Comparative genome analyses of Arabidopsis spp.: Inferring chromosomal rearrangement events in the evolutionary history of A. thaliana

Comparative genome analysis can be a powerful tool to address questions pertaining to the evolution of the structure of genomes (Paterson and Bennetzen 2001). In plant families that include model taxa, such as the Brassicaceae, the Solanaceae, and the Poaceae, comparative mapping studies allow evolutionary inferences related to the nature and number of rearrangements that distinguish the genomes of the model species from those of their less well-studied relatives (Bonierbale et al. 1988; Ahn and Tanksley 1993; Prince et al. 1993; Chittendan et al. 1994; Kurata et al. 1994; Periera et al. 1994; Lagercrantz and Lydiate 1996; Gale and Devos 1998; Lagercrantz 1998; Doganlar et al. 2002). Not only do such inferences facilitate the transfer of knowledge from model taxa to their relatives, but they also provide insight into the process of genome restructuring that can lead to reproductive isolation and ultimately speciation (Rieseberg 2001; Hall et al. 2002). The larger the number of taxa analyzed within a family, the more comprehensive the information is for inferring the nature of ancestral genomes and the evolutionary history that led to the speciation of the members of that family. The Brassicaceae (or Cruciferae) is a dicot family divided into 13 tribes with a total of 360 genera (Al-Shehbaz 1973) including the agronomically important genus Brassica and the model species Arabidopsis thaliana. The near-complete annotation of the genomic sequence of A. thaliana, including detailed information of physical map position, copy number of genes and intergenic sequences, and location of duplicated chromosomal segments (The Arabidopsis Genome Initiative 2000; Blanc et al. 2000; Vision et al. 2000; Bowers et al. 2003; Ermolaeva et al. 2003; Raes et al. 2003) provides a solid foundation for comparative mapping studies within this family. A. thaliana, prized for its small genome, is actually an anomaly within the tribe Sisymbrieae of the Brassicaceae. A base chromosome number of 8 is inferred to be the ancestral state in the tribe (Koch et al. 1999), and reduction in chromosome number to n = 5–7 is assumed to be a derived state (Koch et al. 1999, 2000). A. thaliana, with its five chromosomes, is the single most extreme example of this genome reduction in the tribe. The nature and rate of occurrence of the chromosomal events that generated the reduced A. thaliana genome from the ancestral n = 8 genome are not well understood. As many as 90 rearrangements are estimated to differentiate Brassica nigra [n = 8] from A. thaliana (Lagercrantz 1998). Two recent mapping studies in which the A. thaliana genome was compared with the Capsella genome (n = 8; Boivin et al. 2004) and with the Arabidopsis lyrata ssp. petraea genome (A. l. petraea; n = 8; Kuittinen et al. 2004), suggest that genome reduction was the result of chromosomal fusion rather than chromosomal elimination. Whatever the nature of these rearrangements, they must have occurred ∼5 Mya, the estimated time of divergence (Koch et al. 2000, 2001; Kuittinen and Aguade 2000) between A. thaliana and its closest known relative A. lyrata (n = 8, genome ∼116%–123% larger than that of A. thaliana; Dart et al. 2004; Schmuths et al. 2004). Further insight into the evolutionary history of the A. thaliana genome is likely to be gained from a three-way comparison of the genomes of A. thaliana and its close relatives, A. lyrata and Capsella spp. In the study reported here, we used the wild, perennial, outcrossing Arabidopsis lyrata subspecies lyrata (hereafter A. l. lyrata; n = 8, diploid; Dart et al. 2004) which occurs in North America (in contrast to the A. l. petraea subspecies which occurs in Europe; Ramos-Onsins et al. 2004), to construct a low-density genetic map with 104 restriction fragment length polymorphism (RFLP) markers distributed on eight linkage groups, and we compared this map to the A. thaliana genome. We then performed comparative statistical analyses of rearrangements observed in A. thaliana versus A. l. lyrata and A. thaliana versus Capsella (based on data from Boivin et al. 2004), using an extension of the Bayesian method reported by Durrett et al. (2004). This method assigns probability estimates to the parsimony solutions as well as alternative scenarios, and is therefore more powerful than parsimony alone to infer the most probable events that underlie genome evolution. Based on these studies, the minimum number of major rearrangements that distinguish A. thaliana from the n = 8 genomes and their rate of occurrence are inferred.