Chromosome rearrangements in evolution: From gene order to genome sequence and back.

S ince Sturtevant’s 1926 genetic proof of inversions (1) and Wright and Haldane’s discovery of conserved linkages (2), classical geneticists have compared pairs of linkage maps to infer chromosomal rearrangements and define evolutionarily conserved chromosomal segments. The data have been limited to whatever sample of the gene complement has been mapped in both sequences, and by statistical uncertainty of gene order and map distances. It is a far cry from the beads-on-a-string maps underlying, for example, the first systematic comparative mapping in mammals (3) to today’s comprehensive primary data for documenting the structural evolution of mammalian genomes: billions of base pairs of nearly complete genomic sequences, with their vast intergenic distances, highly dispersed upstream and downstream regulatory elements, overlapping genes, alternative exons, somatic rearrangements, paralogs, gene families large and small, apparently randomly scattered pseudogenes, massive assemblydefeating arrays of repetitive sequences of all sorts, great volumes of transposoninserted material, and confusing recent accretions of highly paralogous material in subtelomeric (4) and pericentromeric regions (5). Making sense of these diverse elements in the genome, understanding how they are organized within the genome of each species, and characterizing the changes in genome organization during evolution are critical problems in comparative genomics. Kent et al. (6), in this issue of PNAS, and Pevzner and Tesler (7, 8) now add new insight into our knowledge of evolutionarily conserved genome structure. We are obliged to greatly expand the neat repertoire of classical evolutionary processes affecting genome structure: inversion and reciprocal translocation; chromosome fusion and fission; gene, segment, and chromosomal duplication and loss; polyploidization [even in mammals (9)] and return to diploidy, to account not only for the various highly productive mechanisms for inserting external material, for the proliferation of repetitive sequencing, for massive ongoing sequence conversion, e.g., in the Y chromosome (10), but also for the complex patterning of frequency and size of the chromosomal fragments involved in the more traditional processes. Even if we wish to describe only the major structural changes that differentiate two species, genomic sequence alignment to find corresponding orthologous segments must overcome a bewildering inventory of very short segments and local rearrangements, multiply aligned regions, and unaligned zones, which represent the ‘‘noise’’ in this analysis, although they are, of course, of vital interest in their own right. Existing methodology for discerning conserved segments in gene orders despite small rearrangements and error (11) are grossly inadequate for ‘‘cleaning up’’ comparisons at the genome sequence level.