Portal de Periódicos da CAPES

Smc5/6 maintains stalled replication forks in a recombination-competent conformation

2009; Springer Nature; Volume: 28; Issue: 2 Linguagem: Inglês

10.1038/emboj.2008.273

1460-2075

Anja Irmisch, Eleni Ampatzidou, Ken‐ichi Mizuno, Matthew J. O’Connell, Johanne M. Murray,

Genomics and Chromatin Dynamics

Article21 January 2009free access Smc5/6 maintains stalled replication forks in a recombination-competent conformation Anja Irmisch Anja Irmisch Genome Damage and Stability Centre, University of Sussex, Brighton, UK Search for more papers by this author Eleni Ampatzidou Eleni Ampatzidou Genome Damage and Stability Centre, University of Sussex, Brighton, UK Search for more papers by this author Ken'ichi Mizuno Ken'ichi Mizuno Genome Damage and Stability Centre, University of Sussex, Brighton, UK Search for more papers by this author Matthew J O'Connell Matthew J O'Connell Department of Oncological Sciences, Mount Sinai School of Medicine, New York, NY, USA Search for more papers by this author Johanne M Murray Corresponding Author Johanne M Murray Genome Damage and Stability Centre, University of Sussex, Brighton, UK Search for more papers by this author Anja Irmisch Anja Irmisch Genome Damage and Stability Centre, University of Sussex, Brighton, UK Search for more papers by this author Eleni Ampatzidou Eleni Ampatzidou Genome Damage and Stability Centre, University of Sussex, Brighton, UK Search for more papers by this author Ken'ichi Mizuno Ken'ichi Mizuno Genome Damage and Stability Centre, University of Sussex, Brighton, UK Search for more papers by this author Matthew J O'Connell Matthew J O'Connell Department of Oncological Sciences, Mount Sinai School of Medicine, New York, NY, USA Search for more papers by this author Johanne M Murray Corresponding Author Johanne M Murray Genome Damage and Stability Centre, University of Sussex, Brighton, UK Search for more papers by this author Author Information Anja Irmisch1, Eleni Ampatzidou1, Ken'ichi Mizuno1, Matthew J O'Connell2 and Johanne M Murray 1 1Genome Damage and Stability Centre, University of Sussex, Brighton, UK 2Department of Oncological Sciences, Mount Sinai School of Medicine, New York, NY, USA *Corresponding author. Genome Damage and Stability Centre, University of Sussex, School of Life Sciences, Falmer, Brighton, E. Sussex BN1 9RQ, UK. Tel.: +44 1273 877191; Fax: +44 1273 678121; E-mail: [email protected] The EMBO Journal (2009)28:144-155https://doi.org/10.1038/emboj.2008.273 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info The Smc5/6 structural maintenance of chromosomes complex is required for efficient homologous recombination (HR). Defects in Smc5/6 result in chromosome mis-segregation and fragmentation. By characterising two Schizosaccharomyces pombe smc6 mutants, we define two separate functions for Smc5/6 in HR. The first represents the previously described defect in processing recombination-dependent DNA intermediates when replication forks collapse, which leads to increased rDNA recombination. The second novel function defines Smc5/6 as a positive regulator of recombination in the rDNA and correlates mechanistically with a requirement to load RPA and Rad52 onto chromatin genome-wide when replication forks are stably stalled by nucleotide depletion. Rad52 is required for all HR repair, but Rad52 loading in response to replication fork stalling is unexpected and does not correlate with damage-induced foci. We propose that Smc5/6 is required to maintain stalled forks in a stable recombination-competent conformation primed for replication restart. Introduction In eukaryotes, there are three conserved structural maintenance of chromosome (SMC) complexes each built around two SMC proteins. Each SMC protein has N- and C-terminal globular domains separated by two long coiled-coil regions with a central hinge at which the protein folds back on itself. The globular domains thus associate, forming a site for ATP binding and hydrolysis. SMC proteins heterodimerise through their hinges and the non-SMC subunits can bridge the two ATPase domains. Cohesin (Smc1/3) mediates sister chromatid cohesion. Condensin (Smc2/4) mediates chromosome condensation at mitosis (Losada and Hirano, 2005). The third complex, Smc5/6, is also essential and is required for efficient homologous recombination (HR) (Murray and Carr, 2008). The Smc5/6 complex contains six non-SMC subunits, Nse1–6. In Schizosaccharomyces pombe, Nse5 and Nse6 are non-essential, but mutants exhibit repair defects and slow growth (Pebernard et al, 2006). Of the four essential Nse proteins, Nse1 is predicted to be a ubiquitin ligase, Nse2 (MMS21) is a SUMO ligase, Nse3 is a member of the MAGE (Melanoma-antigen encoding gene) super-family and Nse4 (Rad62) resembles Kleisins and bridges Smc5 and 6 (Murray and Carr, 2008). In hypomorphic Smc5/6 complex mutants, HR-dependent DNA structures accumulate following replication fork collapse (Ampatzidou et al, 2006; Branzei et al, 2006). Consistent with this, these mutants show synthetic lethality/sickness with mutants in several HR regulators (Murray and Carr, 2008). Homologous recombination is involved in bypass of some DNA damage during ongoing replication and exerts an effect to restart replication when forks break (Marians, 2000). The ability to restart stalled or broken replication forks is of particular importance when two converging forks stop. If a single fork stops and cannot be recovered, the problem is resolved by adjacent fork convergence. However, in regions of unidirectional replication (the rDNA array and telomeres), a single stopped fork must be recovered because no converging fork will be available (Murray and Carr, 2008). Both the rDNA and telomeres have served as potent readouts of recombination defects. For example, recombination regulators, such as the RecQ helicase and the Smc5/6 complex, are implicated in telomerase-independent telomere maintenance (Multani and Chang, 2007; Potts and Yu, 2007) and suppression of rDNA recombination (Sinclair et al, 1997; Fricke and Brill, 2003; Torres-Rosell et al, 2007). Eukaryotic ribosomal DNA (rDNA) is organised as arrays containing 100–10 000 tandem repeats. rDNA is in the nucleolus where rRNA is synthesised and processed and the ribosomes are assembled (Nomura, 2001). Each rDNA repeat consists of 5.8S, 17S and 26S rDNA genes and a non-transcribed spacer (NTS) containing an autonomously replicating sequence (ARS) plus one or more polar replication fork barriers (RFBs). The RFBs prevent replication proceeding in the opposite direction to transcription (Figure 1A). S. pombe rDNA is heterochromatic and refractory to RNA polymerase II transcription. This silencing is dependent on the chromodomain protein Swi6HP1 (Huang, 2002). Such a large repetitive locus is a potent HR substrate, but rDNA recombination rates are significantly lower than expected, indicating that recombination within the array is suppressed (Petes, 1980). This negative regulation is important to prevent unequal recombination events that would cause gain or loss of repeats. It was recently suggested that the Smc5/6 complex exerts an effect as a negative regulator of HR in the rDNA, as budding yeast smc5/6 mutants showed increased nucleolar Rad52 foci when a double strand break was induced within the rDNA array (Torres-Rosell et al, 2007). Figure 1.Rad52 association in response to both stalled and collapsed forks. (A) Schematic of S. pombe chromosomes 2 and 3 showing the loci analysed, position of the rDNA at both ends of chromosome 3 and an rDNA repeat unit showing transcripts and the non-transcribed spacer (NTS) containing ars3001 and replication fork barriers (RFB). (B, a–c) αGFP ChIP analysis of the loci indicated from smc6+, smc6-X and smc6-74 mutant cells in a rad52-GFP background following arrest in 10 mM HU for 4 h with stable replication forks (cds1+). Increased rDNA enrichment is evident in smc6+ and smc6-X, but not in smc6-74 cells. (B, d–f) The equivalent experiment as in B1-3, with strains containing a cds1 deletion. In this background, HU arrest results in collapsed replication forks. ‘10’ and ‘20’ indicate a control region 10 and 20 kb from the indicated ARS. (C) Representative examples of Rad52–GFP foci in cds1+ and cds1-d cells blocked in HU. In the majority (95–98%) of cds1+ (smc6+ or smc6 mutant cells) blocked in HU, Rad52–GFP showed a diffuse nuclear pattern that was excluded from the nucleolus. In cds1-d cells (smc6+ or smc6 mutant cells), defective in the S phase check point, stalled forks cannot be maintained and multiple Rad52–GFP foci are observed in virtually every cell arrested for 4 h in HU. Download figure Download PowerPoint Here, we analyse two recessive smc6 loss-of-function alleles in S. pombe: smc6-X and smc6-74. Both mutants are DNA damage sensitive and HR defective but are viable at all temperatures (Lehmann et al, 1995; Verkade et al, 1999; Fousteri and Lehmann, 2000). The smc6-X mutation (R706C) maps close to the hinge. The smc6-74 mutation (A151T) maps within a highly conserved arginine finger close to the ATP-binding site of the N-terminal globular domain (Lammens et al, 2004). This suggests that smc6-74 might affect DNA-dependent ATP binding/hydrolysis (ATP-binding site mutations are lethal (Verkade et al, 1999; Fousteri and Lehmann, 2000)). smc6-74, but not smc6-X, can be suppressed by overexpression of a multi-BRCT domain protein, Brc1 (Verkade et al, 1999). This indicates that an Smc5/6 complex function is defective in smc6-74, but not in smc6-X, and that Brc1 can compensate for this smc6-74-specific function. Brc1 is homologous to Saccharomyces cerevisiae Esc4/Rtt107 and shares similarities with human PTIP, both of which have been implicated in stabilizing stalled replication forks (Jowsey et al, 2004; Rouse, 2004; Roberts et al, 2006). Brc1-mediated rescue of the smc6-74 phenotypes requires a number of structure-specific nucleases (Sheedy et al, 2005; Lee et al, 2007). We demonstrated previously that both smc6-74 and smc6-X are defective in resolving HR-dependent intermediates that accumulate at collapsed forks (Ampatzidou et al, 2006). In budding yeast, Rad51/52-dependent DNA structures have similarly been observed to accumulate in an smc5/6 complex mutant mms21(nse2) during MMS exposure (Liberi et al, 2005; Branzei et al, 2006). Thus, Smc5/6 complexes have a function processing HR intermediates generated at collapsed forks, the resolution of which is important for accurate chromosome segregation. Here, we show that, when forks collapse, HR proteins are recruited to replication sites as expected and that this is not Smc5/6 dependent. However, we unexpectedly observed that Rad52 (encoded by rad22 in S. pombe, but referred to here as Rad52) is rapidly loaded onto chromatin when replication forks are stably stalled, and that this is Smc5/6 dependent. Further, we also show that RPA loading to stable stalled replication forks is similarly Smc5/6 dependent. We thus identify an early HR function for Smc5/6 at stable replication forks and suggest that Smc5/6 is required to keep stalled forks in a conformation that allows RPA and Rad52 recruitment, and that this provides potential for replication restart. Results Smc6 is required for Rad52 chromatin association at stalled replication forks Replication forks arrested in HU are stabilised by the Rad3-Cds1 checkpoint (Lindsay et al, 1998). In the absence of cds1, the replisome dissociates from the nascent DNA ends and DNA processing occurs, a process termed ‘fork collapse’ (Cobb et al, 2003; Katou et al, 2003). Checkpoint-defective cells treated with HU undergo genome-wide fork collapse, which rapidly recruits recombination proteins and Rad52 to microscopically visible foci (Lisby et al, 2004; Meister et al, 2005). In contrast, in wild-type cells treated with HU Rad52 foci are not induced. It is thus assumed that recombination proteins do not associate with stalled (i.e., checkpoint stabilised) forks. We reported previously that the dynamics of Rad52 foci in both smc6 mutants upon exposure to HU is similar to that observed in wild-type cells (Ampatzidou et al, 2006). A focus represents the accumulation of multiple Rad52 molecules at a single random site within an individual cell. Foci are also thought to correlate with sites where repair is difficult or delayed. In contrast, chromatin immunoprecipitation (ChIP) data represents an average enrichment of protein at a defined DNA sequence in a population of cells. We thus examined Rad52 association with specific sequences (schematic Figure 1A) by ChIP in HU-arrested cultures. To distinguish between stably arrested and collapsed replication forks, we examined both cds1+ and cds1-d (checkpoint defective) backgrounds. When cds1 was absent, and forks collapse upon HU arrest, Rad52 was enriched at all loci investigated (Figure 1Bd–f). This is consistent with the multiple Rad52 foci observed in HU-arrested cds1-d and smc6 cds1-d double-mutant cells (Figure 1C and Meister et al, 2005; Ampatzidou et al, 2006). In contrast, in cds1+ cells, where stalled forks are stable and the majority of cells exhibit diffuse Rad52 nuclear staining (Figure 1C), we unexpectedly found smc6+, and smc6-X showed HU-dependent enrichment of Rad52 at the rDNA. In contrast, smc6-74 did not exhibit this enrichment (Figure 1Ba–c). We conclude that Rad52 is chromatin associated at the rDNA locus when forks are arrested in a stable configuration and that this association is Smc6 dependent. When forks collapse, Rad52 is recruited to all loci in an Smc6-independent manner. Smc6 is required for rDNA stability The Smc5/6 complex has been shown to be required for rDNA segregation and stability and this most likely reflects its function resolving HR structures formed when replication forks collapse spontaneously. To assess if the Rad52 loading defect observed for smc6-74 at stable stalled forks correlates with rDNA stability, we measured loss of a ura4+ marker integrated into one rDNA repeat (Thon and Verhein-Hansen, 2000) (Figure 2). In smc6+ (wild type) cultures, the ura4+ marker was lost from the rDNA once in 2.08 × 103 cells per generation, leading to ∼2% of growing cells becoming ura4− (Figure 2A). As polII transcription is repressed in the rDNA by silencing (Huang, 2002), ura4+ loss was confirmed by Southern blotting (Supplementary Figure S1). ura4+ loss was elevated ∼10-fold (1 in 1.94 × 102 cells per generation) in smc6-X cultures, confirming Smc5/6 complex as a negative regulator of rDNA recombination (Torres-Rosell et al, 2007). In fission yeast, rDNA repeats are located close to both telomeres of chromosome 3 (Figure 1A). Consistent with a requirement for HR to maintain this telomeric organisation, deletion of rad22 (RAD52) led to ∼50% marker loss (Supplementary Figure S2). Figure 2.Recombination in the rDNA is regulated by Smc5/6. (A) Fluctuation analysis of rDNA marker loss in strains containing a ura4+ marker integrated into a single rDNA repeat (Thon and Verhein-Hansen, 2000). (B) smc6-74 suppresses lethality associated with entry into stationary phase in swi6-d strains. The strains indicated were tested for the loss of viability during entry into stationary phase. Strains were grown in YES medium at 30°C from exponential phase to the stationary phase. (C) Cell density begins to plateau at ∼20 h. Viability is calculated from the number of cells forming colonies when 2 × 103 cells were plated on rich media. Representative example of three independent experiments is shown. Download figure Download PowerPoint Despite the fact that smc6-74 mutants show an equivalent HR defect to smc6-X mutants following HU treatment and DSBs induced by ionising radiation (Verkade et al, 1999; Ampatzidou et al, 2006), ura4+ loss in smc6-74 cultures was equivalent to that observed in smc6+ cultures (1 in 2.73 × 103 cells per generation). Thus, smc6-74 cells appear to be defective in a different aspect of rDNA HR repair compared with smc6-X cells, and this correlates with the reduced Rad52 association at stable stalled forks. Smc6-74 regulates recombination in the rDNA in the absence of silencing In the course of our experiments, we observed that smc6-X cells have a silencing defect (Supplementary Figure S3). To establish if this caused the increased marker loss, we repeated the experiments in a swi6-null background. swi6 encodes heterochromatin protein 1 in S. pombe, and is absolutely required for heterochromatin formation. In smc6+ cultures, the ura4+ loss rate was increased 7.3-fold when swi6 was deleted (Figure 2A). Rate of loss in smc6-X swi6-d was increased 14.5-fold over the smc6+swi6+ (wt) strain. Thus, loss of silencing contributes to marker loss but is not the only cause of rDNA instability in smc6-X mutants. In contrast, ura4+ loss was significantly reduced in the smc6-74 swi6-d strain compared with the smc6+ swi6-d strain (0.5-fold lower (P=0.0107)). Thus, the smc6-74 mutation represses the increased rate of marker loss in the rDNA in the absence of silencing. We noted that swi6-d and smc6-X swi6-d strains, but not smc6-74 swi6-d strains, lost viability upon entry into stationary phase, a phenotype reported for the hyper-recombinogenic fbh1 (F box DNA helicase)-null mutant (Morishita et al, 2005; Osman et al, 2005). We thus examined the survival of swi6-d strains transiting from logarithmic growth into stationary phase. swi6-d and smc6-X swi6-d cells lost viability in stationary phase (after 20 h), although more slowly than fbh1-d cells (Figure 2B and C). In contrast, smc6-74 swi6-d strains maintained viability under these conditions. This allele-specific rescue of the loss of viability of swi6-d cells entering stationary phase may suggest that recombination between repetitive sequences occurs less frequently in the smc6-74 mutant background. Rad52 chromatin association requires fork stalling and is not confined to the rDNA The Smc6-dependent Rad52 chromatin association at the rDNA during HU treatment could either be a specific response to forks stalled by HU or could occur in unperturbed S phase and become apparent in HU-treated cultures because cells become synchronised in S phase. To distinguish between these two possibilities, we synchronised smc6+ cells and assayed Rad52–GFP association in an unperturbed cell cycle (Figure 3A, left panel). Cell cycle progression was monitored by flow cytometry (Figure 3B) and by mitotic and septation indices (Figure 3C). No cell cycle-dependent enrichment of Rad52–GFP was observed either in the region of an early firing origin or at the rDNA locus. Thus, Rad52 chromatin association is dependent on fork stalling. Figure 3.Rad52 is chromatin associated at unique origins when replication stalls. (A) qPCR analysis of Rad52 ChIP at early and late replicating and rDNA origins following synchronisation of Rad52–GFP cells in G2 by cdc25-22 block and release. Parallel analysis of Mcm4 in Mcm4–GFP cells was used to monitor the timing of Rad52 association. Cells were monitored for cell cycle progression by FACS (B) and septation index (C). In an unperturbed cell cycle, G1 is very short and cytokinesis is coincident with S phase, giving rise to the 2-4C shoulder at 90 min (left). When cells are blocked in HU (right), cytokinesis proceeds as normal but S phase is slowed. This leads to uncoupling of septation and replication and the appearance of a 1C peak. Thus, the FACS profile of cells blocked in HU moves from a 2C to 1C peak with time. The 1C peak is maintained at 150 min, but by 210 min later, release has moved towards the 2C, consistent with replication being slowed but not completely stopped in HU. In the presence of HU (right-hand panels), Mcm4 (line) accumulated at early, late and rDNA origins as the cells progressed through G1, peaking at 90 min. It then declined at the early origin, indicative of origin firing but remained highly enriched at rDNA and late origins until 210 min, consistent with the repression of late firing origins. rDNA origins fire early, but only one in four fires and most Mcm4 is displaced by passive replication. In contrast to Mcm4, Rad52 levels (bars) declined, as cells progressed through G1 but increased at early and rDNA regions at 105 min and at the late origin at 210 min, when Mcm4 levels start to decline. No consistent increase in Rad52 association with any loci tested was observed in S phase in an unperturbed cell cycle (−HU, left-hand panel). Download figure Download PowerPoint To determine the timing of Rad52 association upon fork stalling, we next compared the profile of Rad52 chromatin association with that of Mcm4 (cdc21-GFP) in HU-treated synchronised cells released into S phase (Figure 3A, right panel). Mini-chromosome maintenance proteins, Mcm2–7, are loaded onto origins of replication in G1 (pre-replication complex) and, upon initiation, form the replicative helicase that unwinds DNA at the fork (Forsburg, 2008). Thus, the loss of Mcm4 from an origin provides a measure of initiation. We characterised Mcm4 association by ChIP. Cells were synchronised in late G2 and released into 10 mM HU. Cells progressed through mitosis, G1 and entered S phase, which in fission yeast is coincident with septation (Figure 3B). Mcm4 accumulated at the early origin, ars2004, between 30 and 90 min after release (G1 to early S phase) and is largely absent again at 120 min, indicative of early origin firing. However, as indicated by the 1C DNA peak (Figure 3B), replication stalls in HU after initiation. At a late firing origin, ars2018, Mcm4 similarly accumulated between 30 and 90 min, but remains present at 120 min, consistent with late origin firing being delayed by the Rad3-Cds1 checkpoint (Patel et al, 2006; Hayashi et al, 2007). At the rDNA ars3001, Mcm4 is enriched between 30 and 90 min and remained significantly enriched at 120 min. ars3001 is early firing (Kim and Huberman, 2001), but although every rDNA repeat contains an ARS element, only approximately one in four fire. Furthermore, active rDNA origins are clustered and thus interspersed with large domains where initiation is repressed (Pasero et al, 2002). In these regions, Mcm4 will be displaced by passive replication of repressed origins and so the association of Mcm4 in the rDNA will provide a measure of ongoing replication as opposed to origin firing. We next examined Rad52 chromatin association in parallel rad52-GFP cultures (FACS shows an equivalent cell cycle profile (Supplementary Figure S4)). Rad52 levels decreased at all loci as the cells progressed through mitosis into G1. Rad52 levels then increased at the early firing origins (ars2004 and ars3001) at 105 min after release into HU and were further increased at ars3001 in the rDNA at 120 min. No increase was observed at the late origin at this time point. This indicates that Rad52 associates with chromatin as or after DNA replication forks have passed. In the presence of HU, replication is slowed but not completely stopped. In both budding and fission yeast, late origins do eventually fire (Meister et al, 2006; Alvino et al, 2007). By FACS analysis, the 1C DNA peak was maintained until ∼150 min, but by 210 min the cells were starting to progress through S phase. Rad52 association with the late origin region was delayed until 210 min, when Mcm4 levels were starting to decline. More Rad52 appeared to be associated with the rDNA than with either the early or late origins. This may reflect the fact that the region is repetitive, or that replication was ongoing within the array for longer (Mcm4 was still enriched at 240 min). However, Rad52 association with the DNA remained high throughout the time course both for the rDNA and the early origin region, and the degree of association did not reflect distance from the origin (not shown; the data for 3 kb from origins is representative). Taken together, these data demonstrate that Rad52 chromatin association following replication fork stalling occurs both at unique loci and in the rDNA but is more prevalent or easier to detect in the rDNA. Smc6-dependent Rad52 chromatin association is reversible When replication forks are collapsed by treating checkpoint-defective cells with HU, Rad52 foci form rapidly and persist for many hours following the release from HU. Rad52 foci are not observed in HU-arrested checkpoint-proficient cells (Figure 1 and data not shown). To establish if the association we observe by ChIP is reversible and thus distinct from that observed as foci, we examined Rad52 association at the rDNA repeats and euchromatic loci by ChIP in an HU block and release time course (Figure 4). We also wished to verify that the lack of Rad52 association observed in smc6-74 did not reflect a difference in cell cycle progression in the presence of HU, and thus we examined smc6-X and smc6-74 in parallel. Cell cycle progression was monitored by flow cytometry (Figure 4E) and cells scored for mitotic index and septation (Figure 4C). Figure 4.Smc6 is required for the association of Rad52 on replication inhibition. (A) Schematic of time course of HU block and release. (B) αGFP ChIP analysis of the loci indicated from smc6+, smc6-X and smc6-74 mutant cells in a rad52-GFP background. Cells were arrested for 4 h in 10 mM HU and released into fresh media. Samples were taken at the times indicated in (A) and cell cycle progression was monitored by mitotic and septation indices and FACS (representative experiments (C) and (E)). (C) Proportion of aberrant mitotic cells. (D) Western blot analysis with α-GFP of whole cell extracts (ChIP input samples) shows that Rad52 levels do not vary significantly during HU treatment and release and are equivalent between smc6+ and smc6 mutant strains. (E) FACS analysis of smc6+, smc6-X and smc6-74 cells in 10 mM HU and on release shows that cell cycle progression is similar in smc6+ and smc6 mutant strains. S. pombe is a haploid organism that spends ∼70% of the time in G2. In asynchronous cultures (time 0), there is a strong 2C peak. By 2.5 h (generation time in rich media at 30°C), all the cells are arrested with a 1C DNA content. On release from HU at 4 h, the profile moves from 1C to 2C, showing that the cells complete S phase within 1 h of release. Two hours after release from HU, a 4C peak appears (coincident with the peak of septation), indicating that the cells have entered the next S phase. Download figure Download PowerPoint Cell cycle progression in the smc6 mutants was virtually identical to wild type both during the block and upon release. smc6-X has a slightly longer cell cycle, and mitosis is delayed about 10 min. Rad52–GFP immunoprecipitation of untreated smc6+, smc6-X and smc6-74 extracts enriched all loci 3- to 5-fold compared with untagged control extracts (Figure 4B). In smc6+ and smc6-X backgrounds, treatment with HU for either 2.5 or 4 h resulted in 10- to 20-fold enrichment of Rad52 at the rDNA RFB and ars3001 loci, and 8- to 15-fold enrichment at the 17S rDNA locus. The unique non-rDNA loci showed a similar but less dramatic trend (for smc6+ 2.5 h compared with asynchronous samples P=0.017, Student's t-test (paired)). Importantly, we reproducibly failed to observe HU-dependent enrichment at any loci in smc6-74 extracts. As no differences in total Rad52 levels in the ChIP samples were observed (Figure 4D), this confirms that Smc6 is required to promote Rad52 chromatin association in response to replication fork stalling. Following release from HU arrest, the level of Rad52 enrichment declined rapidly. At 0.5 h (G2) most enrichment was lost, and at 1.5 h (M-G1) enrichment was reproducibly less than that observed in asynchronous cultures (Figure 3B). Thus, Rad52 becomes chromatin associated during HU treatment but dissociates upon release as cells enter mitosis. It is important to note that this profile contrasts directly the kinetics of Rad52–GFP foci formation in HU-treated cells (Meister et al, 2005; Ampatzidou et al, 2006). Very few HU-arrested cells show Rad52–GFP foci and most have diffuse nuclear Rad52–GFP that is significantly reduced in the nucleolus (Figure 1 and Meister et al, 2005; Coulon et al, 2006). After release from HU, one or two foci become visible in ∼25% of cells at 0.5 h, and this is true in smc6+, smc6-X and smc6-74 strains (Ampatzidou et al, 2006). Thus, our ChIP data provide a distinct profile of Rad52 chromatin association that is dependent on Smc6. This is not representative of the repair events that occur at random collapsed forks, which are visualised by foci, and that are independent of Smc6. Smc5/6 indirectly regulates Rad52 chromatin association The association of Rad52 with stalled forks is unexpected. Recombination is temporally separated from replication, and recombination is proposed to reset collapsed and/or damaged forks but has not been assigned a function at stable stalled forks. Rad52 is required for Rad51-dependent HR and has Rad51-independent strand annealing activity that may be important to stabilise stalled forks. We therefore examined whether Rad51 association was reflecting Rad52 association (Figure 5A). In contrast to Rad52, Rad51 levels were not found to increase significantly when cells were treated with HU for 4 h. We conclude that Rad52 levels do not reflect Rad51-dependent HR at stable stalled forks. As rDNA marker loss is increased in smc6-X and suppressed in smc6-74 when silencing is lost, we propose that Rad52 association allows a potential for recombination. This would be consistent with Rad51-dependent strand invasion events being tightly regulated by a number of helicases (Srs2, RecQ, Fbh1) (Branzei and Foiani, 2007). Figure 5.Smc6 is required for the association of RPA on replication inhibition. Analysis of (A) Rad51 (encoded by rhp51) and (B) RPA (encoded by rad11) chromatin association in HU. (A) αGFP ChIP analysis of the loci indicated from smc6+, smc6-X and smc6-74 mutant cells in an rhp51-GFP background fol