Abstract In yeast, four-stranded, biparental “joint molecules” containing a pair of Holliday junctions are demonstrated intermediates in the repair of meiotic double-strand breaks (DSBs). Genetic and physical evidence suggests that when joint molecules are resolved by the cutting of each of the two Holliday junctions, crossover products result at least most of the time. The double-strand break repair (DSBR) model is currently accepted as a paradigm for acts of DSB repair that lead to crossing over. In this study, a well-defined mammalian gene-targeting assay was used to test predictions that the DSBR model makes about the frequency and position of hDNA in recombinants generated by crossing over. The DSBR model predicts that hDNA will frequently form on opposite sides of the DSB in the two homologous sequences undergoing recombination [half conversion (HC); 5:3, 5:3 segregation]. By examining the segregation patterns of poorly repairable small palindrome genetic markers, we show that this configuration of hDNA is rare. Instead, in a large number of recombinants, full conversion (FC) events in the direction of the unbroken chromosomal sequence (6:2 segregation) were observed on one side of the DSB. A conspicuous fraction of the unidirectional FC events was associated with normal 4:4 marker segregation on the other side of the DSB. In addition, a large number of recombinants displayed evidence of hDNA formation. In several, hDNA was symmetrical on one side of the DSB, suggesting that the two homologous regions undergoing recombination swapped single strands of the same polarity. These data are considered within the context of modified versions of the DSBR model.
Activation-induced cytidine deaminase (AID) produces DNA breaks in immunoglobulin genes during antibody diversification. Double-stranded breaks (DSB) in the switch region mediate class switch recombination, and contribute to gene conversion and somatic hypermutation in the variable regions. However, the relative extent to which AID induces DSB in these regions or between these and other actively expressed sequences is unknown. Here, we exploited an enhancer-trap plasmid that identifies DSB in actively expressed loci to investigate the frequency and position of AID-induced vector integration events in mouse hybridoma cells. Compared to control cells, wild-type AID stimulates plasmid integration into the genome by as much as 29-fold. Southern and digestion-circularization PCR analysis revealed non-uniformity in the integration sites, with biases of 30- and 116-fold for the immunoglobulin kappa light chain and mu heavy chain genes, respectively. Further, within the immunoglobulin mu gene, 73% of vector integrations map to the mu switch region, an enhancement of five- and 12-fold compared to the adjacent heavy chain variable and mu gene constant regions, respectively. Thus, among potential highly transcribed genes in mouse hybridoma cells, the immunoglobulin heavy and light chain genes are important AID targets, with the immunoglobulin mu switch region being preferred compared to other genomic sites.