Molecular and evolutionary characteristics of the fraction of human alpha satellite DNA associated with CENP-A at the centromeres of chromosomes 1, 5, 19, and 21

In this paper, we show that the molecular mechanisms by which these arrays evolve are identical in multiple chromosomes: i) accumulation of crossovers that homogenise and expand the arrays into different domains and subdomains that are mostly unshared between homologues and ii) sporadic mutations and conversion events that simultaneously differentiate them from one another. Individual arrays are affected by these mechanisms to different extents that presumably increase with time. Repeats associated with CENP-A, where the centromere is formed, are subjected to the same evolutionary mechanisms, but constitute minor subsets that exhibit subtle sequence differences from those of the bulk repeats. While the DNA sequence per se is not essential for centromere localisation along an array, it appears that certain sequences can be selected against. On chromosomes 1 and 19, which are more affected by the above evolutionary mechanisms than are chromosomes 21 and 5, CENP-A associated repeats were also recovered from a second homogeneous array present on each chromosome. This could be a way for chromosomes to sustain mitosis and meiosis when the normal centromere locus is ineluctably undermined by the above mechanisms.We discuss, in light of these observations, possible scenarios for the normal evolutionary fates of human centromeric regions.Although human alpha satellite DNA sequences have been studied for decades, a number of their structural and evolutionary characteristics remain obscure. It is generally accepted that sequences constituting highly homogeneous arrays, including those within which the active centromere is formed, evolve in a concerted way [1]. In view of this concerted evolution, many authors have supposed that the repeats are homogenised with high efficiency, both intra-chromosomally and between homologues. At the same time, it has been shown that meiotic recombination is highly suppressed in the centromeric chromosomal regions [2-5]. Indeed, it was recently