Using comparative FISH and genomics, we have studied and compared the evolution of chromosome 3 in primates and two human neocentromere cases on the long arm of this chromosome. Our results show that one of the human neocentromere cases maps to the same 3q26 chromosomal region where a new centromere emerged in a common ancestor of the Old World monkeys ∼25-40 million years ago. Similarly, the locus in which a new centromere was seeded in the great apes' ancestor was orthologous to the site in which a new centromere emerged in the New World monkeys' ancestor. These data suggest the recurrent use of longstanding latent centromeres and that there is an inherent potential of these regions to form centromeres. The second human neocentromere case (3q24) revealed unprecedented features. The neocentromere emergence was not accompanied by any chromosomal rearrangement that usually triggers these events. Instead, it involved the functional inactivation of the normal centromere, and was present in an otherwise phenotypically normal individual who transmitted this unusual chromosome to the next generation. We propose that the formation of neocentromeres in humans and the emergence of new centromeres during the course of evolution share a common mechanism

A neocentromere at 3q26 was observed in a father and his daughter on a chromosome 3 with deleted centromeric region. No alpha-satellite DNA was detectable at the 3q26 neocentromere, but it was weakly positive with anticentromere (CREST) antibodies. Electron microscopy showed that the neocentromere formed microtubule-associated kinetochores with normal morphology and of the same size as the kinetochores of other large chromosomes. The deleted centromere formed a small linear marker chromosome that reacted strongly with anticentromere antibodies, but showed reduced kinetochore size. The 3q26 neokinetochore was stable under adverse growth conditions, which often caused kinetochore loss in the original 3-centromere on the small marker.

Eukaryotic centromeres are composed of centromere DNA and the multiple proteins directly or indirectly associated with it. One important DNA-binding protein in the centromere is DNA topoisomerase II (topo II). In the genome in general, topo II has two functions, one structural and one enzymatic, the latter catalyzing DNA strand-passage reactions. It has been demonstrated that topo II accumulates at centromeres during the first part of mitosis, and disappears again at anaphase, but it has not been clear whether it serves a structural or an enzymatic function at the centromere. To investigate this issue, we developed the topo II-induced self-primed in situ assay (Topo-SPRINS). In this assay, DNA breaks created by topo II are stabilized with the topo II inhibitor VM-26 in vivo, and used as 'primers' for localized DNA synthesis in vitro. The assay revealed that topo II has enzymatic activity at mitotic centromeres and that the activity is relatively constant across centromeres. Furthermore, the activity was observed at a neocentromere, and, in multicentric chromosomes, the activity was restricted to the active centromere. The topo II activity is thus selectively present at functioning centromeres, indicating that it plays a role in mitotic centromere function.

Two dicentric human chromosomes were investigated with light and electron microscopic techniques. One chromosome, with a translocation tdic(5;13)(p12;p12), behaved as a dicentric in about half the cells: it had two primary constrictions; C- and Cd-banding showed two centromeres; and the CREST antikinetochore antibody reacted with the two centromeres with equal affinity. Electron microscopic analysis of sectioned metaphases showed that the dicentric could develop kinetochores at both centromeres simultaneously. The other dicentric chromosome, tdic(21;21)(q22;q22), occasionally showed two primary constrictions, but both C- and Cd-banding distinguished between an active and an inactive centromere, and the CREST antibody reacted only weakly with the inactive centromere. Electron microscopy showed kinetochore development at only one centromere.

A stable, dicentric human chromosome, which is known from light microscopy to show a 50:50 distribution between monocentric/dicentric appearance, was examined by conventional electron microscopy and after labelling the centromere with anticentromere antibodies from CREST serum. Both centromeres of the chromosome developed kinetochores whether in monocentric or dicentric configuration. The eight monocentrics observed had all developed kinetochores at the centromere outside the constriction; at least six of them also had kinetochores at the centromere in the constriction. The dicentrics from glutaraldehyde fixed cells had spindle microtubules attached to both kinetochore sets irrespective of monocentric/dicentric configuration. The chromosome thus appeared to use both centromeres, either equally or with one serving a chromatid adhesion function while the second was used for transport along the spindle.