Dynamic Rearrangements Determine Genome Organization and Useful Traits in Soybean^,,

Abstract

Soybean (Glycine max) is a paleopolyploid whose genome has gone through at least two rounds of polyploidy and subsequent diploidization events. Several studies have investigated the changes in genome structure produced by the relatively recent polyploidy event, but little is known about the ancient polyploidy due to the high frequency of gene loss after duplication. Our previous study, regarding a region responsible for bacterial leaf pustule, reported two homeologous Rxp regions produced by the recent whole-genome duplication event. In this study, we identified the full set of four homeologous Rxp regions (ranging from 1.96 to 4.60 Mb) derived from both the recent and ancient polyploidy events, and this supports the quadruplicated structure of the soybean genome. Among the predicted genes on chromosome 17 (linkage group D2), 71% of them were conserved in a recently duplicated region, while 21% and 24% of duplicated genes were retained in two homeologous regions formed by the ancient polyploidy. Furthermore, comparative analysis showed a 2:1 relationship between soybean and Medicago truncatula, since M. truncatula did not undergo the recent polyploidy event that soybean did. Unlike soybean, M. truncatula homeologous regions were highly fractionated and their synteny did not exist, revealing different rates of diploidization process between the two species. Our data show that extensive synteny remained in the four homeologous regions in soybean, even though the soybean genome experienced dynamic genome rearrangements following paleopolyploidy events. Moreover, multiple Rxp quantitative trait loci on different soybean chromosomes actually comprise homeologous regions produced by two rounds of polyploidy events.

Genome duplication is common in plants (Wendel, 2000; Adams and Wendel, 2005; Cui et al., 2006) and particularly widespread in angiosperms (Otto and Whitton, 2000; Jaillon et al., 2007). Many modern diploid plants are in fact paleopolyploids that possess vestiges of multiple rounds of polyploidy (Blanc and Wolfe, 2004; Adams and Wendel, 2005). Soybean (Glycine max) is considered a paleopolyploid. Cytogenetic studies have provided proof of ancient polyploidy events in soybean and have reported a unique chromosome number for Glycine among other members of the Phaseoleae (Hadley and Hymowitz, 1973; Lackey, 1980). In addition, fluorescence in situ hybridization analyses using bacterial artificial chromosomes (BACs) have allowed the identification of segmental duplications and chromosome-level homeology within the soybean genome (Pagel et al., 2004; Walling et al., 2006). Genetic mapping has confirmed many duplicated regions of the soybean genome (Shoemaker et al., 1996; Lee et al., 1999, 2001; Cai et al., 2008b), and high levels of microsynteny were observed within homeologous BACs screened by RFLP probes (Yan et al., 2003, 2004).

Estimation of evolutionary distances between duplicated genes using EST collections has provided further evidence for genome duplication in soybean by demonstrating two rounds of polyploidy events (Blanc and Wolfe, 2004; Schlueter et al., 2004). However, because of nonoverlapping Ks (substitutions per synonymous site) peaks between soybean and Medicago truncatula, it is unclear whether the more ancient duplications present in the genomes of both of these species occurred prior to the Glycine-Medicago split or individually after taxon divergence. A phylogenetic analysis using 39 gene families suggested that an ancient polyploidy event occurred in the common ancestor of the soybean and M. truncatula before taxon divergence and that a relatively recent polyploidy event occurred independently in the soybean lineage (Pfeil et al., 2005). Given this shared polyploidy, an approximately 25% to 30% greater rate of synonymous substitution in M. truncatula was suggested for two different Ks peaks (Shoemaker et al., 2006). Differences in evolutionary rates between these two species were demonstrated by an analysis of four soybean regions and two M. truncatula homologous regions anchored by tandemly duplicated lipoxygenases (Shin et al., 2008).

Sequencing of homeologous BAC clones anchored by HCBT and FAD2 genes has provided insight into the duplicated genomic regions of soybean produced by the recent genome duplication (Schlueter et al., 2006, 2007b). Genome analysis with 17 homeologous BAC clones supports whole-genome duplication (WGD) rather than segmental duplication in soybean (Schlueter et al., 2007a). Recently, three homeologous regions centered on the Rpg1-b gene were analyzed, and a maximum of four homeologous regions in modern Glycine species were proposed (Innes et al., 2008). The presence of lipoxygenase families in four homeologous regions provides support for the hypothesis that the quadruplicated genome structure of soybean is derived from two different rounds of polyploidy (Shin et al., 2008).

Bacterial leaf pustule (BLP), caused by Xanthomonas axonopodis pv glycines (Xag), is a common bacterial disease of soybean found in many soybean-growing regions of the world where high temperatures and humidity prevail. Early symptoms of BLP are small yellow to brown lesions with raised pustules in the center (Groth and Braun, 1986); these may develop into large necrotic patches and result in premature defoliation and yield loss (Hartwig and Johnson, 1953; Weber et al., 1966; Laviolette et al., 1970). Classical genetic analysis has revealed that resistance to BLP is controlled by a single recessive gene (Feaster, 1951; Hartwig and Lehman, 1951) designated rxp (Bernard and Weiss, 1973). The Rxp locus was mapped 3.9 centimorgans from Satt372 and 12.4 centimorgans from Satt014 on chromosome 17, formerly linkage group D2 (Narvel et al., 2001). Recently, quantitative trait locus (QTL) studies of BLP resistance revealed that Satt372 on chromosome 17 is strongly associated with resistance to six different isolates of Xag under both greenhouse and field conditions (Van et al., 2004). Several minor QTLs were also identified in different Xag isolates. The existence of major and minor QTLs for BLP resistance might be due to WGD. In a previous study, we reported two homeologous Rxp regions produced by a recent duplication event and described a paralogous region in M. truncatula (Van et al., 2008).

Here, we provide a more comprehensive description of the soybean genome structure anchored by Rxp genes. Furthermore, we investigate the evolutionary history of the Rxp regions based on a comparative genomic analysis of M. truncatula. Despite the fact that dynamic structural changes occurred in the Rxp regions in both soybean and M. truncatula after their divergence, genome structure, synteny, and gene functions were found to be highly conserved in the four homeologous regions in soybean. Our data also demonstrate that most of the multiple Rxp QTLs in different soybean linkage groups actually comprise homeologous regions that were produced by two rounds of genome duplication.

RESULTS

Identification of Homeologous Regions

Two chromosomal regions (chromosomes 17 and 5) were identified as homeologous Rxp regions produced by a recent duplication event (Van et al., 2008). To identify other homeologous regions produced by the ancient duplication event, single nucleotide polymorphism (SNP) markers were developed to target the region near Rxp between Satt486 and Satt372 on chromosome 17 using a comparative genomic approach based on the sequence of the homologous region in M. truncatula. Soybean ESTs within the homologous Rxp region reported by Van et al. (2008; M. truncatula chromosome 3: 21,036–21,254 kb) and near that region (M. truncatula chromosome 3: 21,254–21,534 kb) were surveyed. Among numerous soybean ESTs that showed homology based on BLAST analysis, 15 ESTs that had no similarity with BACs on chromosome 17 or 5 were selected (Fig. 1A). Primers were designed from 15 ESTs, and several SNPs were detected from two ESTs (GenBank accession nos. CO979743 and BE021935) between two parental genotypes, Pureunkong and Jinpumkong 2, of a genetic mapping population. CO979743 and BE021935 were mapped onto chromosomes 4 and 6, respectively (Fig. 1A), and these regions were considered to be candidates for other homeologous regions produced by the ancient duplication event.

Figure 1.

Identification of six chromosomal regions in soybean and M. truncatula. A, SNP marker development for identification of homeologous soybean regions related to Rxp. A comparative genomic approach using M. truncatula was used to identify the targeted region near Rxp. Soybean ESTs from the homologous Rxp region reported by Van et al. (2008; M. truncatula chromosome 3: 21,036–21,254 kb) and areas near this region (M. truncatula chromosome 3: 21,254–21,534 kb) were investigated. Developed SNP markers were mapped onto our previous RIL mapping population (Cai et al., 2008a). B, Comparative map of four chromosomal regions of the soybean related to the Rxp locus and their orthologous regions in M. truncatula. Soybean and M. truncatula genomes show a 2:1 relationship in Rxp regions: GmA and GmA′ correspond to MtA, while GmB and GmB′ are orthologous to MtB. The dotted lines between soybean and M. truncatula indicate their orthologous relationship. See Supplemental Table S1 for detailed sequence information. [See online article for color version of this figure.]

Five soybean BACs (gmw1-28M13, gmw1-28O12, gmw1-38P12, gmw1-61M05, and gmw1-102F24) were identified by CO979743 (chromosome 4) using PCR-based screens of the gmw1 BAC library. Based on their insert sizes as estimated by contour-clamped homogeneous electric field electrophoresis and fingerprinted contigs from WebFPC version 2.1 (http://www.agcol.arizona.edu/fpc/WebAGCoL/Soybean/WebFPC/), the largest BAC clone, gmw1-61M05 (FJ686871), was sequenced. Raw data were assembled and five large contigs were obtained. Two gaps were closed by a primer-walking strategy with sequences from both ends of the contigs, leading to three larger contigs. Three additional BACs (gmw1-54G20, gmw1-75B18, and gmw1-86A12) were identified using primers designed from the contig sequences of gmw1-61M05. Again, the largest BAC clone, gmw1-54G20 (FJ686870), was sequenced and assembled as three large contigs. Furthermore, all remaining gaps in gmw1-61M05 were closed by the contig sequences of gmw1-54G20. Thus, the two larger contigs with one gap covering 241,394 bp were constructed based on the sequences of gmw1-61M05 and gmw1-54G20.

BLAST searches of three chromosomal regions (chromosomes 17, 5, and 4) using the soybean genome via the soybean Phytozome site (Glyma0.1b; http://www.phytozome.com/soybean) indicated that contigs gmw1-29F06_gmw1-24M16 (chromosome 17), gmw1-20O10_gmw2-77P21_gmw1-89M01 (chromosome 5), and gmw1-61M05_gmw1-54G20 (chromosome 4) matched nearly perfectly with scaffold_25, scaffold_35, and scaffold_47, respectively (Fig. 1B). Scaffold_171, which showed homology to scaffold_35, was also identified as an Rxp homeologous region. Using sequences from scaffold_47, scaffold_157, and scaffold_158, a fourth homeologous region was identified. This region was confirmed by the position of BE02195 on scaffold_157; this EST was previously mapped onto chromosome 6 (Fig. 1A).

In silico mapping of the scaffolds onto the soybean linkage map confirmed our previous mapping data, including the candidate homeologous regions: scaffold_171 and scaffold_25 onto chromosome 17, scaffold_35 onto chromosome 5, scaffold_47 onto chromosome 4, and scaffold_157 and scaffold_158 onto chromosome 6. Based on sequence homology, these scaffolds were designated as GmA, GmA′, GmB, and GmB′, respectively (Fig. 1B; Supplemental Table S1). The identification of four Rxp homeologous regions supports the quadruplicated structure of the soybean genome.

Identification of Orthologous Regions in M. truncatula

Two regions homologous to the soybean Rxp locus were identified in M. truncatula by BLAST analysis of the M. truncatula genome sequence (Mt2.0; http://www.medicago.org/genome/index.php). These regions were labeled MtA (chromosome 8: 22,601,040–24,533,986 bp) and MtB (chromosome 3: 18,607,393–21,534,813 bp) based on sequence similarity and their relationship with homologous regions in the soybean genome (Supplemental Table S1). The soybean and M. truncatula genomes showed a 2:1 relationship with regard to the Rxp regions: GmA, GmA′, MtA and GmB, GmB′, MtB (Fig. 1B). Each pair of soybean regions had a corresponding orthologous region in M. truncatula and showed a closer relationship with one of the two orthologous M. truncatula regions at the sequence level.

Comparison of Homeologous Rxp Regions in Soybean

An alignment of the four Rxp homeologous regions demonstrates their syntenic relationships (Fig. 2). Two similar but distinctive pairs are evident: pair A (GmA versus GmA′) and pair B (GmB versus GmB′). The two members of pair A show a higher level of synteny to one another compared with pair B, and the same is true for pair B, indicating that both of these pairs were produced by a recent duplication event. Although small insertions/deletions and inversions were observed between GmB and GmB′, most of the duplicated region was conserved. In contrast, relatively lower levels of sequence conservation were observed between members of pair A and pair B. Numerous syntenic breakpoints were detected in pair members, but significant synteny was still maintained. The presence of two pairs of homeologous chromosomal regions in different linkage groups reflects the quadruplicated genome structure of soybean generated by an ancient polyploidy event and a more recent polyploidy event.

Figure 2.

Alignment of four homeologous Rxp regions in soybean. Gray lines between scaffolds indicate syntenic regions. Blue and yellow blocks denote forward and reverse directions, respectively. The paleopolyploidy event responsible for generating homeologous pairs and their synonymous substitution rates are shown on the right.

A total of 1,699 genes were predicted in the four homeologous regions: 268 genes in GmA, 293 genes in GmA′, 700 genes in GmB, and 438 genes in GmB′. Putative functional descriptions of the predicted genes were obtained using the UniRef100 database (Supplemental Table S2). The gene density in these regions is similar for all four regions: one gene per 7.31 kb in GmA, one gene per 6.95 kb in GmA′, one gene per 6.57 kb in GmB, and one gene per 7.23 kb in GmB′. These predictions are not significantly different from previous estimates of gene density (Young et al., 2003; Mudge et al., 2005; Shin et al., 2008). The average G/C content was nearly the same among the four regions: 31.77% in GmA, 31.99% in GmA′, 33.15% in GmB, and 32.18% in GmB′. Among 268 predicted genes in GmA, 189 were present in GmA′, indicating that 71% of the duplicated genes formed by the recent polyploidy event have been conserved. A similar rate of retention has been documented for Rpg1-b homeologous regions (Innes et al., 2008). Furthermore, among 268 genes, 21% and 24% of genes found in GmB (56 genes) and GmB′ (63 genes), respectively, were conserved, showing that nearly three-quarters of genes duplicated by the ancient polyploidy event have been lost.

A search was performed for all putative homologs between all possible pairs of the four regions, and synonymous (Ks) and nonsynonymous (Ka) substitution rates between the Gm-Gm paralogs were calculated (Table I; Supplemental Table S3). The median Ks values of Gm-Gm paralogs within pair A (0.16) and pair B (0.17) were almost identical to the Ks values previously reported for the recent polyploidy event (Blanc and Wolfe, 2004; Schlueter et al., 2004, 2006, 2007b; Innes et al., 2008; Shin et al., 2008; Van et al., 2008). The Ks value of Gm-Gm paralogs between combinations of A and B pairs were greater (0.67, 0.66, 0.66, and 0.62) than the intrapair values. This suggests that these duplicate gene pairs resulted from the ancient polyploidy event (Blanc and Wolfe, 2004; Schlueter et al., 2004; Pfeil et al., 2005; Innes et al., 2008; Shin et al., 2008). The average nucleotide and amino acid identities between Gm-Gm paralogs produced by the recent polyploidy event were 90% and 87.4%, respectively. The Gm-Gm paralogs produced by the ancient polyploidy event were 79.1% similar at the nucleotide level and 74.9% similar at the amino acid level.

Table I.

Median Ks values of pairwise combinations of regions

Combinations of Pairs	Median Ks (No. of Gene Pairs)
Ancient polyploidy
Mt-Mt paralogs	0.85a (2)
MtA-MtB	0.85 (2)
Gm-Mt paralogs	0.75b (65)
GmA-MtB	0.72 (14)
GmA′-MtB	0.80 (15)
GmB-MtA	0.77 (15)
GmB′-MtA	0.71 (21)
Gm-Gm paralogs	0.65c (251)
GmA-GmB	0.67 (58)
GmA-GmB′	0.66 (66)
GmA′-GmB	0.66 (62)
GmA′-GmB′	0.62 (65)
Taxon divergence
Gm-Mt orthologs	0.59d (186)
GmA-MtA	0.63 (62)
GmA′-MtA	0.54 (56)
GmB-MtB	0.59 (33)
GmB′-MtB	0.58 (35)
Recent polyploidy
Gm-Gm paralogs	0.17 (407)
GmA-GmA′	0.16 (211)
GmB-GmB′	0.17 (196)

^a,b,cThe median Ks value for paralogous genes indicates differences in evolutionary rates between soybean and M. truncatula.

^b,dA comparison of Gm-Mt paralogs and Gm-Mt orthologs indicates a polyploidy event in the most recent common ancestor of soybean and M. truncatula.

The four soybean Rxp regions differ in size (Fig. 2). GmB and GmB′ are approximately 2.3- and 1.5-fold longer than GmA and GmA′, respectively (detailed sequence information is available in Supplemental Table S1). Size variation among homeologous regions may be due to retrotransposon amplifications, which have been found to play a dramatic role in genome size expansion (Eichler and Sankoff, 2003). The approximate numbers of retroelements were estimated based on functional descriptions of predicted genes (Supplemental Table S2). GmB (69 elements) contained about 6- to 8-fold more retroelements than GmA (eight elements) and GmA′ (12 elements) and nearly 3-fold more than GmB′ (22 elements). Compared with genome size and the number of predicted genes, GmB had many more retroelement insertions than the other regions. It has been shown in soybean that a chromosomal rearrangement that placed one duplicate region of a pair adjacent to a centromere led to a striking increase in retrotransposon content and size of the region (Innes et al., 2008). In this study, however, GmB is approximately 15.73 Mb away from the centromere of chromosome 4, and this was similar to the distances between each of other three homeologous regions and its centromere (http://www.phytozome.com/soybean). Nevertheless, only GmB was found to contain pericentromeric repeats, while others were not (http://www.phytozome.com/soybean). Pericentromeric chromosomal regions were reported to be greatly expanded by transposable elements and repetitive DNA in Arabidopsis (Arabidopsis thaliana; Hall et al., 2006). Also, tandem repeats with interspersed retroelements were concentrated in pericentromere of soybean (Lin et al., 2005). Thus, pericentromeric repeats in GmB could be responsible for this expansion in length of one homeologous region.

Comparative Genome Analysis between Soybean and M. truncatula

We compared the four homeologous soybean regions and two homologous regions in M. truncatula to better understand the evolution of the soybean genome (Fig. 3). Comparative alignment using BLASTZ revealed that GmA and GmA′ are more similar to MtA than to MtB (Fig. 3A). Three big inversion blocks are present, but the high sequence similarity between the MtA and GmA/GmA′ sequences indicates that they are orthologs. Similarly, GmB and GmB′ show higher synteny to MtB than to MtA (Fig. 3B). Again, MtB appears to be orthologous to GmB/GmB′, although MtB is shorter than GmB.

Figure 3.

Alignment of homologous regions between soybean and M. truncatula: GmA versus MtA and MtB (A) and GmB′ versus MtA and MtB (B). Gray lines between contigs indicate regions of synteny. Blue and yellow blocks denote forward and reverse directions of homologous sequences, respectively. M. truncatula regions segregate in a diploid fashion, and syntenic blocks are designated with Roman numerals. Detailed sequence information for the M. truncatula blocks is available in Supplemental Table S1.

The relationships of the homologous regions suggested by the BLASTZ results were clarified by estimating the Ks and Ka values for homologous genes in soybean and M. truncatula (Table I; Supplemental Table S3). After gene prediction for the homologous M. truncatula regions was performed, putative soybean and M. truncatula homologs were compared. The Mt-Mt paralogs had a greater median Ks value (0.85) than the Gm-Gm paralogs (0.65), consistent with previous analyses, although the absolute values were slightly larger (Blanc and Wolfe, 2004; Schlueter et al., 2004). The greater Ks value for the Gm-Mt paralogs (0.75) than for the Gm-Mt orthologs (0.59) suggests that the older of the two WGD events in soybean occurred prior to the Glycine-Medicago split. Compared with the soybean homeologous regions, the M. truncatula regions were fractionated into syntenic blocks, which are designated with Roman numerals (Fig. 3). GmA shows colinearity with blocks I-a, IV, and VI and no synteny with block V in MtA. In contrast, GmB has synteny with blocks I-b and III in MtB, whereas MtA does not show colinearity. Due to this fractionated synteny, it was difficult to find colinearity between MtA and MtB, although they still showed synteny with the homologous soybean regions. This nonoverlapping structure of homologous regions in M. truncatula allowed us to identify only two Mt-Mt paralogous genes; their median Ks value (0.85) is almost identical to the previously reported Ks value of the duplication in M. truncatula (Table I; Mudge et al., 2005; Cannon et al., 2006).

Conserved QTLs among Homeologous Rxp Regions

Numerous QTLs are associated with the four Rxp homeologous regions produced by two rounds of WGD in soybean (Fig. 4). The majority of QTLs for BLP resistance are located on or near homeologous Rxp regions: resistance in the field and to six isolates of X. axonopodis (8ra, OCS-F, OCS-G, SDL2178, LMG7403, and LMG7404) are located on chromosome 17 (GmA), resistance to LMG7403 is located on chromosome 5 (GmA′), and resistance to SDL2178 is located on chromosome 4 (GmB; Van et al., 2004; Fig. 4). Moreover, several QTLs reported in these regions are conserved across homeologous regions, for example, QTLs for soybean cyst nematode (Heterodera glycines Ichinohe) resistance on chromosomes 17, 5, and 6; QTLs for seed protein content on chromosomes 5, 4, and 6; QTLs for seed size on chromosomes 17 and 6; and QTLs for seed weight, yield, and oil content in all four homeologous regions. The existence of conserved QTLs indicates that the functions of these duplicated genes have been retained after the WGD events. Additionally, QTLs related to sclerotinia stem rot resistance have been reported on chromosome 10, which shows high homology to block V.

Figure 4.

Conservation of QTLs, including QTLs for BLP resistance, between homeologous regions in the soybean genome. Blue boxes indicate the positions of scaffolds on the soybean composite genetic map, and dotted lines indicate a homeologous relationship. SSR markers linked to BLP resistance QTLs (adapted from Van et al., 2004) are highlighted in red boldface font. The relative positions of QTLs conserved between homeologous regions are indicated with colored arrows next to the genetic map: insect or fungus resistance QTLs (red); seed size, weight, and yield QTLs (green); oil and protein content QTLs (blue).

DISCUSSION

Evolutionary Change before Speciation

The comparative approach described above was designed to uncover the full set of quadruplicated homeologous regions in soybean to help further expand our understanding of the duplicated nature of the soybean genome and estimate the evolutionary changes that occurred in the Rxp regions after divergence from the ancestral legume. Synonymous site calculations helped establish the relative timings of speciation and polyploidy events in these regions. Comparison of the Ks value for Gm-Mt paralogs (0.75) and Gm-Mt orthologs (0.59) supports a shared polyploidy event in the ancestor of Glycine and Medicago; different rates of evolution in these two lineages were also detected (Table I). The Ks values tended to decrease: Mt-Mt paralogs (0.85), Gm-Mt paralogs (0.75), and Gm-Gm paralogs (0.65). These three Ks values actually represent the same polyploidy in the ancestral genome; thus, the rate of synonymous substitution in M. truncatula appears to be 30% greater than that in soybean. Insights provided by Ks values are not limited to specific regions; for instance, the Ks values for homologous Rpg1 regions (Innes et al., 2008) can be interpreted as follows. The Ks value for GmwH3-Mt paralogs (0.798) was clearly larger than the Ks values for GmwH1-Mt orthologs (0.487) and GmwH2-Mt orthologs (0.493), suggesting the shared polyploidy event prior to the Glycine-Medicago split. Moreover, the Ks value for GmwH3-Mt paralogs (0.798) was also larger than the Ks values for GmwH1-GmwH3 paralogs (0.488) and GmwH2-GmwH3 paralogs (0.459). These values support a greater rate of synonymous substitution in M. truncatula.

Evolutionary Change after Speciation

After the shared polyploidy event and following divergence from the Medicago lineage, soybean experienced an independent WGD event approximately 14.5 million years ago (Schlueter et al., 2004). Therefore, as a consequence of two polyploidy events in the ancestor of Glycine, four homeologous regions are expected to be present in the soybean genome (Innes et al., 2008; Shin et al., 2008). In this study, we were able to identify and analyze all four homeologous regions; these regions contain two similar pairs (Fig. 2). High levels of sequence conservation were observed between the two homologous chromosome pairs derived from the recent duplication event (intrapairs of A and B; see “Results”), including intergenic sequences, as has been reported previously (Schlueter et al., 2006, 2007b; Innes et al., 2008; Shin et al., 2008; Van et al., 2008). Furthermore, many paralogous genes generated by the recent WGD event have been retained within pairs (Supplemental Table S3). This observation contrasts with what has been found in studies of the maize (Zea mays) genome; the maize genome lost approximately half of its duplicated genes in nearly the same period of time (approximately 11.9 million years; Ilic et al., 2003; Lai et al., 2004; Messing et al., 2004). Several studies have suggested that the highly duplicated structure of the soybean genome reflects slow and incomplete diploidization in soybean (Schlueter et al., 2006, 2007b; Shin et al., 2008). Due to this distinctive feature of the soybean genome structure, it has been technically difficult to amplify a single copy of a target genomic region.

The slow process of diploidization in soybean is also reflected in the nature of the evolutionary changes in the duplicated regions after the ancient polyploidy event. After the shared polyploidy event, very few Mt-Mt paralogs were maintained, resulting in nonsynteny between MtA and MtB, in contrast to what was observed for soybean (Figs. 2 and 3). Furthermore, about one-quarter of duplicated genes (21%–24%) were retained after dynamic structural changes in the Glycine lineage. Such a high level of gene conservation is rare among legume families (Cannon et al., 2006).

Comparative Genomic Analysis of Soybean and M. truncatula

A key insight of this study is that the level of synteny between soybean and M. truncatula varies depending on the evolutionary relationship of their homologous regions. Theoretically, two different chromosomal regions should be found in the M. truncatula genome for a given soybean region: one should be paralogous, while the other should be orthologous to soybean. However, in practice, when the sequence is not long enough, the likelihood of discovering only one homologous region in M. truncatula is high. In our previous study of Rxp regions (Van et al., 2008), Ks values between the soybean homeologous region and their syntenic region in M. truncatula (0.7654 and 0.6877) were somewhat greater than that of Gm-Mt orthologs in other reports (Blanc and Wolfe, 2004; Shin et al., 2008). Moreover, the Ks values in our previous study were much closer to the Ks value for Gm-Mt paralogs (0.75) than to that for Gm-Mt orthologs (0.59) in this study. Thus, the M. truncatula syntenic region (Van et al., 2008) could be considered as paralogous (MtB) instead of orthologous (MtA) to the two soybean regions (GmA and GmA′).

This can be explained by the diploidization of the M. truncatula homeologous chromosomes (Fig. 3). Due to reciprocal deletion of a chromosomal segment, no synteny was found within the M. truncatula homeologous pair with the exception of block I (blocks I-a and I-b). Therefore, the M. truncatula region orthologous to the Rxp BACs (GmA and GmA′) appears to have been deleted during the diploidization process. Conflicting observations about synteny between these two species can be interpreted as follows. Highly syntenic regions in the vicinity of the soybean rhg1 gene (Mudge et al., 2005) had orthologous relationships, while a limited network of synteny was observed in the HCBT regions (Schlueter et al., 2008), possibly due to a paralogous relationship resulting from the absence of an orthologous M. truncatula region in the fractionated genome.

Chromosomal Rearrangements

The common ancestral genome of soybean and M. truncatula was reconstructed by consolidating homologous regions; the evolutionary changes in the Rxp regions are illustrated in flow-chart form in Figure 5. Six blocks were predicted in the common ancestor, but some blocks appear to have been differentially lost or translocated during polyploidy events and/or taxon divergence. In M. truncatula, only block I was retained, while the remaining blocks were reciprocally lost as a result of diploidization. In contrast, in soybean, prediction of genome rearrangement was more difficult due to the independent polyploidy event in soybean. Soybean scaffolds with homology to block V are located on two different chromosomes (chromosomes 10 and 20) and show similar levels of synteny and synonymous substitution rate (Ks = 0.17) to recent polyploidy pairs (Supplemental Fig. S1; Supplemental Table S3), based on the assumption that this region was translocated to a certain chromosome prior to the recent polyploidy event. Furthermore, block II homologous regions are located at chromosomes 4 and 6 adjacent to the Rxp homeologous regions (GmB and GmB′), indicating dynamic genome rearrangements following the recent polyploidy event. Furthermore, the reconstructed ancient Rxp locus that existed prior to the recent polyploidy event was likely partially segregated in a diploid manner as seen in M. truncatula, indicating that the independent polyploidy event in soybean may have had a critical influence on the structure of the modern Glycine genome. Details of the mechanisms that can explain such translocations of chromosomal segments have not been elucidated; further studies are required.

Figure 5.

Model of evolutionary changes in the Rxp regions. The common ancestral genome of soybean and M. truncatula was reconstructed by consolidating homologous regions. The hypothetical genome comprised six blocks that were differentially retained in the two species after polyploidy and speciation events.

Maintenance of Gene Function

Sequence conservation of homeologous chromosomal segments suggests functional conservation across duplicated chromosomes; two rounds of polyploidy events in soybean may have resulted in the formation of multiple BLP resistance loci. Among the QTLs reported by Van et al. (2004), three loci were closely linked to homeologous Rxp regions, and each of them was associated with different, isolate-specific resistance (Fig. 4; Supplemental Fig. S2). This indicates that duplicated Rxp loci have retained their ancestral function and diverged in a strain-specific manner. The retention of QTL function between homeologous chromosomes has been observed in several cases, such as Brassica species (Axelsson et al., 2001) and salmonid fishes (O'Malley et al., 2003; Somorjai et al., 2003; Leder et al., 2006; Moghadam et al., 2007; Gharbi et al., 2009). In Brassica species, the functional preservation of duplicated QTLs for flowering time has been reported (Axelsson et al., 2001); moreover, duplicated genes controlling flowering time appear to act in an additive manner to modulate flowering time (Schranz et al., 2002). Multiple QTLs for resistance to various X. axonopodis isolates would have offered a slight fitness advantage against a wide range of pathogens and, therefore, increased the probability of survival under varying environments and conditions. In addition to BLP resistance loci, several QTLs including soybean cyst nematode resistance were also conserved among the homeologous regions (Fig. 4). Therefore, multiple QTLs are often detected for a given trait in soybean; this could be a consequence of polyploidy events with multiple genes controlling the trait of interest.

As mentioned above, we have presented some evidence supporting the conservation of QTL function across homeologous regions in soybean. However, due to the inherently poor resolution associated with QTL-mapping experiments (Kearsey and Farquhar, 1998), the genes responsible for the QTLs could actually be located elsewhere near the duplicated regions. Therefore, more precise localization of BLP resistance QTLs with a high-density map are needed for a robust estimation of duplicated QTLs and their relationship to paleopolyploidy events in soybean. Furthermore, selected recombinants within an recombinant inbred line (RIL) population that contain genetic variations between homeologous regions could be tested whether their resistances to BLP appear to act in an additive manner or not.

Another surprising observation was that block V, thought to have been translocated prior to the recent polyploidy event, was also related to BLP resistance (Fig. 4). A QTL for BLP resistance to the field and 8ra isolate is located near the block V homologous region on chromosome 10. Moreover, a QTL related to sclerotinia stem rot resistance has been reported on the same locus that controls BLP resistance. By comparing the sequences of homeologous Rxp regions and block V homologous regions, the origin of the gene that controls resistance to BLP may become clearer, as well as the factors that control disease resistance in a strain-specific manner. This information should facilitate efficient candidate gene selection, which can be confirmed by further functional studies.

Identifying homeologous chromosomes and defining evolutionary relationships in a model legume are critical for understanding the origin and evolution of particular traits in the paleopolyploid soybean. The 7× assembly of the soybean genome is now available at the U.S. Department of Energy Joint Genome Institute Web site. Determining the extent of sequence variation between homeologous soybean regions may help to understand mechanisms of their unique functions. Furthermore, comparative genomic analysis with a model legume can provide insights into the evolutionary fate of duplicated genes. Our data highlight the dynamic structural organization of the soybean genome and emphasize the functional conservation of homeologous regions; furthermore, the data presented in this study can be used to help identify the rxp gene.

MATERIALS AND METHODS

EST-Based SNP Detection and Linkage Mapping

Soybean (Glycine max) ESTs similar to Medicago truncatula homologs were identified by BLAST searches against the soybean EST database (http://www.ncbi.nlm.nih.gov/blast). ESTs that shared no similarity with BACs on chromosome 17 or 5 were selected. Primers were designed based on the sequence of selected soybean ESTs using Primer3 software (Rozen and Skaletsky, 2000). These ESTs were amplified and sequenced as described by Van et al. (2004). After SNPs were identified between the two soybean parents of Pureunkong and Jinpumkong 2 (our previous RIL mapping population; Cai et al., 2008a), SNP genotyping using primer single-base extension was performed on a Victor3 microplate reader (Perkin-Elmer Life Science) as described by Cai et al. (2005).

Identification, Sequencing, and Assembly of BACs

Homologous soybean BACs related to the Rxp locus were identified by the same primer set used for selection of EST CO979743, which was also used for SNP detection. The primer sequences to amplify CO979743 were F (5′-CACAACCACCATCAGACTCAAT-3′) and R (5′-TGTCAGTTGGTGTTCCATAAGC-3′). The optimal annealing temperature was determined by a pilot experiment with soybean ‘Williams 82’ using a PTC-225 Peltier Thermal Cycler (MJ Research) prior to BAC library screening. The optimized cycling conditions were as follows: 94°C for 2 min, followed by 35 cycles of 94°C for 30 s, 68°C for 30 s, and 72°C for 30 s, and a final 2-min extension at 72°C. PCR was performed in a total volume of 11 μL containing 100 ng of Williams 82 genomic DNA or BAC pool, 15 pmol of each forward and reverse primer, 10 mm deoxyribonucleotide triphosphate mix, 1 μL of 10× buffer, 6.9 μL of double-distilled water, and 0.4 unit of Taq DNA polymerase (Vivagen). PCR-based screening was conducted with multidimensional pools of the Williams 82 soybean BAC library (Marek and Shoemaker, 1997) using the same conditions as described above. Final screening was carried out with candidate BAC clones obtained directly from a working copy of the library. During the entire screening procedure, Williams 82 was used as a positive control.

The insert size of the BAC clone was estimated by contour-clamped homogeneous electric field electrophoresis after NotI digestion. Electrophoresis was performed using a 1% agarose gel in Tris-acetate EDTA buffer for 14 h at 6 V cm⁻¹ with an initial pulse time of 5 s and a final pulse time of 15 s. MidRange PFG Marker I (New England Biolabs) was used as a fragment size marker.

Soybean BACs were sequenced by a hybrid approach using both the Genome Sequencer (GS) FLX system and an ABI 3730xl sequencer. Sequences were determined by combining 20× coverage of GS FLX reads and 2× coverage of ABI 3730xl reads. Raw data of these reads were assembled using Phrap with default parameters. Remaining gaps were closed by primer walking from both contig ends.

Identification and Mapping of Scaffolds

Soybean scaffolds were identified by BLAST analysis of the 7× WGS assembly (Glyma0.1b; http://www.phytozome.com/soybean) using BAC sequences as queries. These scaffolds, representing three chromosomal regions of soybean, were BLASTed against the soybean genome database to identify other scaffolds that were produced by polyploidy. All of the scaffolds were mapped in silico onto the Soybean Consensus Map 3.0 (http://soybase.org) by BLAST searches against a database comprising sequences of simple sequence repeat (SSR)-containing clones (Song et al., 2004) and SNP-containing ESTs (Choi et al., 2007). Only the best hits among the scaffolds were selected to avoid mispositioning due to duplicated sequence. Relative positions of QTLs were estimated based on the soybean composite map at Soybase (http://soybase.org).

Sequence Analysis

Homologous M. truncatula regions were identified by BLAST analysis of the M. truncatula genome assembly version 2.0 (http://www.medicago.org/genome/index.php) using soybean scaffold sequences as queries. Pairwise sequence comparisons were performed with the BLASTZ program (Schwartz et al., 2003), and results were visualized using GBrowse (http://www.gmod.org/ggb/gbrowse.shtml) and SynBrowse (http://www.synbrowse.org).

Gene prediction was carried out using FGENESH with an Arabidopsis (Arabidopsis thaliana) matrix, and predicted genes were annotated using BLASTP against the UniProt Reference Clusters (UniRef100). Putative homologs between predicted genes in each chromosomal region were predicted using BLASTP, and only bidirectional best hits (e value < 1e-10, identity > 40%) across regions were accepted. For better accuracy, manual curation was conducted by comparing the pairwise alignments of homologous regions. The Ks values of putative homologs were calculated using PAML (Yang, 1997). The median Ks value of homologs between two regions was used to estimate the absolute date of duplication events. Ks values less than 0.05 and greater than 1.0 were not included in the calculation (Schlueter et al., 2004).

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers FJ686870 (gwm1-54G20) and FJ686871 (gmw1-61M05).

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure S1. Alignment of two soybean chromosomal regions homologous to block V.

• Supplemental Figure S2. Comparison of soybean Rxp homeologous regions and QTLs previously reported by Van et al. (2004).

• Supplemental Table S1. Detailed sequence information of soybean and M. truncatula used in this study.

• Supplemental Table S2. Descriptions of predicted genes based on UniRef100.

• Supplemental Table S3. Pairwise comparisons of Ks values between homologous genes.