|Home | About | Journals | Submit | Contact Us | Français|
When replication forks stall during DNA synthesis, cells respond by assembling multi-protein complexes to control the various pathways that stabilize the replication machinery, repair the replication fork, and facilitate the reinitiation of processive DNA synthesis. Increasing evidence suggests that cells have evolved scaffolding proteins to orchestrate and control the assembly of these repair complexes, typified in mammalian cells by several BRCT-motif containing proteins, such as Brca1, Xrcc1, and 53BP1. In Saccharomyces cerevisiae, Esc4 contains six such BRCT domains and is required for the most efficient response to a variety of agents that damage DNA. We show that Esc4 interacts with several proteins involved in the repair and processing of stalled or collapsed replication forks, including the recombination protein Rad55. However, the function of Esc4 does not appear to be restricted to a Rad55-dependent process, as we observed an increase in sensitivity to the DNA alkylating agent methane methylsulfonate (MMS) in a esc4Δ rad55Δ mutant, as well as in double mutants of esc4Δ and other recombination genes, compared to the corresponding single mutants. In addition, we show that Esc4 forms multiple nuclear foci in response to treatment with MMS. Similar behavior is also observed in the absence of damage when either of the S-phase checkpoint proteins, Tof1 or Mrc1, is deleted. Thus, we propose that Esc4 associates with ssDNA of stalled forks and acts as a scaffolding protein to recruit and/or modulate the function of other proteins required to reinitiate DNA synthesis.
To replicate a genome, DNA replication forks not only act processively to synthesize DNA, but when needed, they may pause, back up, and/or switch templates. These replication fork gymnastics are not only important for programmed gene rearrangements, but they are also required for the replication of some intrinsically difficult to replicate sequences and for the replication of damaged DNA. However, because these processes may also result in genome instability, they must be tightly regulated. The control of these processes is often referred to as the intra-S or replication stress checkpoint, which in Saccharomyces cerevisiae is controlled by the kinases Mec1 and Rad53 [1–4]. When a replication fork stalls, replication protein A (RPA) binds to the exposed single-stranded DNA (ssDNA) and recruits Ddc2, which also binds Mec1 . Mec1 phosphorylates Mrc1, which along with Tof1 is required to stabilize the replication fork so that synthesis may resume when the stress is alleviated [5–12]. In addition, phosphorylated Mrc1 plays a role in the phosphorylation of Rad53 by recruiting Rad53 to Mec1 and/or by promoting Rad53 autophosphorylation [6,7]. Once activated, Rad53 phosphorylates other proteins involved in replication and recombination [13,14], such as primase [15,16] and Rad55 .
The activation of the replication stress checkpoint not only stabilizes stalled replication forks, but also facilitates repair. In Escherichia coli, homologous recombination (HR) plays a central role in the recovery of stalled and broken replication forks , and a similar role is envisioned for recombination in eukaryotes . Central to HR is the formation of joint molecules that provide a physical linkage between interacting DNA duplexes. In S. cerevisiae, these joint molecules are formed by Rad51-ssDNA nucleoprotein filaments, which catalyze the homology search and DNA strand exchange [19–21]. Assembly of Rad51-ssDNA filaments on RPA-coated ssDNA requires the action of two mediator proteins, Rad52 and the heterodimer of Rad55 and Rad57 (Rad55–57) [22–24]. Rad52 function is facilitated by the physical interaction between Rad52 and both Rad51 and RPA in addition to its preference for binding ssDNA , while Rad55–57 function likely draws on the physical interactions between Rad55 and both Rad51  and ssDNA . Rad52 is required for all types of HR in vegetative and meiotic cells. While Rad55–57 is not required for spontaneous recombination in vegetative cells, it is critical for DNA damage-induced recombination in vegetative cells and for meiotic recombination [19–21]. In addition, Rad55 is phosphorylated in response to replication fork blocks and DNA damage in a Mec1- and Rad53-dependent manner and may thus help to coordinate the different facets of the stress response .
In general, there is increasing evidence that cells integrate and control the various responses to DNA damage by assembling specific proteins into multiprotein repair foci. Formation of these foci is likely mediated by specialized scaffolding proteins, which themselves do not necessarily possess any catalytic activity, but which facilitate repair by coordinating the functions of different enzymes, increasing their local concentration, and/or directly stimulating their activity. Binding to the scaffolding proteins may be mediated post-translationally by protein phosphorylation. For example, Brca1 C-terminal (BRCT) motifs [25,26] are thought to be phosphopeptide-binding domains that mediate protein-protein interactions involved in fork repair and other aspects of the DNA damage response [27–29]. In mammalian cells, several mediator proteins, such as XRCC1 and 53BP1, appear to employ BRCT motifs to coordinate the activities of DNA repair and recombination proteins [30–34]. In S. cerevisiae, the observation that many replication stress response and HR proteins are associated in S-phase nuclear foci [35,36] suggests that scaffolding proteins may exist to coordinate the assembly of replication fork repair foci in this organism, as well.
ESC4 (also known as RTT107) was first identified as part of the S. cerevisiae DNA damage response in a screen for deletion strains that are hypersensitive to methane methylsulfonate (MMS) . Based this sensitivity, the presence of six BRCT motifs, and the hyper-dependency on post-replication repair of DNA damage in its absence, we speculated that Esc4 might help recognize or repair stalled replication forks . Consistent with this idea, it was then shown that Esc4 is phosphorylated in a Mec1-dependent manner and is required for the resumption of DNA synthesis after damage  and for normal progression through S-phase . Here, we report genetic and biochemical data that suggest that Esc4 physically associates with stalled or collapsed replication forks, and helps mediate their repair. The evidence suggests that Esc4 acts as a scaffolding protein that facilitates the assembly of replication fork maintenance and repair foci.
Standard protocols were used to culture yeast in rich and synthetic media  and to construct strains by targeted deletion [41,42]. Gene deletions were verified by PCR, and multiple clones were analyzed for phenotype. Sequences of primers used to construct deletion cassettes and verify their proper insertion are available upon request.
Strains used to determine genetic interactions are derivatives of BY4741 and the kanMX4 null deletion strains of the Saccharomyces Genome Deletion Project  (OpenBiosystems). For phosphorylation studies, the stop codon of ESC4 was replaced with a 13myckanMX6 cassette , generating the C-terminal fusion protein Esc4-13myc. Microscopy of Esc4 was done by replacement of the ESC4 stop codon with a cassette encoding GFP(S65T) . For a list of strains used in this work, see Table 1.
Using the LexA-based yeast two hybrid system (Origene, Rockville, MD), Esc4 bait constructs were screened against a S. cerevisiae genomic fragment library in pJG4–5 (kindly provided by R. Brent, Molecular Sciences Inst.) in yeast strain EGY48 carrying the reporter plasmid pSH18–34. Esc4 residues 2 to 660, followed by a GGSGG peptide linker, were cloned into pNLexA, yielding pNLexA-Esc42–660. Transformation of yeast strain EGY48 containing the reporter plasmid and pNLexA-Esc42–660 with the pJG4–5 library yielded ~107 transformants; 3 × 107 were screened to identify galactose dependent Leu+ clones that were also blue in the presence of X-gal. In the second screen, residues 836–1070, preceded by a GGSGG peptide linker were cloned into pEG202, yielding pEG202-Esc4836–1070. Unlike pNLexA-Esc42–660, cells containing pEG202-Esc4836–1070 did not transform well with the library. Thus, strain EGY48 containing the pJG4–5 library (108 transformants) was mated with strain RFY206 containing the reporter plasmid and pEG202-Esc4836–1070 to identify interactors. Residues 661–835 were excluded from the screens, since constructs containing this region resulted in galactose-independent autoactivation of lacZ transcription. Library inserts isolated in both screens were specific for the Esc4 bait used in their isolation and did not cross-interact. The pJG4–5 library inserts were amplified by PCR and digested with MseI to identify unique, positive clones. A total of 14 and 3 unique clones were identified in the pNLexA-Esc42–660 and pEG202-Esc4836–1070 screens, respectively. Plasmids were rescued from these clones into E. coli KC8, amplified in E. coli XL1-Blue, and sequenced to identify library inserts. The specificity of the interaction between Esc4 bait constructs and isolated library plasmids was confirmed against LexA alone, as well as other non-specific LexA-bait fusion constructs.
Esc4 was visualized by introducing a genomic C-terminal myc-tag. Growth and MMS sensitivity of the ESC4-13myc tagged strains were similar to ESC4 strains, implying that Esc4 retains its cellular function in the presence of the 13myc fusion. Co-immunoprecipitation of Esc4 and Rad55 was performed using previously described methods  with an anti-Rad57 antibody. Cell lysates for phosphorylation assays were prepared in the presence of protease inhibitors (Complete Mini, Roche) and phosphatase inhibitors (cocktails I and II, Sigma). Lysates were normalized by Bradford assay. Visualization of Esc4-13myc by immunoblotting was done with anti-myc antibody (9E10, Santa Cruz Biotechnol.), horseradish peroxidase-conjugated mouse secondary antibody (Upstate), and ECL(+) (AP Biotech).
Midlog phase cultures were normalized by cell density and serially diluted. Dilutions were plated onto rich media containing MMS and colonies were counted or photographed after three days.
The diploid Esc4-GFP strain was treated with ethidium bromide to create a ρ° variant and facilitate nuclear staining ; localization of Esc4 was not affected by loss of mitochondria. Cells were grown at room temperature (21 °C) in synthetic complete media to allow for optimal GFP chromophore formation  and processed as described  with minor modifications. Cultures were grown to early log phase and MMS was added. After 90 m, cells were treated with 5% sodium thiosulfate, washed twice in fresh media, and mounted in 1.5% low-melt agarose. Images were acquired with a CoolSnapHQ cooled CCD camera (Roper Scientific) mounted on a Nikon E600 microscope with a Plan-Apochromat 100×, 1.4 N.A. objective, using filter sets for GFP (λex=470; λem=525) and 4'-6-diamino-2-phenylindole (DAPI, λex=360; λem=460), and the acquisition software IP LAB 3.9.2 (Scanalytics). Images of ~300 cells were scored for morphology and nuclear foci. The percentage of cells containing ≥3 nuclear foci was calculated for each cell morphology (unbudded, budded, double budded), and the average of three independent determinations is reported.
Methods for the determination of deletion frequency between the five, direct and inverted repeat δ sequences at the SUP4 locus in strain J599 and its derivatives were based on reported procedures [44–46]. J599 contains the SUP4-o allele linked to a selectable URA3 marker, as well as the two ochre suppressible mutations, ade2-1 and can1–100. Cells were grown to 5 × 107 cells/mL, washed with saline (9 g/L), and plated onto YPD to determine the total number of viable cells, and onto synthetic media lacking Arg containing 5 μg/mL adenine, 60 μg/mL canavanine, and 1 g/L 5-fluoroorotic acid (FOA) to detect deletion events. The red, FOA resistant, canavanine resistant colonies arising on the selective plates after 5 d were counted and attributed to recombination processes that led to a simultaneous loss of SUP4-o and URA3. The frequency of marker loss was calculated from the average number of colonies observed among the eight to twenty-four independent cultures of each strain per experiment; reported values are the average of three experiments. Representative clones from each experiment were assayed by PCR to determine whether deletion occurred via recombination between the direct or inverted repeats .
Replication forks stall during DNA synthesis for a variety of reasons, and different mechanisms have evolved to ensure their stabilization and reactivation . The absence of such mechanisms results in chromosomal instability. ESC4 was first identified in a screen for genes involved in the formation of silent chromatin (R. Sternglanz, personal communication and ref. ) and a screen for genes involved in the regulation of retrotransposon mobility . It has six BRCT domains, and weak sequence homology to the S. pombe protein Brc1, a protein required for chromosome stability  that appears to function with Slx1 to initiate recombinational repair . We originally identified ESC4 in a genome-wide screen for genes involved in the tolerance of MMS-induced damage . Our initial characterization identified a significant increase in the MMS sensitivity of an esc4Δ strain upon deletion of RAD18. However, epistasis analysis with genes involved in nucleotide excision repair (rad14Δ), recombination repair (rad52δ), and cell cycle control (rad9Δ, rad24Δ) indicated that ESC4 function is independent of these processes . A second genome wide screen for MMS-sensitive mutants also identified esc4Δ and demonstrated that the absence of ESC4 leads to a defect in S-phase progression . Further studies indicated that Esc4 is a nuclear protein that is phosphorylated in a Mec1-dependent manner and involved in the resumption of DNA replication after damage [38,54]. Finally, Esc4 phosphorylation by Mec1 was shown to depend on Slx4, but not Slx1 . These observations prompted us to further investigate the role that Esc4 might play in stabilizing and/or repairing stalled or collapsed replication forks.
We first conducted a yeast two hybrid screen to identify any potential interactions between Esc4 and proteins involved in the damage response. A genomic fragment library was employed to ensure that the screen would not be biased against genes that require induction by DNA damage. The sequence of Esc4 is distinguished by four well-defined BRCT motifs in the N-terminal portion of the Esc4 sequence, and at least one BRCT motif in the C-terminal portion that is readily identified by sequence homology, although two motifs have been reported in the literature  (Figure 1). Therefore, screens were performed with two different Esc4 bait constructs encompassing either the N-terminal (residues 2–660) or C-terminal (residues 836–1070) domain.
Library inserts that were independently isolated at least three times are shown in Figure 1. It is immediately apparent that the N-terminal domain of Esc4 is predicted to interact with a number of different proteins that are involved in DNA processing or repair, including Rad55, Mms22, Tof1, and Sgs1. A complex containing Esc4 and Mms22 was identified previously by immunoaffinity precipitation with Mms22-Flag . Mms22 forms nuclear foci both spontaneously and in response to the topoisomerase II poison, etoposide, and MMS22 is also epistatic with ESC4 in response to etoposide . DSBs resulting from etoposide treatment are predicted to be repaired by the single-strand annealing mechanism of HR, but while the frequency of this repair is lower in a mms22Δ mutant, cells containing double deletions of MMS22 and the other genes involved in HR are more sensitive than the single mutants . It has been suggested that Mms22 processes stalled replication forks and prepares them for repair by other pathways, and Esc4 may have a similar function.
Rad55 forms an obligate heterodimer with Rad57 and like Esc4 is phosphorylated in a Mec1-dependent manner in response to DNA damage [17,38]. To test whether Esc4 binds in vivo to Rad55 in its physiological relevant form, the Rad55–57 heterodimer, we used anti-Rad57 antibodies to immunoprecipitate the Rad55–Rad57 complex and detected Esc4-myc in these immunoprecipitates (Figure 2). The reverse experiment, immunoprecipitation of Esc4 and detection of Rad55–57, is not feasible because Esc4 is a relatively abundant protein, whereas the cellular concentration of Rad55–57 is extremely low . The interaction between Rad55–57 and Esc4 did not depend on MMS treatment, indicating that the damage dependent phosphorylation of Esc4 and/or Rad55 is not required for complex formation. Rad55 is also constitutively phosphorylated at its C-terminal region (VIB and WDH, unpublished result), which may contribute to its recognition by Esc4. The physical interaction between Esc4 and Rad55 suggests a role for Esc4 in HR. The Rad55–57 heterodimer orchestrates the assembly of the Rad51 filament on RPA-coated ssDNA, and it is possible that Esc4 also functions at this early step of recombination.
Given the interaction between Esc4 and Rad55, we next investigated the genetic relationship between ESC4 and genes involved in HR. HR is mediated by the RAD52 group of genes [19–21]. We and others previously reported that an esc4Δ rad52Δ mutant exhibits a small, but reproducible increase in MMS sensitivity compared to esc4Δ and rad52Δ single mutants, implying that Esc4 function is at least partially independent of Rad52 [37,38]. RAD51, RAD54, RAD55, and RAD57 form a RAD52 subgroup that mediates the single-strand invasion pathway of HR [19–21]. We observed a significant increase in MMS sensitivity with deletion of ESC4 in rad51Δ, rad54Δ, and rad55Δ strains (Figure 3). No growth defects were observed in the absence of damage. RAD59 encodes a protein that along with Rad52 catalyzes the single-stranded annealing (SSA) pathway of HR [19–21]. The esc4Δ rad59Δ double mutant was significantly more sensitive to MMS than would be predicted based on the sensitivity of the single mutants (Figure 3), and we conclude that Esc4 function is at least partially independent of both the single-strand invasion and SSA pathways of DNA damage repair.
MRE11, RAD50, and XRS2 form another subgroup of the RAD52 group [19–21]. Deletion of ESC4 in the single mutants, mre11Δ, rad50Δ, and xrs2Δ, yielded viable cells with slight growth defects and synergistic sensitivity to MMS (Figure 4A). Because esc4Δ was previously identified in a high-throughput screen for genes synthetically lethal with rad50Δ , we also sporulated heterozygous double mutants of esc4Δ and mre11Δ, rad50Δ, and xrs2Δ to specifically evaluate whether any of the resulting double mutants is inviable. In each case we observed a growth defect in the double mutant spores (Figure 4B). In contrast, normal, viable spore progeny were observed in esc4Δ/esc4Δ diploids, as well as heterozygous mutants of esc4Δ and rad52Δ or rad55Δ (data not shown). We conclude that Esc4 functions in damage resistance pathways that are at least partially independent of the Mre11-Rad50-Xrs2 complex.
In addition to the RAD52 group genes, repair of stalled replication forks also requires the actions of specialized, context-specific enzymes including nucleases, DNA helicases, and topoisomerases. For example, RAD27 encodes a protein with 5'–3' exonuclease and flap exonuclease activity; deletion of RAD27 renders cells particularly reliant on recombination to repair ssDNA gaps formed during lagging strand synthesis . We observed that deletion of ESC4 and RAD27 leads to a slight growth defect on rich media and a dramatic increase in sensitivity to MMS compared to either single deletion mutant (Figure 5A). This suggests that Esc4 facilitates recombinational repair of ssDNA.
esc4Δ was isolated previously in a screen for genes synthetically lethal with mutations in the DNA helicases, SRS2 and SGS1, as well as in the gene encoding the type I topoisomerase Top3 [58,60], which can function in concert with Sgs1. Thus, ESC4 appears to be required when Srs2, Sgs1, and Top3 are absent. Interestingly, Sgs1 is one of the proteins predicted to bind Esc4 based on the above described two-hybrid screen. While this physical interaction remains to be confirmed in independent assays, it may reflect a more intimate role of Esc4 in the Sgs1-Top3 pathway.
In addition, several nucleases involved in processing stalled replication forks or recombination intermediates have been identified by their functional overlap with Sgs1/Top3. These include the Mms4-Mus81 and Slx1–Slx4 endonucleases [61–64]. Deletion of ESC4 in haploid mutants of mms4Δ or mus81Δ had no significant effects on growth, but resulted in a significant increase in sensitivity to MMS relative to the single mutants (Figure 5B). In contrast, deletion of SLX1 or SLX4 did not increase the MMS sensitivity of the esc4Δ mutant (Figure 5C–D). While the lack of sensitivity of the slx1Δ strain precludes a definitive conclusion regarding its genetic relationship to esc4Δ, the similar MMS sensitivity of the slx4Δ esc4Δ and esc4Δ mutants suggests that the corresponding proteins function in the same pathway. Slx1–Slx4 is a DNA structure-selective DNA endonuclease that has been shown to be required to resolve replication intermediates specifically in the rDNA of budding yeast [63,64]. It has been proposed that the Slx1–Slx4 heterodimer functions to cleave stalled or converging replication forks resulting in SSA. In addition, it has recently been reported that Slx4 interacts physically with Esc4 and promotes the phosphorylation of Esc4 by Mec1, independently of Slx1 . The epistasis between esc4Δ and slx4Δ suggests that the corresponding proteins function in the same pathway, possibly involving rDNA replication. This function may explain why the rad52Δ esc4Δ double mutant is slightly more sensitive than the rad52Δ single mutant, as SSA in rDNA has been shown to be Rad52-independent [63,64].
In all, the data suggests that Esc4 functions with Slx4, and perhaps the Sgs1-Top3 complex, and plays an important role in recognizing and/or repairing ssDNA gaps at stalled or collapsed replication forks.
To investigate the phosphorylation of Esc4, we treated ESC4-13myc cells with MMS and subjected the cell lysates to immunoblotting (Figure 6). The abundance of Esc4 in the cell was sufficient for direct detection in whole cell lysates. Treatment with MMS led to a clear retardation of Esc4 over time to a slower migrating band (Figure 6A). Incubation of cell lysates with phosphatase converted the Esc4 band to a faster migrating species (compare Figure 6B lanes 2 and 3). These results are consistent with phosphorylation of Esc4 in response to MMS-induced DNA damage and agree with previously reported results based on visualizing Esc4 with an anti-phospho-SQ/TQ antibody .
We investigated whether the gel shift was dependent on a specific protein kinase in the known DNA damage checkpoint kinase cascade. Deletion of MEC1, the main effector of the damage-response kinase cascade, fully abolished the observed shift (Figure 6C), as observed in a previous study . Deletion of RAD53 led to a reproducible decrease in the intensity of the band corresponding to phosphorylated Esc4 (Figure 6D, lanes 3 and 4), although interpretation of this data is complicated by the apparent decrease also observed in the intensity of the band corresponding to the unmodified protein. Deletion of the downstream kinases DUN1 or CHK1 had no effect on the band migration, and thus they do not appear to play a role in Esc4 phosphorylation. Deletion of RAD51, RAD54, RAD55, MRC1, or TOF1 also had no effect (Figure 6D – 6F). We propose that after binding to the ssDNA of a stalled replication fork, Esc4 is phosphorylated by Mec1 and to a lesser, but significant extent by Rad53.
To examine the subnuclear localization of Esc4 we constructed a homozygous diploid ESC4-GFP(S65T) strain. The Esc4-GFP fusion did not increase the MMS sensitivity of the cells and thus the protein was presumed to retain its function. During asynchronous growth, the Esc4-GFP signal was coincident with the nuclear DAPI signal, as also reported in a high-throughput assay . Esc4-GFP was found in one of three nuclear distributions (Figure 7A). Nearly all of the G1 (unbudded) and G2/M (large-budded) cells displayed even, diffuse nuclear localization without any observable foci, and treatment with MMS had little effect on these cells. In contrast, we observed 1 or 2 foci in about half of the untreated S phase (small budded) cells, typically located at the edge of the nucleus, and diffuse localization in the other half. Treatment with MMS increased the number of foci in S phase cells, such that ~60% of them contained multiple, nuclear foci (Figure 7B). Similar behavior was also observed upon treatment with HU (Figure 7C). The change in Esc4-GFP localization from an even distribution throughout the nucleus, to distinct foci in S-phase cells, suggests a role for Esc4 in a DNA replication process. In addition, the apparent increase in the number of foci upon treatment with MMS supports a role for Esc4 in the stabilization and/or repair of stalled replication forks.
To determine whether the foci observed in undamaged S-phase cells could be the result of stalled replication in the rDNA, we examined Esc4-GFP in the presence of plasmid expressing Nop1-RFP (kindly provided by R. Rothstein). Nop1-RFP was clearly visible within the nucleus, indicative of the rDNA (Figure 7D). At least a fraction of the Esc4 foci appeared to be coincident with Nop1, suggesting that Esc4 may associate with rDNA. This is consistent with the genetic interactions between ESC4 and SLX4, discussed above. However, foci were also apparent that clearly did not co-localize with Nop1 suggesting that the Esc4 foci are not limited to the rDNA. Thus, we propose that in undamaged cells, Esc4 foci may be located in the rDNA where they are involved in the resolution of converging replication forks, as well as outside the rDNA, where they may be associated with stalled DNA replication forks. Esc4 foci in undamaged cells may also be storage reservoirs of protein, unassociated with any DNA.
Interestingly, the foci are not dependent on the presence of MEC1, RAD55 or MMS22 (data not shown), suggesting that neither phosphorylation nor the proposed interactions with Rad55 and Mms22 control foci formation. To determine whether Esc4 might bind the ssDNA associated with stalled replication forks, we examined the behavior of Esc4 in mrc1Δ and tof1Δ strains (Figure 7D and 7E). In the absence of either Mrc1 or Tof1, unconstrained replisome progression results in physical separation of the replication proteins from the replication fork and the accumulation of ssDNA at an otherwise undamaged replication fork . While the number of Esc4 foci in mrc1Δ and tof1Δ strains in the presence of MMS was identical to that observed in wild-type cells, the number of foci in these deletion strains was also greatly elevated in the absence of MMS treatment. The large increase in the number of spontaneous Esc4 foci observed in the mrc1Δ and tof1Δ strains, relative to their wild-type counterparts, suggests that Esc4 binds stalled replication forks that accrue ssDNA.
To investigate the role of Esc4 in recombination, we characterized recombination among the five δ sequences at the SUP4 locus [45,46] (Table 2). The δ sequences are derived from the long terminal repeats of the yeast Ty transposon, are 71 to 97% homologous, and are arranged in both direct and inverted orientations around the SUP4-o gene. The SUP4 locus may be difficult to replicate due to the δ repeats, as well as natural pause sites. While recombination at this locus may be repaired by error-free means, such as sister chromatid exchange [45,65], synaptic misalignment during recombination or an SSA mechanism may result in deletion of the SUP4-o gene .
We found that wild-type cells exhibit a low occurrence of deletions, on the order of just a few events for every 107 cells, with recombination between direct and inverted repeats contributing about equally. Conversely, cells lacking Rad51 or Rad55 exhibit a measurable increase in deletion events. Nearly 100% of the deletions observed in the rad51Δ and rad55Δ mutants occur by recombination between direct repeats, in agreement with the previous conclusion that deletion of RAD51 or RAD55 increases SSA . Qualitatively, these results agree with previous reported values ; discrepancies in the magnitude of the observed frequencies are attributed to differences in the method of calculating the recombination events. We found that deletion of ESC4 has little effect on deletion frequency, but does result in a bias for recombination between direct repeats. Deletion of ESC4 in a rad51Δ or rad55Δ mutant, restores recombination between inverted repeats to a modest level, with a small reduction in overall frequency. Thus, Esc4 appears to affect recombination between the long terminal repeats present at SUP4 in a complex manner that depends on the presence of other recombination proteins. The data suggest that Esc4 is not a core recombination factor but rather an accessory component, consistent with the idea that Esc4 is a scaffolding protein.
The physical, genetic, and biochemical data suggest that Esc4 associates with stalled or converging replication forks and helps to mediate their stabilization and repair. We suggest that Esc4 functions as a scaffolding protein that helps coordinate the various functions of different repair proteins. This is consistent with the physical interactions with Rad55 and Mms22, as well as the other predicted interactions with Sgs1 and Tof1. The genetic data suggests that while Esc4 is required for fork repair, its activities are not limited to a single process. In addition, the epistasis between esc4Δ and slx4Δ [63,64] and the localization of Esc4 to rDNA in undamaged cells suggests that these activities include the repair of forks that collapse within the rDNA repeats.
Elegant cytological work has suggested the presence of multi-protein complexes that localize multiple DNA lesions to a single repair center. In addition, DNA replication in prokaryotes and eukaryotes is organized in replication centers, where multiple forks are serviced by an overabundance of essential and non-essential replication factors [66,67]. In mammalian cells, several BRCT-containing proteins appear to function as scaffolds that help to organize such multi-protein complexes to provide an optimal disposition of repair and possibly replication factors . The importance of these scaffolding proteins is not manifest in obvious replication defects or a strong sensitivity to DNA damage in their absence. Rather the absence or mutation of such scaffolding proteins is strongly correlated with disease, such as cancer, suggesting that these proteins make critical contributions to genome stability through the subtle control of the various recombinational outcomes at stalled replication forks. We envision a similar role for Esc4.
We thank R. Kolodner, R. Rothstein, R. Brent, and M. Longtine for the generous gifts of materials. This work was supported by the National Institutes of Health (R01 GM068569 to FER and R01 CA92276 to WDH) and the Susan G. Komen Breast Cancer Research Foundation (PDF02-1450 to FER).