HR requires a template with sufficient sequence identity to the damaged strand in order to direct repair. In mammalian cells, the sister chromatid is the primary template for HR compared to the homologous chromosome . SCEs occur naturally as events associated with normal DNA replication and upon replication fork stalling/collapse. Formation of SCEs is intimately associated with DNA replication because eukaryotic cells exposed to DNA-damaging agents in G2 show elevated SCE levels only after completing a subsequent replication cycle .
Although the molecular mechanisms controlling SCE are not fully understood, HR between sister chromatids is principally responsible for SCE in higher eukaryotic cells . This process is considered to be conservative and error-free, since no information is generally altered during reciprocal interchange by HR. It is known that not all types of DNA damage give rise to SCE. DNA DSB agents can not efficiently induce SCEs. In contrast, SCEs can be induced by various genotoxic treatments causing replication arrest. S phase-dependent agents, such as mitomycin C (MMC) and UV light are among the most effective inducers of SCE , presumably the conditions that increase the cellular burden of SSBs or subsequent DSBs creation during replication stress generally induce SCE efficiently. Thus, the simplest pathway by which SCE likely occurs is through HR-mediated restart of a broken DNA replication fork when it encounters a nick or gap in one parental strand  (Figure 2A).
Many HR proteins have been reported to promote SCE in chicken DT40 cells. HR defective mutants, including mutants of RAD51, RAD54, and the RAD51 paralogs (i.e. RAD51B, C, and D and XRCC2), consistently have reduced SCE . However, in mammalian cells, the results are more complex. Rad54 knockout mice cells show little or no reduction in spontaneous SCE, but there is a noticeable deficiency in MMC-induced SCE [28, 29]. Moreover, some RAD51 paralog mutants show modest reductions in SCE, but isogenic rad51d mutant lines in both chinese hamster ovary and mouse fibroblasts show no decrease in spontaneous SCE [29, 30]. Consistent with these studies, we observed that BRCA1 has no obvious role in spontaneous SCE (unpublished data), although BRCA1 promotes replication-stress induced SCE. Although HR is considered to be the pathway for formation of SCEs, the observation that in HR-deficient cells, the background SCE levels are comparable to the parental cells suggests that spontaneous SCEs do not originate from HR. On the contrary, HR seems to be involved in the formation of induced SCEs . In summary, the variation in phenotypes between spontaneous and induced SCE suggests that more than one molecular pathway is responsible for SCE in response to replication stress.
In contrast to HR proteins, several proteins were found to suppress SCE. The helicase protein, BLM, appears to be important in this process since loss of the BLM gives rise to an elevated frequency of SCE during DNA replication . BLM suppresses SCE via multiple processes, including through association with topoisomerase IIIα (hTOPO IIIα) [33–35] and/or RAD51 . It has been suggested BLM and hTOPO IIIα together effect the resolution of a recombination intermediate containing a double Holliday junction. Although it is believed that BLM works as an anti-recombinase, in Drosophila DmBlm was found to be required specifically to promote the SDSA, a type of HR associated with GC but not cross-over (Figure 1). This result was confirmed in the chicken DT40 B lymphocyte line by demonstrating that Ig GC frequency was drastically reduced in BLM−/− cells . Thus, BLM suppresses SCE but promotes GC.
Recent work in our lab showed that ATR suppresses SCE upon replication fork collapse, although ATR has no role in SCE when the replication forks stall . HU, which functions as an inhibitor of ribonucleotide reductase, slows down fork progression by reducing dNTP pools, leading to stalled replication forks that after prolonged treatment collapse into DNA DSBs . We found that ATR depletion leads to an increased rate of SCE in the cells treated with HU for 18 hr when DSBs are efficiently created. Conversely, ATR depletion suppressed I-SceI-induced GC . Although it is not clear how ATR suppresses SCE, there are several possibilities. First, the similar effect of ATR and BLM deficiency on SCE and GC suggest that both proteins act in the same pathway, presumably ATR suppresses SCE via regulation of BLM. BLM is phosphorylated by ATR on two residues, Thr99 and Thr122, and has a role in the recovery from S-phase (16) . Surprisingly expression of BLM containing T99A and T122A substitutions in human BLM defective cells was able to suppress the hyper-SCE phenotype, which is the same as expression of wild type BLM, indicating that substitution of Thr99 and Thr122 with alanine did not prevent BLM from suppressing spontaneous SCE . Thus, BLM phosphorylation by ATR has no direct role in spontaneous SCE. However, the possibility that BLM phosphorylation by ATR is important to SCE induced by replication stress has not been tested. Alternatively, the SCE repression by ATR may operate in part by impeding the resection of cutting free DNA ends. It has been reported that the MEC1 replication checkpoint suppresses the formation of RAD52 foci and prevents HR at chromosome breaks induced by the HO endonuclease in yeast . This repression operates at least in part by impeding resection of DNA ends, which is essential to generate the 3′ ssDNA tails that are the primary substrate of HR. Interestingly, the MEC1 pathway does not prevent recombination at stalled forks, presumably because they already contain ssDNA , which is consistent with that the concept that ATR has no role on SCE following replication fork stalling but suppresses SCE following fork collapse after DSBs are produced . Lastly, the elevated SCE frequency following ATR depletion may be related to the specific locations where the increased breakages occur. Chromosomal fragile sites are the regions of the genome which exhibit gaps or breaks on metaphase chromosomes under conditions of partial replication stress . Common fragile sites with or without associated breakages are the preferred location for SCE in aphidicolin treated cultures [43, 44]. SCEs were found to be distributed nonrandomly across fragile sites and nonfragile sites; and among the fragile sites, the high frequency SCE sites were highly correlated with the high frequency breakage sites , indicating that SCE are preferentially induced at common fragile sites with broken ends. ATR protein was found to bind to three regions of FRA3B under conditions of replication stress, and a deficiency of ATR results in a dramatic increase in fragile site breakage [45, 46]. Thus, defective ATR signaling could result in DNA breakages at the sites which are the hotspots for SCE.