Comparative Genetic Mapping in Boechera stricta , a Close Relative of Arabidopsis 1[C][W][OA]

The angiosperm family Brassicaceae contains both the research model Arabidopsis ( Arabidopsis thaliana ) and the agricultural genus Brassica . Comparative genomics in the Brassicaceae has largely focused on direct comparisons between Arabidopsis and the species of interest. However, the reduced genome size and chromosome number ( n 5 5) of Arabidopsis complicates comparisons. Arabidopsis shows extensive genome and chromosome reshufﬂing compared to its close relatives Arabidopsis lyrata and Capsella rubella , both with n 5 8. To facilitate comparative genomics across the Brassicaceae we recently outlined a system of 24 conserved chromosomal blocks based on their positions in an ancestral karyotype of n 5 8, rather than by their position in Arabidopsis. In this report we use this system as a tool to understand genome structure and evolution in Boechera stricta ( n 5 7). B. stricta is a diploid, sexual, and highly self-fertilizing species occurring in mostly montane regions of western North America. We have created an F 2 genetic map of B. stricta based on 192 individuals scored at 196 microsatellite and candidate gene loci. Single-nucleotide polymorphism genotyping of 94 of the loci was done simultaneously using an Illumina bead array. The total map length is 725.8 cM, with an average marker spacing of 3.9 cM. There are no gaps greater than 19.3 cM. The chromosomal reduction from n 5 8 to n 5 7 and other genomic changes in B. stricta likely involved a pericentric inversion, a chromosomal fusion, and two reciprocal translocations that are easily visualized using the genomic blocks. Our genetic map will facilitate the analysis of ecologically relevant quantitative variation in Boechera .

Comparative genetic mapping between related organisms within a phylogenetic framework is a powerful method for understanding genome evolution. Comparative mapping in the grass family (Poaceae) has been successful in detecting collinear genomic regions between a number of domesticated cereal and forage crops, leading to the formulation of the crop circle with rice (Oryza sativa) at the center (Moore et al., 1995;Devos, 2005). Rice was selected as the reference point because of its small genome and vast genomic resources, and not because it was phylogenetically well positioned to facilitate comparisons within the family. An analogous situation occurs in the dicot family Brassicaceae, which contains both the model species Arabidopsis (Arabidopsis thaliana) as well as the domesticated Brassica species. To date, most comparative genomics in the Brassicaceae has largely focused on direct comparisons between Arabidopsis and the species of interest. However, several of the factors that made Arabidopsis ideal for genome sequencing, particularly its reduced genome size and chromosome number (157 Mb, n 5 5; AGI, 2000;Johnston et al., 2005), complicate its use as a standard in comparative genomics. Recent phylogenetic results have demonstrated that genome and chromosome reduction in Arabidopsis are derived characteristics from its close relatives with n 5 8.
In a recent taxonomic reclassification of the Brassicaceae based on molecular phylogenetic results ( Fig.  1), Arabidopsis and its closest relatives, including the n 5 8 species Arabidopsis lyrata and Capsella rubella, were placed within the tribe Camelineae Beilstein et al., 2006). Comparative genetic mapping between Arabidopsis, A. lyrata, and C. rubella found that genome organization of A. lyrata and C. rubella is largely conserved, and that the n 5 5 genome of Arabidopsis is the derived state due to a minimum of three inversions, two reciprocal translocations, and three reciprocal translocation/fusion events (Boivin et al., 2004;Kuittinen et al., 2004;Koch and Kiefer, 2005;Yogeeswaran et al., 2005;. Hence, the n 5 8 karyotype is likely the ancestral state for the Camelinae, and potentially for much of the Brassicaceae since 38% of the family has a base-chromosome number of x 5 8 Warwick and Al-Shehbaz, 2006). Using the hypothesis of an ancestral karyotype of n 5 8 similar to the genome structure of A. lyrata and C. rubella,  investigated a number of chromosomal reductions in the Brassicaceae using comparative chromosome painting (CCP). To do so they applied multicolor chromosome painting using contiguous bacterial artificial chromosome (BAC) pools of Arabidopsis arranged according to the genetic map of A. lyrata and C. rubella. The results revealed that karyotypes with reduced chromosome number (n 5 6, 7) of two taxa from the tribe Camelineae (Neslia, Turritis) and one taxon from Descurainieae (Hornungia) shared conserved chromosome segments that can be related to the ancestral karyotype. Furthermore, the results suggested a common mechanism for chromosome number reduction via a pericentric inversion followed by a reciprocal translocation/fusion event Schranz et al., 2006b).
In addition to comparative mapping done within the Camelineae, there is also a wealth of comparative analyses between the economically important Brassica species from the tribes Brassiceae and Arabidopsis. There has been some difficulty in establishing syntenic relationships between Brassica and Arabidopsis due to the derived nature of the Arabidopsis genome, the paleopolyploid nature of Brassiceae genomes (Lagercrantz, 1998;Lysak et al., 2005;Parkin et al., 2005;Kim et al., 2006), and the relatively large phylogenetic distance between the two genera (Bailey et al., 2006;Beilstein et al., 2006;Koch et al., 2007). Despite these difficulties, a comprehensive comparison between Brassica napus and Arabidopsis identified 21 conserved syntenic blocks shared by B. napus and Arabidopsis genomes, which could be duplicated and rearranged to represent 90% of the B. napus genome. These conserved blocks represent collinear regions that have been maintained since the divergence of the Arabidopsis and Brassica lineages some 20 million years ago (Yang et al., 1999;Koch et al., 2003;Lysak et al., 2005).
An important step toward a unified comparative genomics system across the Brassicaceae can be accomplished by integrating the genomic block system used to show the collinear regions identified between B. napus and Arabidopsis (Parkin et al., 2005) with the concept of the n 5 8 ancestral karyotype shared by A. lyrata and Capsella . We recently proposed a set of 24 genomic blocks (A-X; Schranz et al., 2006b). The order, orientation, and color coding of these blocks are based on their positions in the ancestral karyotype, but using Arabidopsis locus names to define the interval of each block ( Fig. 2A). Often the end points of the blocks correspond to pericentromeric or telomeric regions found in one or more species. This suggests that there may be common mechanisms involved in genome evolution across the crucifers.
Using our system of conserved blocks we can further explore the genomic organization of crucifers within a phylogenetic framework (Schranz et al., 2007). In the aforementioned tribal reclassification of the Brassicaceae , the tribe Camelineae (containing Arabidopsis) is most closely related to the small tribe Halimolobodeae (containing about 40 species) and the Boechereae (with about 110 species; Fig. 1). The Boechereae, made up mostly of species in the genus Boechera, are supported as being a monophyletic assemblage based on the fact that they are almost exclusively a North American group and differ from the Camelineae and Halimolobodeae in having a basechromosome number of x 5 7 rather than x 5 8. Thus, they represent an excellent system to examine karyotype evolution accompanying chromosome number reduction.
The genus Boechera contains an array of morphologically and ecologically diverse taxa that have mainly radiated in alpine, montane, and desert regions of western North America. The group has great potential for studies of ecology and evolution (for review, see Mitchell-Olds, 2001;Dobeš et al., 2007). Numerous studies have analyzed molecular and phenotypic diversity of Boechera species. This includes the molecular evolutionary analysis of gene families (Bishop et al., 2000;Schein et al., 2004;Benderoth et al., 2006), the phylogeography of haplotypes (Dobeš et al., 2004b;Song et al., 2006), the occurrence of supernumary B chromosomes (Böcher, 1951;Sharbel et al., 2004), variation in breeding systems and ploidy (Sharbel and Mitchell-Olds, 2001;Schranz et al., 2005;Schranz et al., 2006a), drought tolerance (Knight et al., 2006), morphological and taxonomical diversity (Rollins, 1993;Al-Shehbaz, 2003;Windham and Al-Shehbaz, 2006), and the evolved responses to pathogen or insect pests (Roy, 1993;Roy and Kirchner, 2000;Jones et al., 2006). The further elucidation of these and other patterns of variation would be greatly aided by the creation and analysis of segregating genetic stocks.
Boechera stricta is one of the most morphologically and molecularly well defined Boechera species, making it a good candidate for genetic and genomic studies. Genetic and molecular analyses indicate that B. stricta is predominantly inbreeding, diploid, and sexual and most accessions form a monophyletic group (referred to as lineage II in Dobeš et al., 2004b;Schranz et al., 2005), whereas many other species in the genus are apomictic, of hybrid origin, and/or triploid (Sharbel and Figure 1. Relationships, number of species, and base-chromosome numbers for the eight tribes of lineage I of the Brassicaceae. A recent taxonomic classification for the Brassicaceae  proposed 24 tribes and recognized three monophyletic lineages (I-III). Within lineage I there is additional support for the monophyly of the three tribes Camelineae (including Arabidopsis, A. lyrata, and C. rubella), Boechereae (including B. stricta), and Halimolobodeae. Most species of the eight tribes within lineage I have a base-chromosome number of x 5 8. Karyotype number reduction has occurred several times, including in the Boechereae (x 5 7).

Figure 2.
Comparative genome organization of the inferred ancestral karyotype (n 5 8) based on published genetic maps of A. lyrata and C. rubella (A), and B. stricta using our F 2 genetic mapping results (B). A, Genome blocks of the ancestral karyotype (AK) are labeled by the letters A to X. Each block is one of eight colors, corresponding to each chromosome. Centromeric positions are indicated by the colored circles. Since only the Arabidopsis genome is currently sequenced, the boundaries of the blocks are defined by their flanking Arabidopsis Genome Initiative At locus names. Each block is considered to be in the upright orientation in the ancestral karyotype. Blocks that are inverted relative to Arabidopsis are indicated by upside-down text of the At locus names. B, Genetic map and genomic blocks for B. stricta. The seven LGs are labeled as BstLG1 to BstLG7. Marker positions (in cM) are shown on the left hand and the corresponding marker name shown on the right hand of each LG. Genomic blocks, as defined above, are arranged onto the LGs based on sequence similarity of the markers to Arabidopsis. Three LGs are completely conserved (Bst4 5 AK4, BstLG6 5 AK6, and BstLG7 5 AK7). Incongruity of color/letter order of the blocks indicates genomic rearrangements in Boechera relative to the ancestral karyotype. Two blocks (A and C) are subdivided in Boechera (A 1 and A 2 ; C 1 and C 2 ). Blocks that are inverted in Boechera (blocks C 1 and C 2 ) are represented by their names being inverted and by a downward pointing arrow. -Olds, 2001;Schranz et al., 2005). Previously we made reciprocal crosses between 92 different Boechera lines from 19 different species to a common B. stricta genetic tester (the line SAD12; Schranz et al., 2005). Most of the intraspecific B. stricta crosses resulted in successful and highly fertile F 1 progeny. Our common crossing parent (SAD12) also was the focus of a partial genomic sequencing project . By analyzing over 39,000 paired end sequences from a collection of medium-sized insert clones, we found that genic and intergenic regions are very similar to Arabidopsis, and conserved microsynteny can be used for rapid identification and cloning of Boechera genes . We have also sequenced several Boechera BACs (Schein et al., 2004;Benderoth et al., 2006), showing that gene order is nearly identical to Arabidopsis. Such conserved microsyteny should facilitate the generation of molecular markers needed to examine patterns of macrosynteny.

Mitchell
In this report, we explore the macrosynteny between B. stricta and other crucifer species by comparative genetic mapping using the genomic block system based on the ancestral karyotype (Schranz et al., 2006b). In doing so, we investigate the karyotype reduction from n 5 8 to n 5 7 that occurred within the tribe Boechereae. Our construction of an F 2 linkage map will further our understanding of patterns of genome evolution in crucifers and facilitate future ecological genomics work in Boechera.

Genetic Markers
By analysis of a large collection of paired endsequenced clones from B. stricta  we have designed nearly 200 molecular markers with high similarity to Arabidopsis to facilitate comparative analyses. A summary of the markers developed and method of scoring is presented in Supplemental Table  S1. For microsatellite markers we preferentially selected long repeats (with an average total repeat length of 33.8 bp made up from an average of 13.7 repeat units) to expedite genotyping (Supplemental Table S2). We designed primers to approximately 250 simple sequence repeat (SSR) loci, of which 58 were placed onto our genetic map (Supplemental Table S2).
The end-sequenced clones were also used to design primers for genes and/or regions of interest, often in conjunction with other ongoing research projects and objectives. These sequences can be divided into three categories: (1) candidate genes of interest (e.g. flowering time and glucosinolate production); (2) random nuclear loci selected as part of a project looking at the patterns of polymorphism in Boechera and other relatives; and (3) targeted regions necessary for synteny comparisons to Arabidopsis. Analysis of these sequences was done to identify single-nucleotide polymorphisms (SNPs) between the two mapping parents. In general, the level of polymorphism was very low, with an average of three polymorphisms per 1,000 bp (B. Song, unpublished data). SNPs were scored using four different methodologies. The majority were scored using a Custom 96-plex GoldenGate Genotyping BeadArray from Illumina. Of the 96 selected SNPs, we obtained genotypic data from 94 of the loci (a 96.8% success rate). This is a much higher success rate than typically seen with custom arrays designed for analysis of human SNPs. We were also very successful in converting SNPs to TaqMan markers as well as cleaved amplified polymorphisms/derived cleaved amplified polymorphisms markers. In addition, nearly all of our molecular markers (193) were scored as codominant, and only three markers were scored as dominant (Con_5393, Pul, and R6.D10).

Linkage Groups
All genotypic data were analyzed with JoinMap 4 using the regression mapping algorithm and Kosambi cM units for genetic linkage analysis (Stam, 1993). Using the most stringent log of the odds (LOD) score threshold of 10.0, all 196 markers resolved into the expected seven linkage groups (LGs). We designate these seven groups as BstLG1 to 7. The genetic map covers a total of 725.8 cM, with an average LG length of 104 cM. The average spacing between markers is 3.9 cM, and no gaps exceed 19.3 cM. The genome size of B. stricta has been estimated to be 264 Mb (approximately 1.73 the genome size of Arabidopsis; Schranz et al., 2006a), giving us a ratio of approximately 362 kb/1 cM.

Comparative Genome Analysis and Evolution
The genetic position and comparative genetic data for each marker is summarized in Table I and includes the position of the marker within the B. stricta genetic map (LG and relative position in cM), homology of the marker (or based on the similarity of the paired end sequence) to Arabidopsis (given to the At gene it matches or the nearest At gene based on genomic homology), and its assignment to a chromosomal block compared to the ancestral karyotype (as defined by letter and color in Schranz et al., 2006b and summarized in Fig. 2).
The genetic map of B. stricta is largely collinear to the ancestral karyotype based on the genetic maps of A. lyrata and Capsella. Furthermore, almost all markers occur and are collinear within their expected genomic blocks (Fig. 2). Only two genomic blocks (blocks A and C) had to be subdivided based on our B. stricta mapping results due to a single pericentric inversion (see below). In the ancestral karyotype block A was defined as the interval from At1g01560 to At1g19330. In B. stricta this interval is subdivided into block A 1 on Bst LG1 (At1g01560-At1g13640) and block A 2 on Bst LG2 (AT1G15190-At1g19330). The boundary between block A and block B was delineated based on mapping results from B. napus (Parkin et al., 2005) and is one of the Table I. Genetic mapping and molecular marker information for B. stricta F 2 map ;, Most significant BLAST score to an intergenic region, position of nearest gene is then given.   few genomic block boundaries that is not defined by a centromeric or telomeric region in Arabidopsis or the ancestral karyotype. Block C must also be divided into two. Block C 1 on BstLG1 spans the interval from AT1g43245 to At1g51310 and block C 2 on BstC2 spans the interval from At1g55490 to At1g56145. The only block that could not be definitively placed on the map was block D, with one marker with homology to block D on BstLG1 and one marker on BstLG7. There were a total of eight markers whose placement on the map did not agree with expectations based on Arabidopsis. One of these, PhyE, mapped to the same position as a close homolog, PhyA, on LG1. Hence, this may represent a case of cross hybridization of the primers to a related sequence. Another gene, FLC2, mapped to LG5 rather than to its expected position on LG6. However, we have cloned two copies of the FLC gene in Boechera (M.E. Schranz, unpublished data) and hence the duplicate copy of this locus has likely been integrated at this new genomic position.
We specifically designed a number of probes from the B. stricta end-sequence collection that had no significant similarity to Arabidopsis. Interestingly, several of these (Bst011023 on BstLG2, R6.A09 on BstLG5, and Bst002440 on BstLG6) map to the likely pericentromeric regions based on comparison to the ancestral karyotype. The observation that these markers occur within regions with low levels of recombination on their respective LGs supports the possibility that these markers are specific to pericentromeric regions.
Based on comparison to the ancestral karyotype of n 5 8, we can infer several details of B. stricta (n 5 7) genome structure and evolution (Figs. 2 and 3). First, three of the B. stricta LGs are almost completely collinear and unchanged when compared to those in the ancestral karyotype (BstLG4 5 AK4, BstLG6 5 AK6,  LG as shown in Figure 2B.  and BstLG7 5 AK7; Fig. 2). This result further validates the derived nature of the Arabidopsis karyotype (n 5 5) and confirms the existence of a common ancestral karyotype (n 5 8), as many of the blocks that are collinear in A. lyrata, Capsella, and now Boechera are on different chromosomes in Arabidopsis. For example, blocks S and T are collinear in A. lyrata, Capsella, and Boechera, but not in Arabidopsis where they are found on chromosomes At4 and At5, respectively . Second, we can surmise that the other four B. stricta LGs were derived by a series of rearrangements that can account for the karyotype reduction from n 5 8 to n 5 7. In total, there likely was a pericentric inversion, a chromosomal fusion, and two reciprocal translocations involving five of the eight chromosomes in the ancestral karyotype (Fig. 3).
While we don't know the evolutionary sequence of these events, we can hypothesize a parsimonious series of changes. A reciprocal translocation occurring between the centromeric regions of AK3 and AK8 (Fig. 3A) would result in the formation of BstLG3 (with block F and G now fused with W and X) and a second chromosome (made up of blocks V and H). A chromosomal fusion between the telomeric regions of block H and block K (of AK5) would result in the formation of BstLG5 (Fig. 3A). The combination of the reciprocal translocation and chromosomal fusion events would account for the karyotype reduction from n 5 8 to n 5 7. However, our mapping data do not allow us to definitively resolve which centromere was lost (AK3, AK8, or AK5), or by what mechanism this loss occurred. It may well be that there were additional chromosomal rearrangements (such as one or more pericentric inversions).
A pericentric inversion of AK1 with one breakpoint between blocks A 1 and A 2 and the other between blocks C 1 and C 2 would result in a rearranged chromosome (Fig. 3B). A reciprocal translocation between the centromeric regions of the rearranged AK1 and AK2 would result in BstLG1 (including blocks D, C 1 , and Figure 3. A parsimonious model of genomic changes describing the evolution of the B. stricta (n 5 7) genome from the ancestral karyotype (n 5 8). A, A reciprocal translocation occurring between the centromeric regions of AK3 and AK8 would result in the formation of BstLG3 (with blocks F and G now fused with W and X) and a second chromosome (made up of blocks V and H). A chromosomal fusion between the telomeric regions of block H and block K (of AK5) would result in the formation of BstLG5. It is equally possible that the chromosomal fusion between block H (of AK3) and block K (of AK5) occurred before the reciprocal translocation event. The combination of the reciprocal translocation and chromosomal fusion events would account for the karyotype reduction from n 5 8 to n 5 7. B, A pericentric inversion of AK1 with one breakpoint between blocks A 1 and A 2 and the other between blocks C 1 and C 2 would result in a rearranged chromosome. A reciprocal translocation between the centromeric regions of the rearranged AK1 and AK2 would result in BstLG1 (including blocks D, C 1 , and A 1 ) and BstLG2 (including blocks C 2 , A 2 , B, and E). [See online article for color version of this figure.] A 1 ) and BstLG2 (including blocks C 2 , A 2 , B, and E). Again, we cannot rule out the possibility of additional changes.
Based on comparison to the ancestral karyotype, we can also conjecture the centromeric positions in Boechera (Fig. 2). Centromeres with conserved positions could lie between the following sets of blocks: I and J on BstLG4, L and M on BstLG5, P and Q on BstLG6, and S and T on BstLG7. Centromeres that are potentially rearranged in Boechera relative to the ancestral karyotype might lie between blocks B and E on BstLG2 and between blocks G and W on BstLG3.

Segregation Distortion
By testing for deviation from the expected Mendelian 1:2:1 ratio of markers we found 25% were significant at the 0.05 significance level. This is less than the approximately 35% segregation distortion seen in A. lyrata (Kuittinen et al., 2004), but higher than that observed in many intraspecific crosses done in crop plants (Jenczewski et al., 1997). Most distorted loci (42/49) were due to a low frequency of observed homozygous SAD12 genotypes, despite the fact that the SAD12 line was the maternal parent. Segregation distortion was nonrandom, with most distorted loci observed in blocks of multiple colinear markers. In total, there were seven regions in which more than one marker was distorted. There were two particularly large blocks of distortion (on the bottom of BstLG6 and the top of BstLG7) together containing 32 of the 49 distorted loci. We did not find any markers showing segregation distortion to significantly deviate between the observed and expected genotypic frequencies at the 0.10 significance level, suggesting selection on the gametic stage.

Recombination Suppression
BstLG1 showed very low levels of recombination over much of its length, with most recombination only occurring near the ends of the LG. Most of the markers on this LG have homology to Arabidopsis chromosome 1. The equivalent region in Arabidopsis covers at a minimum 6.7 Mb. There is more recombination seen within these syntenic intervals in the published maps of both Arabidopsis and in A. lyrata. The low levels of recombination seen in BstLG1 makes marker order and placement very ambiguous. Such a lack of recombination across the interval in B. stricta could be due to an intraspecific inversion present between the two parental types or some other mechanism suppressing recombination, such as heterochromatization.

DISCUSSION
The dicot family Brassicaceae is an excellent group in which to examine patterns of genome and sequence evolution (Schranz et al., 2007). Not only does it contain the model species Arabidopsis and the domesti-cated Brassica crops, but it is also the focus of several complete genome-sequencing projects, including A. lyrata, C. rubella, Thellungiella halophila (5Eutrema halophila), and B. rapa. Added to this is the partial genome sequencing data of Arabis alpina (N. Warthmann and D. Weigel, personal communication) and B. stricta (approximately 0.153 genome coverage; Windsor et al., 2006). Comparative genetic mapping provides the framework for analysis of these genomic sequences. Comparative mapping has found significant synteny in genomic blocks conserved across a number of species, suggesting that there are common mechanisms involved in genome evolution across the family (for review, see Schranz et al., 2006b).
In this study, we contribute to our understanding of crucifer genome evolution by comparative analysis of our F 2 linkage map for B. stricta. Overall, our genetic map is highly collinear with the n 5 8 genetic maps from A. lyrata and Capsella. However, we found that the genome evolution and karyotype reduction to n 5 7 in B. stricta did not occur by one simple chromosomal fusion event, but rather involved several chromosomal changes including a pericentric inversion, a chromosomal fusion, and two reciprocal translocations. Most of these chromosomal changes can simply be represented with our genomic block system. Although the evolution from n 5 8 to n 5 7 is more complex than one simple fusion it is much easier to interpret than if compared to Arabidopsis (Schranz et al., 2006b). Our results support the hypothesis that there are common mechanisms involved in crucifer genome evolution such that changes tend to occur at certain points while maintaining most of the genome in large syntenic blocks. In addition, we found one LG that has a large degree of recombination suppression as well as two regions showing significant segregation distortion. Our genetic mapping will also facilitate future analysis of quantitative trait variation within Boechera.

Markers and Genetic Map Construction
Previously, we had end sequenced a large number of B. stricta l clones to investigate sequence similarity and microsynteny with Arabidopsis . In this study, we utilized these sequences to develop nearly 200 SSR and SNP molecular markers for genetic mapping. We successfully genotyped 94 SNPs simultaneously using a custom Illumina bead array. The Illumina bead array technology has been extensively used in studies of human polymorphism, including the Human HapMap project (Altshuler et al., 2005). Only recently has it been used for genetic mapping in plants (Rostoks et al., 2006). Nearly all of our SSR and SNP markers were developed from sequences with high similarity to Arabidopsis, allowing us to investigate patterns of macrosynteny between these two species and other crucifer genomes.
All of our markers could be unambiguously placed into one of the seven LGs using the most stringent LOD score threshold of 10. Our overall sample size, number of markers, marker density, and genome coverage is higher than that in genetic maps constructed for either C. rubella or A. lyrata (Boivin et al., 2004;Kuittinen et al., 2004;Yogeeswaran et al., 2005). With an average distance between markers of less than 4 cM and with no gaps greater than 20 cM, our genetic map provides excellent opportunities for precise quantitative trait locus studies, positional cloning, and comparative genomic analyses.

Comparative Genomics and Genome Evolution of Boechera
In a recent review, we proposed a framework for comparative genomics for the Brassicaceae based on a set of 24 conserved syntenic blocks (Schranz et al., 2006b). Herein, we demonstrate the utility of these genomic blocks for illustrating and understanding the genome evolution of B. stricta from the tribe Boechereae.
Recent molecular phylogenetic analyses have shown that the tribe Boechereae is closely related to the potentially polypheletic tribe Camelineae, containing both the genus Arabidopsis and the genus Capsella (Bailey et al., 2006;Beilstein et al., 2006;Koch et al., 2007). The two tribes along with several other tribes are strongly supported as a monophyletic assemblage within the Brassicaceae (referred to as lineage I; Fig. 1; Bailey et al., 2006;Beilstein et al., 2006;Koch et al., 2007). The basechromosome number for most of the tribes within lineage I is x 5 8 ( Fig. 1; Al . There are several examples of karyotype reduction , including the reduction of Arabidopsis to n 5 5, and increases in chromosome number often due to polyploidy within lineage I (Marhold and Lihova, 2006;Warwick and Al-Shehbaz, 2006). Interestingly, the entire tribe Boechereae shares the base-chromosome number of x 5 7, presumably from a reduction from x 5 8 seen in other members of lineage I. A priori, one might expect that this reduction occurred via a single chromosomal fusion event. However, our results demonstrate that the karyotype evolution of B. stricta was more complex, involving at a minimum a pericentric inversion, a chromosomal fusion, and two reciprocal translocations (Fig. 3). The proposed types and likely mechanisms of changes are very similar to other examples of karyotype reduction in the Brassicaceae . This study suggests that rearrangements often occur at pericentromeres or at telomeres containing nucleolus organizer regions (NORs).
In A. lyrata there are a total of five NORs. NORs on different chromosomes associate nonrandomly in interphase nuclei of both Arabidopsis and A. lyrata (Armstrong et al., 2001;Pecinka et al., 2004;Berr et al., 2006). The pairing of NORs of nonhomologous chromosomes has been hypothesized to play an important role in facilitating rearrangements between chromosomes . In B. stricta, there is only a single terminal NOR (Kantama, 2005), hence, if the ancestor of B. stricta also contained five NORs than four have been lost. By comparison of the maps of A. lyrata and B. stricta we can hypothesize that a conserved NOR could be positioned at the top of BstLG4 (above block I). Intriguingly, the other NORs seen in A. lyrata are all at the sites of rearrangements in B. stricta. Of particular importance are the NORs at the ends of block H (on AL3) and block K (on AL5). This is the site of chromosomal fusion seen on BstLG5, suggesting a possible mechanism for chromosomal fusion by recombination of the two NORs. This process would also lead to the elimination of these two NORs. Another NOR at the bottom of block C (on AL1) could have been lost during the pericentromeric inversion event occurring in B. stricta. Finally, there is a NOR at the top of block S (on AL 7). On BstLG7 we have a single marker coming from block D above block S, suggesting this NOR also was lost. Support for the hypothesis that chromosomal rearrangements often occur at NORs comes from the analysis of telomerase-deficient Arabidopsis lines where frequent chromosomal fusions between NORs have been observed (Siroky et al., 2003).
We also have detected a number of rearrangements that likely occurred between pericentromeric regions of the ancestral karyotype during the evolution of the B. stricta genome. Specifically, we hypothesize reciprocal translocations between the pericentromeric regions of AK1 and AK2 and between AK3 and AK8. Rearrangements involving pericentromeric regions have also been observed in rearrangements of other crucifer species, including Arabidopsis Kawabe et al., 2006;.
NORs and pericentromeric regions are known to be highly dynamic genomic regions, characterized by long stretches of repetitive DNA (Hall et al., 2004;Fajkus et al., 2005). It is plausible that nonhomologous chromosome association via pairing of repetitive elements in either pericentromeric regions or NORs could be involved in recombination events leading to chromosomal rearrangements. Our results further confirm a generalized set of conserved mechanisms contributing to Brassicaceae genome evolution. One of the major conclusions is that only limited regions of the genome are involved in rearrangements, whereas most of the genome is maintained in discrete and conserved blocks.

Recombination Suppression of BstLG1
BstLG1 has suppressed recombination within the central region of the chromosome. Most of the markers correspond to genic, and not pericentromeric, regions found in Arabidopsis chromosome 1. One possibility to explain the recombination suppression is an intraspecific inversion within BstLG1 between the two parental lines. It is important to note that this would be a separate inversion event than that shown in Figure  3B. Additionally, we do not know whether such intraspecific chromosomal polymorphisms exist within B. stricta. Another possibility is that recombination is suppressed because one of the parental BstLG1 chromosomes contains large tracts of heterochromatin.
Heterochromatic chromosomes (Het) and supernumary heterochromatic-B chromosomes have been detected in asexually reproducing (apomictic) Boechera lines (Sharbel et al., 2004;Kantama, 2005;Sharbel et al., 2005). The Het chromosome is derived from B. stricta as based on hybridization of pericentromeric repeats (Kantama, 2005;Schranz et al., 2006a) and genetic analysis has demonstrated that the Het chromosome can be crossed into B. stricta without conferring a dominant apomictic phenotype (Schranz et al., 2006a). These two results suggest that Het chromosomes might have originated and been present within sexually reproducing B. stricta lineages. From cytological studies we know that the maternal B. stricta parent, SAD12, does not carry the Het chromosome (Schranz et al., 2006a), but, potentially the parental line, LTM, might bear a Het chromosome that was transmitted to the F 1 plant. Future cytogenetic studies will be needed to clarify the reason(s) for the recombination suppression of the BstLG1.

Segregation Distortion
A total of 25% of our markers showed significant transmission ratio distortions. This is only slightly higher than the average ratio for intraspecific crosses of agricultural species (18.4% 6 11.0%; Jenczewski et al., 1997). Several recent analyses of interspecific crosses of wild species have reported much higher levels of distortion (Hall and Willis, 2005;Bratteler et al., 2006), including A. lyrata with an average of 35% (Kuittinen et al., 2004). Hence, the degree of segregation distortion seen for B. stricta is within expectations. Most markers were distorted because of an underrepresentation of SAD12 homozygous genotypes. This was surprising since SAD12 was the maternal parent, suggesting that cytoplasmic incompatibility is not responsible for the distortion.
Segregation distortion can occur at two levels, reflecting selection at either gametic or genotypic levels. Prezygotic selection at the gametic level causes deviation from 1:1 allelic ratios, producing genotypic ratios of p 2 :2pq:q 2 , where p and q indicate allele frequencies after selection. In contrast, postzygotic selection causes deviation from this predicted genotypic ratio, for example if heterozygotes have higher survival or one homozygote has reduced viability. At each distorted locus we calculated allele frequencies and compared observed and predicted genotypic ratios. Results fit a model of prezygotic gametic selection that favors LTM alleles in several genomic regions. We found no evidence for postzygotic selection favoring particular genotypes. Thus, there is no indication of heterotic influences on plant viability.
Gametic selection can occur via competition between pollen and/or ovules with different genotypes caused by self-incompatability loci, inbreeding depression due to genetic load, or genetic isolation evolved between the parental populations. B. stricta is a highly self-compatible species with very high levels of homozygosity (Song et al., 2006), thus, we do not expect either self-incompatibility or inbreeding depression to be the cause of the segregation distortion. However, there could be significant genetic isolation between these two populations. The two populations are geographically isolated, being approximately 1,000 km apart. Also, the habitats of the two populations are quite different (Fig. 4); the SAD12 locality is a sagebrush grassland in a river valley, whereas the LTM locality is a subalpine meadow. The two sites differ in levels of precipitation and temperature. Additionally, the two parental genotypes cluster into different groups based on STRUCTURE analyses of 229 single-copy nucleotide loci and the Wright's fixation index value between the two groups is 0.4 (B. Song and T. Mitchell-Olds, unpublished data). Finally, we also know that the two populations have different chloroplast haplotypes. SAD12 has chloroplast haplotype DG and LTM has haplotype AH based on analysis of the trnL intron-trnL/F IGS region (Schranz et al., 2005). The LTM haplotype (AH) is one of two main haplotypes found to have likely colonized formerly glaciated areas of North America, with the AH haplotype spreading into the Northwest and AS spreading in the Southwest, Northcentral, and Northeast (Dobeš et al., 2004b). The SAD12 haplotype (DG) is rare and is a geographically restricted haplotype that may represent an isolated glacial refugial population (C. Dobeš and M. Koch, personal communication).

Future Directions
CCP using multicolored BACs arranged according to the ancestral karyotype  would be invaluable for resolving several remaining questions about genome evolution in Boechera. First, CCP would confirm the order and orientation of the genomic blocks in B. stricta and help resolve the uncertain placement of block D. Second, CCP could identify rearrangements that were not detected by our mapping due to low marker density or that occur in regions of low recombination (such as centromeres or in BstLG1). This will help us resolve whether or not there is recombination suppression in BstLG1 due to an intraspecific chromosomal inversion. Third, CCP could identify the location of centromeres, NORs, and heterochromatic regions of our Boechera genetic map. This information will be important in future quantitative trait analyses. Locating centromeres would also resolve which centromere has been lost (potentially one between blocks V and H) and by what mechanism this occurred. Finally, CCP could be used to address whether any of the genomic rearrangements identified in our genetic mapping of B. stricta are unique to this species or if they are shared with all members of the Boechereae tribe (with x 5 7) and/or with members of the closely related tribe Halimolobodeae (with x 5 8). We also could test for shared breakpoints by tracking a handful of markers spanning the block boundaries in F 2 families in crosses made in other taxa.
Comparative genetic mapping and CCP could also help resolve conflicting phylogenetic signals by tracking specific and rarely occurring genomic changes (Lysak and Lexer, 2006). For example, such approaches could be used to identify monophyletic groups within the polyphyletic Camelineae (Bailey et al., 2006;Koch et al., 2007) and help resolve ambiguous relationships within the genus Boechera (Schranz et al., 2005).
Our construction of a B. stricta genetic map will also greatly facilitate our analyses of ecologically important quantitative variation and positional cloning of the underlying genes segregating in our cross. There is substantial variation for a number of important traits, including glucosinolate content, flowering time, and drought tolerance that should be amenable to quantitative analyses. Our placement of many candidate gene markers onto our genetic map will facilitate these analyses. Furthermore, we have already advanced our mapping population to the F 5 generation and will soon have recombinant inbred lines for this perennial species. Finally, our success in exploiting microsynteny between Boechera and Arabidopsis for sequencing genes of interest should allow for successful positional cloning of quantitative trait loci (Schein et al., 2004;Benderoth et al., 2006;Windsor et al., 2006).

Plant Materials and DNA Isolations
A genetic cross was made between two highly inbred lines of Boechera stricta (Graham) Al-Shehbaz. The maternal line SAD12 was collected in Gunninson County, Colorado by Dr. Bitty Roy and the paternal line LTM was collected in Lemhi County, Idaho (Fig. 4). Details about the plant populations, locations, and the genetic crossing have been described previously (Schranz et al., 2005). A total of 192 F 2 lines were grown, with seeds placed on moist filter paper in sealed petri dishes and cold treated at 4°C for 3 weeks in the dark. The petri dishes were then transferred to a growth chamber until seed germination. The germinated seedlings were then transferred to 96-well flats. Seedlings were grown for 4 weeks and then transplanted to pots (11 3 11 3 13 cm). The plants were grown in a controlled growth room under long-day conditions (16 h light and 8 h dark).
DNA from each F 2 line was isolated using the Qiagen DNeasy Plant Mini kit and the Qiagen Genomic-tip 100/G kit (Qiagen) was used for the two parental genotypes.

Genetic Marker Development and Analysis
We previously reported our analysis of approximately 39,000 paired end sequences from the SAD12 genotype of B. stricta  that was used as the maternal parent in our genetic cross. We utilized these endsequenced clones to develop a number of different genetic marker systems. Foremost, we used sequences that had strong similarity/homology to Arabidopsis (Arabidopsis thaliana) to facilitate comparative analyses. In several instances, we used the similarity data from the paired end sequences to help infer homology. A summary of the markers developed is presented in Supplemental Table S1. This includes the name of the marker, the type of polymorphism (SSRs or SNPs), the method by which they were analyzed (sequencer, gel based, cleaved amplified polymorphism, derived cleaved amplified polymorphism, Illumina bead array, or TaqMan probe), the primers used to detect and/or score the polymorphism, and the l clone from which it was derived.
To identify SSRs to use as molecular markers we screened the endsequenced clones for SSRs (microsatellites) using the SPUTNIK program (Abajian, 1994). We allowed for no errors (insertions, mismatches, and deletions) to the repeat. Primers that flanked the repeat element and had an amplicon ,300 bp were designed using PRIMER 3 software (Rozen and Skaletsky, 2000;Supplemental Table S2). In addition, we used 10 microsatellite loci reported in previous studies (Clauss et al., 2002;Dobeš et al., 2004a;Schranz et al., 2005;Song et al., 2006). Most microsatellites were scored by analysis on 4% MetaPhor Tris-acetate EDTA agarose gels (Cambrix Bio Science). Primers used in previous studies were run as described previously (Schranz et al., 2005;Song et al., 2006).
Second, we used the end-sequenced clones to design primers for genes and/or regions of interest, often in conjunction with other ongoing research projects and objectives. Primer pairs were designed from the end-sequenced clones using either PRIMER 3 software (Rozen and Skaletsky, 2000) or with PRIMACLADE software when the end sequences were aligned to Arabidopsis genomic sequence (Gadberry et al., 2005). The primers were used to amplify and sequence genomic DNA of the mapping parents (SAD12 and LTM) to identify SNPs. PCR and DNA sequencing was done as previously described  or by Genaissance Pharmaceuticals Inc.
SNPs were identified and formatted en masse using SnpDetector (A.J. Windsor, unpublished data), a script written in Python (http://www.python. org). To identify candidate SNPs, SnpDetector performs pairwise alignments between orthologous sequences by calling the National Center for Biotechnology Information's bl2seq (Altschul et al., 1990;Zhang and Madden, 1997) and/or EMBOSS' water (Rice et al., 2000) programs. Subsequently, SnpDetector parses the alignments generated, detects candidate SNPs, and screens the candidate SNPs relative to the requirements needed for genotyping.
Identified SNPs were scored using four different methodologies. A total of 94 SNPs were analyzed using a Custom 96-plex GoldenGate Genotyping BeadArray from Illumina Inc. Second, 16 SNPs were scored using TaqMan probes (Applied Biosystems). Cleaved amplified polymorphisms were identified using both the SNP2CAPS (Thiel et al., 2004) and BlastDigester (Ilic et al., 2004) programs. Finally, derived cleaved amplified polymorphisms (Michaels and Amasino, 1998) were designed using the derived cleaved amplified polymorphism Finder 2.0 (Neff et al., 2002).

Genetic Map Construction
We used the JoinMap v4 program using the weighted least-squares method for map construction (Stam, 1993). A LOD score threshold of 10 was used to assign all markers to one of seven LGs.

Segregation Distortion
To test for segregation distortion, we examined the fit of each marker to the expected 1:2:1 ratio with x 2 tests using the JoinMap program (Stam, 1993). For those markers that showed significant deviations in segregation at the 0.05 level, we used x 2 tests to investigate whether the observed genotypic frequencies differed from the expected genotypic frequencies (e.g. are the marker classes in Hardy-Weinberg equilibrium). If the allele frequencies significantly deviate from the expected p 2 1 2pq 1 q 2 this is taken as evidence for selection on the zygotic stage. If the results are nonsignificant this could be due to either selection at the gametic stage or on the zygotic stage with additive fitness values.
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers DU667459 to DU708532.

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