The origin of chromosome rearrangements at early stages of AMPD2 gene amplification in Chinese hamster cells.

Gene amplification and chromosomal rearrangements are frequent properties of cancer cells, provoking considerable interest in the mechanism of gene amplification and its consequences - particularly its relationship to chromosomal rearrangements. We recently studied the amplification of the gene for adenylate deaminase 2 (AMPD2) in Chinese hamster cells. Using fluorescent in situ hybridization (FISH), we found that early amplification of the AMPD2 gene is based on unequal gene segregation at mitosis, rather than local over-replication. We observed large inverted repeats of the amplified sequences, consistent with an amplification mechanism involving cycles of chromatid breakage, followed by fusion after replication and, in mitosis, the formation of bridges between the fused sister chromatids that leads to further breaks - a process we refer to as chromatid breakage-fusion-bridge (BFB) cycles. Our previous work left open the question of how this mechanism of gene amplification is related, if at all, to the chromosomal rearrangements that generate the dicentric, ring and double-minute (DM) chromosomes observed in some AMPD2-amplified metaphase cells, which are not predicted intermediates of chromatid BFB cycles, although they could be generated by related chromosome BFB cycles. We have addressed this question using FISH with probes for the AMPD2 gene and other markers on the same chromosome. Our results are not consistent with the chromosome BFB cycle mechanism, in which two chromatids break simultaneously and fuse to generate, after replication, a dicentric chromosome. Rather, they suggest that dicentric chromosomes are generated by secondary events that occur during chromatid BFB cycles. Our results also suggest that DM chromosomes are generated by the 'looping-out' of a chromosomal region, generating a circular DNA molecule lacking a centromere; in this case, gene amplification would result from the unequal segregation of DM chromosomes at mitosis. We conclude that, at early stages of AMPD2 gene amplification, chromatid BFB cycles are a major source of both 'intrachromosomal' gene amplification and genomic rearrangement, which are first limited to a single chromosome but which can then potentially spread to any additional chromosome. It also seems that, occasionally, a DNA sequence including the AMPD2 gene can be excised, generating a DM chromosome and thus initiating an independent process of 'extrachromosomal' amplification.