated B. rapa genome [16]. In B. rapa, anchoring of sequenced BAC clones to linkage maps was achieved by matching alleles amplified from the genomic DNA of Chiifu-401 and BAC clones derived from the same genotype (Chiifu-401) [28], which is the only way to assign BACs uniquivocally to a genetic map, especially where there are extensive duplications and triplications, as in B. rapa genome. In this study, we assigned BACs to the B. napus genetic map through mapping of at least two markers derived from the same BAC. Single BACs were assigned only when two or more markers derived from the same BAC were mapped to the same or adjacent positions. Using this strategy, we assigned 161 sequenced BACs and their associated BAC contigs to the B. napus genetic map, which represented the most gene-rich euchromatic regions distributed across the 10 chromosomes of the A genome. However, this strategy still has limitations since null alleles are very common in Brassica in some lines compared with others, as indicated in this and previous studies [26,27]. There may also be situations where the combination of null alleles within complementary paralogous segments in different populations may lead to mis-assignment. Moreover, one may be more confident in the assignment of SSRs where they amplify single polymorphic bands with sizes similar to the expected fragments from DNA derived from amphidiploid plants.
At the time at which this study commenced, 536 seed BACs had been sequenced and released via NCBI-GenBank. Of these, only 161 were assigned to the A genome LGs of B. napus, with the remaining 375 unable to be assigned due to the lack of polymorphism amongst the markers developed in this study. Clearly, there would be considerable value to assign all the available sequenced BACs to the B. napus LGs. More recently, a total of 1014 sequenced BACs have become available as shown at http://brassica.bbsrc.ac.uk. Even with the imminent availability of sequenced scaffolds from high throughput methods there remains a pressing requirement for alignment with sequenced BACs and unequivocal assignment and orientation to respective chromosomes. Thus, continued development of additional SSR markers from the set of sequenced BACs and sequence scaffolds will facilitate further integration of the A genome LGs in B. rapa and B. napus, and allow assignment of additional BACs and BAC contigs to the A genome in B. napus, especially where these have previously been unassigned. The value of BACs also lies in their ability to be used as probes for chromosome FISH. The mapping of markers derived from previously unassigned sequenced BACs could anchor their source BACs to the A genome in B. napus, which then enabled their reliable assignment to the B. rapa chromosomes based on the integrity of the A genome between the two species. The integration of the A genome LGs in B. rapa and B. napus will facilitate the utilization of genetic resources and the rapid transfer of knowledge from B. rapa to B. napus. Moreover, because the sequenced BACs cover the gene-rich euchromatic regions of the Arabidopsis genome, information from the model plant will also assist in understanding the relationships between gene function, genome structure and evolution in Brassicaceae and should yield valuable information for rapeseed research [41]. Furthermore, the construction of an integrated genetic map will accelerate the pace of gene mapping and cloning in B. napus.

Conclusions
We developed 890 microsatellite markers from 536 anchored seed BACs and used these to construct a linkage map in B. napus integrated with those of B. rapa, which included 219 newly developed BrGMS markers. Among these mapped BrGMS loci, 186 were evenly distributed on the A genome LGs. Most of the BrGMS loci on the A genome LGs in B. napus are collinear with the homoeologous LGs in B. rapa. The mapping of these BAC-specific SSR markers enabled the assignment of 161 B. rapa seed BACs, and also their associated BAC contigs to the A genome in B. napus. The integrated genetic map will accelerate the exploitation of the B. rapa genomic resources for gene tagging and cloning in B. napus and for comparative analysis of the A genome of Brassica species.

Plant materials
A panel of six B. napus breeding lines or cultivars ('S1', 'S2', 'M201', 'M202', 'No2127' and 'ZY821') which were parents of three established mapping populations was used for polymorphism screening of SSR markers [22]. A double-haploid (DH) population, BnaNZDH, with 88 individuals derived from the F1 of a cross between 'No2127' and 'ZY821' [22,42] was used for linkage mapping. Total DNA was isolated from young leaves of the six B. napus breeding lines and the 88 DH lines using the cetyltrimethylammonium bromide (CTAB) method [43], and used as template for PCR amplification.

Source of sequences, microsatellite identification and marker development
The sequences of 536 BACs were obtained from NCBI Genbank and employed for the identification of SSRs. The BAC sequences were submitted to a web-based SSR-discovery tool, SSRPrimer, which integrates the SSR repeat finder Sputnik [44] with the primer design program Primer 3 [45] http://hornbill.cspp.latrobe.edu.au/ ssrdiscovery.html, to identify microsatellites. The criteria used for the microsatellite search were a minimum of 6 repeats for di-nucleotide motifs, four repeats for trinucleotide motifs and three repeats for tetra-or pentanucleotide motifs. One to five SSRs with repeat length greater than 20 bp from each BAC were selected for marker development. The parameters for SSR primer design were set as default: primer length from 18 to 23 bp, GC content over 40%, melting temperature between 55°C and 70°C, and PCR products ranging from 100 to 400 bp in size. These newly developed BAC-derived SSR markers were designated as 'BrGMS', representing Brassica rapa genomic microsatellites. All primers were synthesized by Generay Biotech (Shanghai, China).
All BrGMS markers were used to examine the PCR amplification and polymorphisms using a panel of six rapeseed lines described above. PCR amplification, products separation and staining were performed as described previously [22].