Mechanisms of Radiation Mutagenesis in Cellular and Subcellular Systems

of this work it would seem to be wise to reserve judgement. One should, however, mention the possibility that mutations are induced during the slow replication of DNA very near the replication point and that as pho­ toproducts are progressively removed, DNA synthesis speeds up, and the DNA being replicated contains progressively fewer photoproducts and gener­ ates fewer mutations upon replication. One could thus conceive of a situation where mutations are induced largely before DNA synthesis resumes at its normal rate, and loss of photoreversibility would appear to correlate better with dimer excision than with DNA synthesis. Advance towards an understanding of how unexcised lesions give rise to mutations has come from two approaches. That of Bridges & Munson has been to study the mutagenic process in a strain which is unable to excise pyrimidine dimers (E. coli WP2 Hcr-). Their philosophy is based on the be­ lief that almost every experiment carried out with Hcr+ strains is in effect observing the response of the excision-repair system, which thus masks any response of the mutagenic process. There is a further advantage in using an Hcrstrain because almost all the mutations arise from photoreversible pyrimidine dimers (62, 7 1) . This means that exposure to visible light at any time after exposure to ultraviolet gives a measure of the rate at which dimers give rise to mutations. Using this approach they have been able to show that pyrimidine dimers may persist for several generation times after ultraviolet and that they may give rise to mutations with a low probability (-'1-5 per cent) per replication cycle (69, 109) . This fact in itself implies that when a dimer passes through the replication point, the information needed to specify the purines on the complementary daughter strand must come either from the parental strand opposite the dimer, or from the daughter strand which is MUTAGENESIS IN CELLULAR AND SUBCELLULAR SYSTEMS 153 made complementary to the parental strand opposite the dimer. In the latter case some recombination-like process would have to be involved. Mutations were not established (did not lose their photoreversibility) when DNA replication was halted at the end of a cycle (69), which is in agree­ ment with earlier data with Hcr+ strains suggesting that DNA replication is involved in the establishing of mutations. The work of Rupp & Howard­ Flanders (59) indicated a way in which mutations might occur. They found that when pyrimidine dimers were allowed to pass through the replication point in an Bce strain of K-12, the newly synthesized DNA had a lower than normal molecular weight. Their calculations showed there was approximately one break in the daughter DNA for every dimer on the parental DNA which passed through the replication point, which strongly suggests that a gap was left in the daughter strand opposite the dimer. These gaps disappeared dur­ ing subsequent incubation. It was suggested (71, 1 10) that mutations might occur as mistakes during the filling of these gaps. Calculations by Bridges & Munson (69) suggest strongly (but not conclusively) that once a mutation is established (opposite a dimer?) , both daughter duplexes are mutants at the next replication. This is in agreement with the idea discussed above that in replicating past a dimer, information is normally obtained from the other parental strand or from the daughter strand formed complementary to the other parental strand by some recombination-like process. The other recent step towards an understanding of such a process came from the discovery by Witkin (1 10) that E. coli strains bearing the exrallele from Be-lor BS-2 were almost completely unmutable by ultraviolet although they mutated spontaneously at a normal rate. Strains bearing the exr­ allele are up to three times more sensitive to ultraviolet than wild-type strains. Witkin (1 10) proposed a model with two mechanisms for dealing with unexcised dimers at the replication point, one error-free, and one (exr-depen­ dent) error-prone. More recently she has modified this model in terms of a recombinational mechanism for the repair of dimer-induced replication gaps (72) . The exr gene function is seen as increasing the efficiency of gap-filling while at the same time increasing the probability of error. According to Witkin (72) there is a good correlation between the recombination ability of strains and their ability to cope with un excised dimers. Moreover Exr­ strains appear to be rather poor recipients in conjugation experiments. The suggestion that there are error-prone and error-free components in postreplication repair enables an explanation to be formulated for various ob­ servations hitherto unexplained. For example, the mutation loss which occurs when plates are incubated at low temperatures after ultraviolet could simply reflect a temperature-dependent change in the balance of the error-free and error-prone processes. If the exr system were less efficient at low temperatures one would expect this also to be reflected in an enhanced sensitivity to the lethal effect of ultraviolet and this is in fact observed (93, 1 1 1) . The enhanc­ ing effect of ionizing radiation on ultraviolet mutagenesis (112 , 113) might be