Rainer Sachsa, Michael Cornforthb, Mariel Vazqueza, Javier Arsuagaa, Lynn Hlatkyc, Charles Geardd and David Brennerd
aUCB Math; bUTMB; cHarvard DFCI; dColumbia CPS
Recently developed techniques such as multiplex-FISH (M-FISH) allow the identification of each homologous chromosome pair by its own pseudo-colour. M-FISH has recently been used to obtain quite detailed information on chromosome aberrations caused by sparsely ionizing radiation delivered during the G0/G1 phase of the cell cycle. Computer analysis of such detailed data allows some inferences to be drawn about the underlying biophysical mechanisms, even without identifying the specific enzymes or DNA sequences involved. To the extent that aberration formation is considered a surrogate endpoint for cancer induction, such mechanistic information becomes important for biologically-based extrapolations, from laboratory and epidemiological radiobiological data to the lower doses of primary interest in the context of public health.
We performed a quantitative analysis of some recently published FISH and M-FISH data sets. Several standard pathways of chromosomal aberration production which have recently been debated were analyzed, especially the breakage-and-reunion model (based on non-homologous end joining) and a recombinational "one-hit" misrejoining model (which assumes damage to one chromosome is sufficient to initiate exchanges). Model predictions were calculated using CAS (chromosome aberration simulator) Monte Carlo computer software. The calculations involved approximating DSB (DNA double strand break) induction and misrejoining as being approximately random on scales larger than ~1 Mbp. Systematic allowance was made for deviations from randomness due to "proximity effects", i.e. effects which favor misrejoining of DSBs initially formed close together in the interphase nucleus.
Computer analysis of recent M-FISH data [Loucas and Cornforth Radiat. Res. (2001)], including dose-response curves for simple and complex aberrations, strongly favors the breakage-and-reunion (BR) model over the recombinational misrejoining (RM) model for laboratory doses (~1-4 Gy). In addition, Fig. 1 compares the data of Greulich et al. [Mutat. Res. 452 (2000) 73-81], for the number of pseudo-colors involved in aberrant chromosomes of human lymphocytes that were exposed to an acute 3 Gy dose of 6 MeV photons, to predictions of each model (with parameters adjusted from other data). These data concern a mixture of cells that have undergone 0, 1 or 2 divisions post irradiation, and the modeling took into account lethal effects and/or preferential elimination of some types of aberrations at mitosis. Except for the category of 1 pseudo-color (which is underpredicted because of a known bias against small acentric rings by CAS) the BR model again (somewhat) outperforms the RM model. Older FISH data [Fomina et al. Int. J. Radiat. Biol. 76, (2000) 807-833] gave similar results (compare also [Sachs et al. Int. J. Radiat. Biol. 76 (2000) 129-148]). The implication of the BR model is that low-dose extrapolations of chromosome aberrations do really require a linear-quadratic model, rather than a linear model. That is, barring possible additional effects which are manifest primarily at low doses, there are fewer aberrations at low doses than a linear downward extrapolation would predict.
It was also found that all the standard chromosome aberration models systematically underpredict the number of very complicated aberration patterns, e.g. metaphases with many different pseudo-colors involved or particularly complex misrejoining reactions. One possible explanation of the discrepancies, originally suggested by some results in the Cornforth lab, could be an early form of radiation-induced chromosomal instability, occurring within the first few post-irradiation cell divisions and perhaps even before the first metaphase. Fig. 2 (CM columns) shows an example of how such instability could improve the fit to the data of Greulich et al. In this scenario, a minority of initial misrejoinings are "contagious", giving rise to additional, secondary chromosomal rearrangements which increase the complexity of the aberrations. As genetic instability may be a key aspect of radiation carcinogenesis, elucidating its initiation, which this data may be indicating, is important for biologically-based risk-estimation.
Research supported by DOE grant DE-FG03-00-ER62909 (CG, DB and RKS), NSF grants DMS 9971169 (MV and JA) and DBI 9904842 (LRH), and by NIH grants GM 57245 (RKS) and NIH CA86823 (LRH).