Synthesis Dependent
Strand Annealing
In addition to the model for DNA double strand break repair described in the Introduction, there is another class of models which we refer to here as Synthesis Dependent Strand Annealing or SDSA. In SDSA one strand invades and initiates DNA synthesis. One version is illustrated below. As synthesis proceeds the DNA also unwinds from the template. This provides a DNA strand with which the other tail can anneal.
The evidence for this type of pathway originates from work using different organisms and experimental designs from a number of laboratories. Evidence centers around the relative lack of associated crossover events and the lack of heteroduplex formation in the cleaved copy. The conventional DSBR model is usually interpreted to predict that the Holliday junctions can be resolved such that a substantive fraction (e.g. one-half) of gene conversion events should be associated with crossing over of flanking sequences, providing that the resolution enzymes themselves are unbiased. It also predicts that both repeated sequences flanking the DSB should acquire heteroduplex DNA. Failure to observe these predictions using various experimental systems led to the SDSA-class of models.

Our laboratory has made two key observations supporting this type of pathway in S. cerevisiae. The first comes from an experiment in which an HO cut site is placed within a leu2 sequence and it is repaired from a second leu2 sequence on an autonomous plasmid. This latter copy also contains an insert of repeated sequences positioned so that the invading strands will prime synthesis through the repeated sequences (see diagram). Dissociation of the invading strands will allow them to anneal with each other, followed by DNA synthesis and ligation to yield a repaired product. A prediction of the SDSA model is that the annealing between the two tails can lead to expansions and contraction of the tandem array. Greater than 40% of the gene conversions were accompanied by contractions or expansions. Furthermore they were only located in the recipient copy.
A second experiment comes from separating the donor template into two overlapping sequences and placing one on one chromosome and the other on a different chromosome. One single-strand tail invades one chromosome and the other tail invades the other chromosome. Products reflecting successful repair events can be recovered, consistent with the model where the invading strands dissociate from their respective donors and anneal with each other.
Many versions of SDSA are premised on the lack of associated crossovers. We have investigated this in S. cerevisiae using repeated sequences located on different chromosomes. Gene conversion associated with crossing over can be scored as the formation of a chromosomal rearrangement (reciprocal translocation). A significant fraction of events could be recovered that entailed repeat contraction/expansion and crossing over. This suggests that SDSA was accompanied by crossing over. One model is presented below to illustrate how Holliday junctions could form and which in turn could give rise to crossover events.