Rapid and apparently error-prone excision repair of nonreplicating UV-irradiated plasmids in Xenopus laevis oocytes.

Repair of UV-irradiated plasmid DNA microinjected into frog oocytes was measured by two techniques: transformation of repair-deficient (delta uvrB delta recA delta phr) bacteria, and removal of UV endonuclease-sensitive sites (ESS). Transformation efficiencies relative to unirradiated plasmids were used to estimate the number of lethal lesions; the latter were assumed to be Poisson distributed. These estimates were in good agreement with measurements of ESS. By both criteria, plasmid DNA was efficiently repaired, mostly during the first 2 h, when as many as 2 x 10(10) lethal lesions were removed per oocyte. This rate is about 10(6) times the average for removal of ESS from repair-proficient human cells. Repair was slower but still significant after 2 h, but some lethal lesions usually remained after overnight incubation. Most repair occurred in the absence of light, in marked contrast to differentiated frog cells, previously shown to possess photoreactivating but no excision repair activity. There was no increase in the resistance to DpnI restriction of plasmids (methylated in Escherichia coli at GATC sites) incubated in oocytes; this implies no increase in hemimethylated GATC sites, and hence no semiconservative DNA replication. Plasmid substrates capable of either intramolecular or intermolecular homologous recombination were not recombined, whether UV-irradiated or not. Repair of Lac+ plasmids was accompanied by a significant UV-dependent increase in the frequency of Lac- mutants, corresponding to a repair synthesis error frequency on the order of 10(-4) per nucleotide.