A direct test was made of predictions of the double-strand-break repair (DSBR) model of recombination in Xenopus laevis oocytes. The DNA substrate injected into oocytes had two directly repeated copies of a 1.25-kb sequence and was cleaved within one of them. Different products were expected to result from concerted, conservative events, as predicted by the DSBR model, and from nonconservative events. Only very low levels of recombination products, both conservative and nonconservative, were observed. When individual, apparent DSBR products were cloned and characterized, it emerged that the majority of them had arisen by nonconservative recombination through short, terminal homologies and not from the gene conversion events predicted for DSBR. Two cloned products among 44 tested corresponded to the predications of the DSBR model, but these could also have been generated by other processes. The most efficient recombination events in oocytes are nonconservative and are based on long, terminal homologous overlaps; when these are not available, short, imperfect overlaps support a lower level of nonconservative recombination; genuine, conservative DSBR events occur rarely, if at all.
Homology Requirements and Competition between Gene Conversion and Break-Induced Replication during Double-Strand Break Repair
