Analysis of One-Sided Marker Segregation Patterns Resulting From Mammalian Gene Targeting

Abstract

The double-strand break repair (DSBR) model is currently accepted as the paradigm for acts of doublestrand break (DSB) repair that lead to crossing over between homologous sequences. The DSBR model predicts that asymmetric heteroduplex DNA (hDNA) will form on both sides of the DSB (two-sided events; 5:3/5:3 segregation). In contrast, in yeast and mammalian cells, a considerable fraction of recombinants are one sided: they display full conversion (6:2 segregation) or half-conversion (5:3 segregation) on one side of the DSB together with normal 4:4 segregation on the other side of the DSB. Two mechanisms have been proposed to account for these observations: (i) hDNA formation is restricted to one side of the DSB or the other, and (ii) recombination is initially two sided, but hDNA repair directed by Holliday junction cuts restores normal 4:4 segregation on that side of the DSB in which the mismatch is closest to the cut junction initiating repair. In this study, we exploited a well-characterized gene-targeting assay to test the predictions that these mechanisms make with respect to the frequency of recombinants displaying 4:4 marker segregation on one side of the DSB. Unexpectedly, the results do not support the predictions of either mechanism. We propose a derivation of mechanism (ii) in which the nicks arising from Holliday junction cleavage are not equivalent with respect to directing repair of adjacent hDNA, possibly as a result of asynchronous cleavage of the DSBR intermediate. (Orr-Weaver et al. 1981;Szostak et al. 1983), as revised by Sun et al. (1991), is currently accepted as the paradigm for acts of double-strand break (DSB) repair that lead to crossing over between homologous sequences. The DSBR model explains homologous recombination between a linearized gene-targeting vector and the chromosome (Figure 1). The events involve resection on the two sides of the DSB, invasion by one 39-end, which primes DNA synthesis leading to the capture of the second 39-end, and eventually, formation of the double Holliday junction intermediate. If the recombining sequences differ, the initial events of strand invasion and annealing generate asymmetric heteroduplex DNA (hDNA) tracts on opposite strands on the two sides of the DSB. Outward branch migration of the double Holliday junction intermediate will result in asymmetric hDNA being flanked by symmetric hDNA on one or both sides of the DSB. Alternate sense cleavage of the double Holliday junction intermediate in either the 2,4 or 1,3 mode generates the crossover products in Figure 1, F and G, respectively, with the regions undergoing recombination being preserved as 59 and 39 homologous repeats on the chromosome. Nicks at positions 2,29 and 4,49 in the crossover product in Figure 1F and those at positions 1,19 and 3,39 in the crossover product in Figure 1G correspond to the DNA cuts required to resolve Holliday junctions previously located to the left and right of the DSB, respectively. In the crossover products, the positions of gene conversion tracts toward...

The Mechanism of Gene Targeting in Human Somatic Cells

“…Importantly, the SNP retention curve for MLH1 + cells was indistinguishable from the parental (MLH1 − ) linear retention curve (compare Figure 1E and F with Figure 7C ). Moreover, the hairpins, which are refractory to the spell-checking activity of mismatch repair [12] , [41] , were retained at the same frequency as is predicted by the linear regression of other SNPs, which are substrates for spell-checking. Finally, the percentage of discontinuous gene conversion tracts (a hallmark of spell-checking) did not change significantly in the mismatch repair-proficient, compared to the mismatch repair-deficient, background (compare Table S9 with Table S1 , respectively).…”

“…In order to differentiate the viral DNA from its chromosomal counterpart, each homology arm of the virus was marked with 4 SNPs that generated unique restriction enzyme recognition sites. In addition, a 22 bp hairpin structure, which is refractory to the mismatch repair machinery [12] , [25] that was generated by the inclusion of 3 to 4 SNPs, was also introduced into each homology arm ( Figure 1A ). The homology arms of the targeted and randomly integrated clones could be amplified from the integrated loci ( Figure 1C ) using diagnostic PCRs.…”

“…However, because of the robust competing pathway of DSBR known as non-homologous end joining [8] , gene targeting events occur rarely in mammals [9] – [11] . Indeed, despite valiant efforts — in particular by the Baker laboratory [10] , [12] , [13] — the low targeting efficiency of plasmid-based dsDNA vectors has prohibited a systematic characterization of recombination intermediates in mammalian cells. To gain better insight into the mechanism of human gene targeting it is crucial to establish a more vigorous gene targeting system.…”