|Home | About | Journals | Submit | Contact Us | Français|
In response to replication stress, the Mec1/ATR and SUMO pathways control the stalled and damaged fork stability. We investigated the S phase response at forks encountering a broken template (termed terminal fork). We show that double strand break (DSB) formation can locally trigger dormant origin firing. Irreversible fork resolution at the break does not impede progression of the other fork in the same replicon (termed sister fork). The Mre11-Tel1/ATM response acts at terminal forks preventing accumulation of cruciform DNA intermediates that tether sister chromatids and can undergo nucleolytic processing. We conclude that sister forks can be uncoupled during replication and that, following DSB-induced fork termination, replication is rescued by dormant origin firing or adjacent replicons. We have uncovered a Tel1/ATM and Mre11-dependent response controlling terminal fork integrity. Our findings have implications for those genome instability syndromes that accumulate DNA breaks during S phase and for forks encountering eroding telomeres.
Cells experience replication stress in response to intra-S DNA damage or oncogene-induced stimuli (Bartkova et al., 2006; Branzei and Foiani, 2005; Di Micco et al., 2006). The Mec1/ATR-mediated checkpoint recognizes ssDNA-RPA nucleofilaments at stalled or damaged forks thus preventing fork collapse (Byun et al., 2005; Lopes et al., 2001; Sogo et al., 2002; Zou and Elledge, 2003) and late replicon firing (Santocanale and Diffley, 1998; Shirahige et al., 1998). Specialized sumoylation pathways also control damaged fork stability (Branzei et al., 2006; Branzei et al., 2008). Fragile sites are specific chromosome loci that exhibit an increased frequency of gaps or breaks. They are involved in chromosome rearrangements related to cancer and are specifically prominent in ATR but not in ATM defective cells (Glover et al., 2005). However, ATM generates the signal for ATR activation in response to double strand break (DSB) formation (Cuadrado et al., 2006; Jazayeri et al., 2006).
DNA breaks arise spontaneously or in response to genotoxic events. Cells respond to DSB formation to prevent chromosomal abnormalities (Haber, 2006). The conserved Mre11-Rad50-Xrs2 (MRX) complex (MRN complex in mammals) is implicated in the DSB response. It binds and holds together the broken extremities, thus preventing chromosome fragmentation (Kaye et al., 2004; Lee et al., 2008; Lobachev et al., 2004) and mediates the loading of Tel1/ATM at the break (Falck et al., 2005; Nakada et al., 2003). The MRX complex is phosphorylated by Tel1 (D’Amours and Jackson, 2001; Usui et al., 2001). The DSB response resembles the one at shortening telomeres, which is also mediated by ATM (d’Adda di Fagagna et al., 2003; Herbig et al., 2004). If the DSB is not repaired, Mre11, Sae2, Exo1, and Dna2 resect the 5′ extremities of the DNA ends (Ira et al., 2004; Moreau et al., 2001; Nakada et al., 2004; Zhu et al., 2008) thus generating ssDNA that activates ATR (Jazayeri et al., 2006; Zou and Elledge, 2003). CDK1 activity is required for efficient DSB resection and Mec1 activation (Ira et al., 2004). While the events following DSB formation are being elucidated, it is still unclear how the replicon dynamics and the replication forks react to a broken template and which is the contribution of the DSB-response pathways under these conditions.
We have studied the response to a single DSB in S phase. We show that DSB formation locally triggers dormant origin firing. The fork encountering the break is resolved at the break site and does not restart downstream. The broken arm is replicated by a fork coming from an adjacent replicon. Resolution of the terminal fork at the break does not prevent the sister fork from continuing replication. We also show that Mre11 and Tel1 protect the integrity of terminal forks by preventing the accumulation of cruciform DNA junctions that resemble reversed forks.
We constructed a yeast strain CY7184 (HO strain) to study forks encountering a single DSB on the template (Fig. SI1A). We placed the HO nuclease consensus site at 2 kb on the right of ARS305, an efficient and early origin of replication (Newlon et al., 1993; Poloumienko et al., 2001). The silent donor cassettes HML and HMR needed for DSB repair (Haber, 1998) were deleted, the gene encoding HO was under the control of the galactose-inducible GAL1 promoter and the physiological HO consensus site at the MATa locus was mutated to generate a version no longer recognized by HO (MATa inc). Under these conditions, the HO cut generates a single irreparable DSB (Lee et al., 2000) close to ARS305 (Fig. SI1B). An isogenic HO-inc strain (HO inc) in which the HO consensus site flanking ARS305 was mutated to prevent HO cutting served as a control. Following 1 hr HO expression, in the HO strain more than 90% of the ARS305 locus is cleaved, while the genomic fragment carrying ARS305 remains unchanged in the HO-inc strain (Fig. SI1B).
Haploid cells suffering a single and irreparable DSB undergo a DNA damage checkpoint-induced prolonged arrest in G2 (Lee et al., 1998; Sandell and Zakian, 1993) that correlates with hyperphosphorylation of the Rad53 kinase (Pellicioli et al., 2001). We have analyzed cell cycle and checkpoint parameters in our strains. Following HO induction in G1, part of the culture was maintained in G1 for 6 hr, while the rest was released into S-phase with glucose to repress HO expression (Connolly et al., 1988). HO is rapidly degraded under these conditions (Kaplun et al., 2000). We monitored cell cycle progression (Figure 1A), Rad53 phosphorylation (Figure 1B) and DSB formation and processing (Figure 1C). Both HO and HO-inc strains enter and complete S-phase with similar kinetics (Figure 1A) (Ira et al., 2004; Pellicioli et al., 2001). Rad53 phosphorylation appears at 80 min. when cells have completed S phase, while cells kept in G1 fail to phosphorylate Rad53 (Figure 1B). Cdk1-dependent DSB resection and a threshold of about 10 kb of ssDNA are needed to generate enough checkpoint signal to detect Rad53 phosphorylation (Ira et al., 2004; Vaze et al., 2002). Progression through S phase also requires CDK1 activity but we observed DSB-induced Rad53 phosphorylation only when cells completed S phase. We therefore investigated whether DSB resection occurs in S phase and whether is able to generate enough checkpoint signal.
In HO inc G1 arrested cells, two restriction fragments of 9.1 kb and 2.6 kb can be detected by the ARS305 probe and remain unaltered throughout the cell cycle (Figure 1C). In HO G1 cells, a 0.6 kb band appears as a result of DSB formation while the 9.1 kb band, the uncut locus, is barely detectable. In S phase, first the 0.6 kb band and then the 2.6 kb band decrease in intensity. Concomitantly, additional smeared intermediates accumulate (* and § in Figure 1C). The progressive disappearance of the 0.6 and 2.6 kb bands and the appearance of the smeared products reflect DSB processing (White and Haber, 1990). The (*) intermediates that begin to accumulate between 40 and 50 min. result from erosion of the 2.6 kb band when the resection of the 0.6 kb band is completed. The (§) intermediates are generated when the 2.6 kb band is fully resected and the adjacent genomic fragment of 3.2 kb begins to be processed. In this case the cleavage at the restriction site is prevented by the formation of ssDNA, thus leading to the formation of intermediates with a size between 4 and 5 kb. Hence, DSB resection begins at 50 min., while the majority of the cells are still in S phase, and by 80 min. in most of the cells resection has proceeded for more than 3.2 kb. Thus Rad53 phosphorylation can be visualized only after 80 min. when cells accumulate about 10 kb of ssDNA. We conclude that, although the Mec1-Rad53 checkpoint plays an important role in response to replication stress induced by HU or MMS treatment (Lopes et al., 2001; Tercero and Diffley, 2001), this pathway is unable to efficiently sense the presence of one single and irreparable DSB during S phase.
We have investigated origin firing and replicon dynamics in response to DSB formation. Following a transient induction of HO in G1, wt HO and HO inc cells were released into S phase. Replication intermediates were analyzed by 2D-gels (Brewer and Fangman, 1987). To preserve the replication intermediates we have psoralen-crosslinked the chromatin in vivo prior to DNA extraction (Sogo et al., 2002). We analyzed the fate of the replication intermediates arising from ARS305 (Figure 2A). At 30 min. from G1 release, both HO and HO-inc strains accumulated bubble intermediates indicating that ARS305 had fired (Figure 2A). Hence DSB formation close to an origin of replication does not prevent it from firing. The fact that very large bubble intermediates can be visualized indicates that the ARS305 right fork (305R) reaches the proximity of the break site. In the HO inc strain bubbles and large Y intermediates (that form when one of the forks migrates beyond the restriction fragment), gradually diminished in intensity. At 100 min. the replication intermediates are barely detectable but reappear at 150 min. due to a new round of DNA synthesis (Figure 2A and data not shown). In the HO strain, bubble and large Y signals are more intense than in the HO-inc strain. At least two hypotheses, not mutually exclusive, may explain the relative increase in the intensity of bubble and Y structures in the HO strain. The DSB may transiently slow down the progression of one or both ARS305 forks. For instance, the accumulation of large Ys may result from the transient pausing of the 305R fork in proximity of the DSB. Alternatively, at each time point, a larger population of cells is replicating the ARS305 region following DSB formation. To address whether the 305R fork transiently stalls in front of the DSB, we monitored the replication intermediates within a restriction fragment in which ARS305 is placed asymmetrically (Fig. SI2). If the 305R fork pauses long enough, then large bubble intermediates should be visualized, analogously to what shown for subtelomeric regions (Makovets et al., 2004). We failed to detect large bubbles in the HO strain (Fig. SI2), thus suggesting that, if there is any 305R fork pausing, it does not persist long enough to be detected. Therefore, the 305R fork is rapidly resolved in proximity of the DSB giving rise to linear ends. The previous result leaves open the possibility that the accumulation of replication intermediates in the HO strain may reflect a DSB-induced ARS305 activation also in those cells that were not ready yet to fire it at that particular time. To address whether DSB formation can induce origin activation, we analyzed two dormant origins, ARS313 and ARS314, which are located at 6 and 3 kb respectively from the physiological HO cut site at the MATa locus in strain CY6914. In the absence of the DSB, the genomic locus carrying ARS313 and ARS314 is replicated passively as only Y intermediates are detected, starting from 45 min. (Figure 2B) (Poloumienko et al., 2001). Following DSB formation at MATa, however, Y structures appear 20 min earlier than in the strain lacking the cleavage site. (25 min. in Figure 2B) and bubble-shaped intermediates accumulate at 45 min. Hence, DSB formation accelerates replication of the fragment containing ARS313 and ARS314 and triggers the firing of one or both origins. The digestion strategy described in Figure 2B does not allow us to visualize bubbles arising from ARS314, as potential initiation events from this origin would generate Y intermediates. Moreover, the observation that intermediates resembling termination structures (Greenfeder and Newlon, 1992; Zhu et al., 1992) appear (arrow in Figure 2B), also suggests that both ARS313 and ARS314 can fire within the same cell. The analysis of smaller restriction fragments containing either ARS313 or ARS314 indeed shows that both origins fire in response to DSB formation (Fig. SI3). Hence, DSB formation triggers the firing of both dormant origins, despite the short inter-origin spacing that, in principle, should cause origin interference (Brewer and Fangman, 1993). The DSB unlikely has a global effect on dormant origin firing as we did not observe a faster completion of S phase following DSB formation, as expected if all dormant origins were fired. These results suggest that origin activity is positively influenced by DSB formation. This could also account for the increased intensity of replication intermediates observed at ARS305 in response to DSB formation.
It has been recently shown in yeast that the two diverging sister forks of a single replicon are in close proximity to each other in an unperturbed S phase (Kitamura et al., 2006). Our genetic system allows us to address whether the progression of the two sister forks is coordinated when one of them is challenged by a DSB on the template. We monitored the progression of the 305L fork in the HO strain by analyzing the replication intermediates at three restriction fragments on the left of ARS305. In both HO and HO inc strains, we observed a progressive invasion of Y intermediates at 3, 10 and 26 kb (fragments L305, ARS304 and L304 respectively in Figure 3A) from ARS305. The absence of bubble intermediates in the fragment containing the dormant origins ARS304 indicates that this origin is not fired in the presence of the break and that region is passively replicated from the 305L fork. Based on the observation that, while the 305R fork can replicate only 2 kb before encountering the DSB, the 305L fork proceeds asymmetrically for at least 25 kb, we conclude that sister forks can progress independently from each other, at least in response to DSB formation.
A broken template may be considered a termination point for replication, similarly to a telomere. We asked how replication is achieved on the region downstream of the break, towards ARS306. Three different scenarios can be envisaged. First, the 305R fork somewhat bypasses the break region and restarts downstream. Alternatively, a novel fork forms at the broken arm through an origin independent mechanism. Finally, replication of the downstream region is delayed and carried out by the fork coming from ARS306 (306L fork) which is located 33 kb from the break site.
We analyzed three adjacent restriction fragments immediately on the right of the lesion that cover a region around 12 kb downstream of the break (Figure 3B). In the HO inc strain, an intense and complete Y-arc appears on the R1 fragment (Figure 3B). The Y s have maximum intensity at 30 min. and then progressively decrease. At 150 min. their intensity begins to increase again, due to a new round of DNA synthesis. Fragments R2 and R3 exhibit a similar behavior although, at any given time, the relative intensity of the Y arc is lower compared to the ones of fragment R1. In the HO strain, the Y intermediates reach the maximum intensity on fragment R3, between 30 and 50 min. and then progressively disappear. The Y arcs on fragments R2 and R1 behave similarly, although in fragment R1 the intermediates are very faint. Y-arc intensity was comparable between HO and HO-inc strains in fragments R2 and R3 but the kinetics of accumulation vary, with the Y-arc intensity peaking at 40 min. in the HO strain while the same signal has maximum intensity at 30 min. in the HO-inc strain. This means that replication of the region downstream of the break is delayed by at least 10 min. in the HO strain. This delay does not reflect differences in the timing of S-phase entry between the two strains because, when the same filters were re-hybridized with the R315 probe, recognizing a fragment 200 kb distant on the right of ARS305, Y-signals accumulated with identical kinetics in both strains (Figure 3B). These data suggest that while in the HO inc strain 305R forks progressively invade fragments R1, R2 and R3, in the HO strain, replication is completed by the forks arising from ARS306, located at 35 kb from ARS305. To firmly establish the direction of the forks in HO and HO inc cells we carried out fork direction analysis (Fig. SI4) (Friedman and Brewer, 1995) in the R2 fragment. While the 2D gel pattern in the HO inc strain is compatible with forks arising from ARS305 and moving on the right, in HO cells, the fork direction profile is consistent with replication being carried out by 306L forks (Fig. SI4). We conclude that replication of the region downstream of the break is rescued by forks arising from the most proximal origin, thus ruling out that forks arising form ARS305 restart downstream of the DSB site or that the broken arm is engaged in an origin-independent replication mechanism. However, we did not detect an efficient replication of fragment R1 in the HO strain (Figure 3B). The very faint Y intermediates observed at early time points on fragment R1 may at least in part represent those 5–10% of the 305R forks that replicate a template that did not experience the HO cut or were repaired either by end-joining or by rare gene conversions from MATa-inc. Thus, forks arriving from ARS306 seem to replicate until 2–3 kb distant from the break without invading efficiently the region adjacent the break site. We note that the fragment R1 becomes extensively resected at 70 min. resulting in the progressive decrease in intensity of the monomer spot and the appearance of a higher MW spot on the linear arc due to the loss of a restriction site (Figure 3B arrows on Fragments R1 and R2). Hence, in principle, 5′ to 3′ DSB resection could counteract 306L fork progression in proximity of the break site.
Mre11 is a subunit of the MRX complex, implicated in DSB sensing and processing. Mre11 possesses both endonuclease and 3′–5′ exonuclease activity (Williams et al., 2007). We analyzed by 2D gels the replication profile of the ARS305 region in the presence of the break in mre11Δ mutants (Figure 4A). At 30 min. the quality of the replication intermediates of wt and mre11Δ cells is comparable. At 40 min., specifically in mre11Δ cells, an intense vertical signal (arrow in Figure 4A), originating in the portion of the gel where large Ys are visualized begins to accumulate. This spike migrates with a 2D gel profile typical of large branched and cruciform molecules and persists until 100 min., although with time its intensity decreases. Concomitantly with the decrease in intensity of the X-spike, there is an accumulation of Y-like intermediates with lower mass, which first appear close to the monomer spot (40 min. * in Figure 4A) and then (70–100 min.) distribute along a line that intersects the cruciform spike. Thus, two pathological events occur in mre11Δ cells, the accumulation of cruciform spike first and the one of low-mass Y-like structures later on. The appearance of the low mass intermediates indicates nucleolytic processing (Lopes et al., 2001). While we cannot exclude that exonucleolytic processing might participate to this event, the kinetics of accumulation of such intermediates is unlikely due to the sole action of an exonuclease as, with time, we do not observe a progressive loss of mass but rather the opposite effect: first the low mass Y-like accumulate close to the monomer spot (40 min.) and then (70 min.) close to fully duplicated structures. Moreover, we note that mre11Δ cells are defective in DSB resection (Tsubouchi and Ogawa, 1998). An alternative possibility is that endonucleolytic activities generate these low mass Y-like intermediates. Since Sae2 collaborates with Mre11 in the DSB response and possesses endonuclease activity that is synergistic with Mre11 (Lengsfeld et al., 2007; Sartori et al., 2007), we addressed whether Sae2 plays any role in this process. We found that mre11Δ and mre11Δ sae2Δ mutants exhibit the same 2D gel profile (i.e. accumulation of both cruciform spikes and low mass Y-like structures) (Figure 4C). We conclude that Sae2 is not required for the generation of the low mass Y-like intermediates in the absence of Mre11. Moreover, we found that also sae2Δ cells behave similarly to mre11Δ mutants, thus suggesting that Mre11 and Sae2 are both needed to prevent the same abnormal transitions of replication forks at a DSB site.
We then investigated the contribution of Tel1 at 305R replication forks encountering the DSB. Compared to wt cells, at 30 min. tel1Δ mutants exhibit a specific accumulation of large Y intermediates (Figure 4B), thus suggesting that 305R fork resolution at the DSB is impaired. With time, (50–60 min., arrows) a spike signal resembling the one observed in mre11Δ and sae2Δ mutants begins to accumulate and disappear at late time points. We note that, differently from mre11Δ and sae2Δ, in tel1Δ mutants, low mass Y-like structures are not detected, thus suggesting that 305R forks do not undergo endonucleolytic events. We conclude that Tel1 is required for the efficient resolution of 305R forks at the DSB and to prevent their degeneration into four way junctions. The 2D-gel profile of the mutants described above was dependent on the DSB formation as no aberrant DNA structures were detected in the respective HO-inc controls (data not shown).
We then investigated whether Mre11, Sae2 and Tel1 influence the progression of 306L forks in response to DSB formation. We analyzed the same R1 restriction fragment described in Figure 3B in mre11Δ, sae2Δ and tel1Δ cells (Figure 4D and data not shown). In wt cells, the replication intermediates remain barely detectable throughout the kinetics (also 306L fork pausing at the R1 fragment is observed at 55 min). Conversely, all mutants exhibit a Y arc already at 20 min. and, at later time points, abnormal structures begin to accumulate in mre11Δ, sae2Δ and tel1Δ, analogously to what shown for 305R forks (Figure 4D and data not shown). We conclude that Mre11, Sae2 and Tel1 counteract the progression of the 306L forks in proximity of the DSB site and prevent their degeneration. We also found that Exo1 counteracts the invasion of the R1 fragment by 306L forks without giving any apparent contribution at 305R forks (data no shown). We note that the progression of the 305L forks is not impaired in mre11Δ, sae2Δ and tel1Δ cells (data not shown).
Much of our knowledge on the replication stress response comes from studies on cells exposed to genotoxic agents, which either cause massive intra-S DNA lesions or severe S phase blocks. Under those circumstances, stalled and/or damaged forks generate a large amount of ssDNA tracts and DNA breaks that trigger a robust DNA damage response (DDR) (Branzei and Foiani, 2005). Replication stress can be also generated by oncogenes and those precancer cells that accumulate enough checkpoint signals, as a result of abnormal S phase events, activate the DDR to prevent cancer development (Bartkova et al., 2006; Di Micco et al., 2006; Halazonetis et al., 2008). We have investigated the mechanisms that act specifically during S phase under conditions in which one single genomic lesion is present and when the amount of checkpoint signal is below the threshold needed to activate Mec1/ATR. Our findings might have relevant implications for the early stage precancer cells that have accumulated only a few DNA lesions and in which DDR activation cannot be detected yet. The same may be true for those cells that experience the replication of a single truncated telomere. Moreover, considering that DSBs arise spontaneously as a result of cellular metabolism, our results may also help to elucidate fundamental mechanisms occurring in a normal S phase.
DSB processing and checkpoint activation require CDK1 activity (Ira et al., 2004; Jazayeri et al., 2006). S phase cells have an active CDK1 but, apparently, fail to induce the Mec1/ATR-dependent Rad53 phosphorylation in response to a single DSB. The following evidences indicates that in our system DSB resection occurs also in S phase cells: i) DSB resection can be visualized when cells are in early-middle S phase as judged by the FACS profile; ii) when replication forks travel for long distance before reaching the DSB site, a Mre11-Sae2-Tel1 and Exo1-dependent process counteracts fork progression; iii) DSB resection can be detected in HU arrested cells (Y.D. and M.F. unpublished results). Moreover, recent evidence suggests that DSBs are processed in S-phase (Zierhut and Diffley, 2008). However, when we measured the extension of ssDNA generated during resection, we noticed that cells reach the threshold of ssDNA needed for Rad53 activation approximately when S phase is already completed. We conclude that DSB resection occurs in S phase but that a single DSB, at least in our experimental conditions, is not sufficient to promote a robust Mec1-dependent S phase response. We cannot rule out that replication and resection are somewhat coordinated, however, given also the relative estimated rates of the resection and replication processes, with replication proceeding ~ 40 times faster that resection (2.9±2.3 kb/min vs 4 Kb/hour, (Fishman-Lobell et al., 1992; Raghuraman et al., 1994; Vaze et al., 2002), the relative position of the break site to the most proximal origin, could affect the extent of ssDNA nucleofilaments generated during resection: a broken template that has an efficient origin close by will be rapidly replicated thus generating two DNA ends that will then undergo resection. Conversely, a broken arm located far from an origin would be engaged into resection before being replicated. Considering that the average interorigin spacing in yeast is ~ 46 kb (Lengronne and Schwob, 2002) likely, DSB ends would never become extensively resected before replication fork passage and therefore, one DSB would not generate enough signal to activate the Mec1-Rad53 checkpoint prior to replication completion.
We showed that DSB formation in proximity of an early origin of replication does not prevent its activation, thus confirming previous results on plasmid origins (Raghuraman et al., 1994). Rather we show that one DSB locally triggers dormant origin firing. These findings might have implications for those cells that experience intra-S accumulation of DSBs such as cells treated with ionizing radiation or radiomimetic drugs or cells that express fragile sites. Interestingly, DNA combing analysis has shown that the inter-origin spacing is reduced in a variety of pathological situations leading to genome instability (Ge et al., 2007; Rao et al., 2007). Genetic inactivation of the ATR, Mec1 and Rad53 dependent pathways causes replication fork collapse, fragile site expression and DNA breaks associated with derepression of late/dormant origins of replication (Casper et al., 2002; Cha and Kleckner, 2002; Lopes et al., 2001; Santocanale and Diffley, 1998; Shirahige et al., 1998; Sogo et al., 2002). It is possible that at least a sub-population of these origins might fire in checkpoint defective cells as a consequence of DSB formation in their proximity. The analogy between our experimental conditions and a rad53 checkpoint defective context is emphasized by the finding that also in these mutant multiple firing events were observed within a small genomic region (Sogo et al., 2002). Since Mec1 and Rad53 have been implicated in directly preventing late origin firing in response to replication stress, it is possible that a two steps mechanism takes place in response to intra-S damage: when the number of DNA lesions is such that the amount of ssDNA is below the threshold for Mec1 activation, cells, locally enhance origin activity or even trigger dormant origin firing; later on, when ssDNA is above 10kb, Mec1 is activated thus causing global repression of late origins. This scenario would underline the importance of having a threshold for checkpoint activation and origins that fire throughout S phase.
It has been suggested that the firing of dormant origins may be used to rescue replication of regions in which forks have collapsed (Ge et al., 2007; Ibarra et al., 2008). Along this view, the local activation of origins following break formation might have the selective advantage to rescue replication of the broken chromosomal arm. Considering that ARS313 and ARS314 are located very close to the MATa locus where HO physiologically cuts, it is tempting to speculate that they might have been evolutionary selected as a reservoir of silent replicons when breaks form. We also note an intriguing similarity between the break-induced origin activation described in this study and what has been shown at short telomeres, which unlike normal-length telomeres, replicate early in S phase due to the early firing of subtelomeric origins (Bianchi and Shore, 2007).
Although the mechanism by which DSB formation enhances/induces local origin activation is still elusive, it is possible that supercoiling release or chromatin modification in the region surrounding the DSB site could mediate this effect. It is interesting to note that the level of histone acetylation has been reported to influence origin activity (Vogelauer et al., 2002) and that histone acetylation is influenced by DSB formation (Peterson and Cote, 2004). Another chromatin modification that maybe important is the phosphorylation of histone H2A, which extends about 25–30 kb on either side of a DSB (Kim et al., 2007). We are in the process of testing these ideas.
Evidence comes from the study of replication factories in yeast showing that two chromosomal loci with the same replication timing but located at the opposite sites of the same replicon, move close together upon DNA replication and separate from each other afterwards (Kitamura et al., 2006). However, coupling of sister forks does not appear to be a strict condition for DNA replication. Indeed, asymmetric progression of sister forks has been occasionally observed (Brewer et al., 1992; Marheineke et al., 2005; Rao et al., 2007). From a topological point of view, coupled or uncoupled sister forks might have different outcomes: if the sister replisomes do not rotate and remain immobile, then positive supercoil will accumulate ahead of the forks. If the two sister forks are uncoupled, their independent rotation along the DNA duplex may form precatenane junctions between sister chromatids. We note that accumulation of positive supercoils can promote fork reversal in the absence of the replisome (Postow et al., 2001) while precatenanes and/or their derivatives may facilitate the formation of sister cromatid junctions or even give rise to fork reversal (Bermejo et al., 2008; Cotta-Ramusino et al., 2005). Recent evidence indicates that in E. coli, sister forks can proceed independently and that precatenanes between sister chromatids likely form (Reyes-Lamothe et al., 2008; Wang et al., 2008). We show that when a DSB blocks one sister fork, the other can progress independently. Therefore, at least in response to DSB formation the two sister forks are uncoupled and this is true even in mre11Δ mutants where the terminal fork collapses. Although during an unperturbed S phase sister forks stay associated to each other it is possible that, specifically when one fork terminates at a DSB, uncoupling occurs. This might imply a damage-induced regulatory pathway that influences sister fork coordination. However we found that the DSB induced sister fork uncoupling does not depend on Mre11, Tel1 or Rad53 (data not shown). Since sister fork uncoupling is a prerequisite for precatenane formation, our data support other studies that indicate the formation of precatenanes during eukaryotic DNA replication (Lucas et al., 2001).
We show that there is no reactivation of a replication fork downstream of the DSB and that under these conditions replication completion is ensured by a fork arriving from an adjacent origin. The most obvious explanation is that, differently from single stranded lesions where leading strand re-priming seems to occur (Heller and Marians, 2006; Lopes et al., 2006) the DSB acts as a fork terminator causing irreversible replisome dissociation.
It has been shown that in conditions that induce stalled forks, Mec1 and Rad53 protect them from fork reversal and processing (Lopes et al., 2001; Sogo et al., 2002; Tercero and Diffley, 2001). Likewise, sumoylation pathways have been shown to prevent X-structures accumulation at damaged replication forks (Branzei et al., 2006; Branzei et al., 2008). In this study we have uncovered a pathway that controls terminal fork integrity. Terminal forks arise when DNA synthesis reaches the end of a broken chromosome or a shortened telomere. Here we show that the response to terminal forks is mediated by Mre11 and Tel1. We note that the Rad53-dependent replication checkpoint and the NHEJ pathway do not contribute to the stabilization of terminal forks (data not shown). It would be of interest to address whether our findings are applicable also to shortened telomeres. There is a peculiarity that characterizes specifically terminal forks: forks reaching a DSB, and perhaps a shortened telomere, experience a loose topological context. At this stage, it is unclear whether the Mre11-Tel1 checkpoint senses terminal forks by monitoring simply the presence of DNA ends and/or the abnormal topological context arising as a consequence of break formation. In any case the loose topological context caused by DSB formation or by telomere shortening, might influence certain transitions at terminal forks.
We could not detect 305R fork pausing in front of the DSB suggesting that the fork structure at the break is rapidly resolved giving rise to linear ends. Considering the resolution of our 2D-gel analysis, we cannot distinguish between the possibilities that the fork replicates until the very end of the fragment or it is resolved by an unknown mechanism a few hundred nucleotides before the break (Figure 5). If however, DSB resection starts prior to fork arrival, as in the case of 306L forks, then fork progression is somewhat impeded. This sort of barrier is relieved in the absence of Mre11, Sae2, Tel1 and Exo1. This could be due to the action of the resection itself, leaving a sufficiently long ssDNA region that the fork simply stops. It is also possible that the resection apparatus or the conformation of the ssDNA nucleofilaments generated by DSB processing may physically impede fork progression. Based on our results and on what we know on the enzymatic properties of the MRX and Sae2 complex, we propose the following model. Replication forks reach the DSB region bound by the MRX complex, Sae2 and Tel1. The cooperative action of these enzymes might then generate an endonucleolytic cleavage at one arm of the fork thus resolving it into linear intermediates (Figure 5). This mechanism of terminal fork resolution would have the advantages to prevent both a replication-mediated premature dismantling of the DDR machinery at the DSB and the generation of two sister chromatid ends that, otherwise, could be engaged into pathological transitions such as sister fusions. In the absence of Mre11 or Sae2 two abnormal intermediates form at terminal forks: a spike of cruciform structures and a class of Y-like structures with low mass. We note that terminal fork degeneration in mre11 mutants is not a consequence of the irreparable nature of the DSB as the same abnormal intermediates were observed in diploids mre11Δ/Δ in which one of the homologues is intact (data not shown). We speculate that in mre11Δ a two-step degeneration process takes place at terminal forks. First fork reversal coupled with branch migration would generate cruciform intermediates and then endonucleolytic cleavage of reversed forks would give rise to the low mass Y-like structures. Cruciform intermediates accumulate at terminal forks also in tel1Δ cells. Since damage-induced Mre11 and Sae2 phosphorylation depends on Tel1, (D’Amours and Jackson, 2001; Usui et al., 2001) the tel1Δ phenotype may result from impaired MRX and Sae2 phosphorylation. Alternatively, terminal fork stability might depend on downstream targets of the Tel1/ATM-Mre11 pathway, including replisome proteins such as the MCM complex (Cortez et al., 2004), H2A and/or chromatin remodeling factors (Morrison et al., 2007; Shroff et al., 2004), or nucleases such as Slx4 (Flott et al., 2007). Although the exact nature of these cruciform structures at terminal forks is still unknown, we note that these intermediates create transient sister chromatid bridges that might contribute to facilitate sister telomere fusion in MRN or ATM deficient cells (Ciapponi et al., 2004). It has been shown that DSB resection is almost totally MRX-dependent in G2 and only partially dependent on Mre11 in cycling cells (Diede and Gottschling, 2001; Ira et al., 2004) and that a MRX independent checkpoint is active specifically in late-S phase (Grenon et al., 2006). One possible explanation for these observations is that mre11Δ cells progressing through S-phase in the presence of DSBs convert the DNA lesions into reversed forks that are then processed by alternative pathways, thus bypassing the requirement for MRX in break resection and checkpoint activation.
In conclusion we have uncovered a novel ATM-Mre11-dependent pathway that acts at terminal forks and that together with the stalled and damaged fork responses, mediated by Mec1/ATR and SUMO, controls the integrity of replicating chromosomes in response to replication stress.
All strains (SI table 1) derive from CY6914, which was obtained by backcrossing 3 times JKM179 with W303 (Lee et al., 1998; Thomas and Rothstein, 1989). Gene deletions were obtained through the one-step PCR method (Wach, 1996). The rad53K227A mutation was integrated by linearizing with EcoRI pCH8 plasmid (Pellicioli et al., 1999). A 117 bp BglII-HincII MATa fragment containing the HO cleavage site (cs) was amplified from the pMV40 plasmid (Kostriken and Heffron, 1984) together with the HPH-MX marker (Goldstein and McCusker, 1999) and integrated close to ARS305 to produce CY7184. The resulting strain has a 40 bp deletion from 41800 to 41839 on chromosome III (SGD coordinates) that is substituted by the HOcs-HPH cassette. The HO-inc mutant (CY7382) is a survivor from galactose of CY7184 with a TG insertion in the HOcs.
Yeast strains were grown in YPA-Raffinose 2% w/v media. Early log phase cells were arrested with 2 μg/ml α-Factor. Galactose 2% was added for 1 hour to induce HO expression together with 1μg/ml α-Factor to maintain the G1 arrest. Cells were then released in S-phase, by centrifugation, washing with 1 volume of YPA-glucose 2% and re-suspension in fresh YPA-glucose 2% media at 23°C.
psoralen-crosslinking was performed as described (Gasser et al., 1996). DNA extraction was performed according the “QIAGEN genomic DNA Handbook,” using genomic-tip 100/G columns. 2D-gel procedure, was described (Brewer and Fangman, 1987). 2D-gel analysis of large, 9.5 kb fragments (as in SI-2) was performed according to (Krysan and Calos, 1991). Quantifications were described (Lopes et al., 2003) as well as fork-direction analysis (Friedman and Brewer, 1995) (http://fangman-brewer.genetics.washington.edu/fork-D.html).
Table 1 List of the yeast strains used in this study.
Supplementary Information 1. The genetic system.
(A) The donor cassettes HMLα and HMRa were deleted. The HO cut site at the MATa locus was mutated to MATa-inc. A 117 bp fragment from the MATa locus containing the HOcs and flanked by the HPH marker cassette was inserted at 2 kb distant from ARS305. (B) DSB formation efficiency in the HO strain. The HO endonuclease was induced for 1 hour in log phase cells. Genomic DNA was collected, digested with EcoRV and analyzed by southern blot with the ARS305 probe (dashed line) The HO-inc strain in which the HOcs close to ARS305 is mutated to HO-inc is used as a control.
Supplementary Information 2. Asymmetric digestion of the ARS305 locus.
The experimental procedure is the one described in figure 1. Genomic DNA from HO (CY7184) and HO inc (CY7382) strains was digested with BglI, MluI and analyzed by 2D-gels with the ARS305 probe. The ARS305 origin is asymmetrically placed in the fragment analyzed and therefore mainly Y-intermediates would be detected. In case of transient stalling of the 305R fork large bubble shaped intermediates would form in the fragment analyzed.
Supplementary Information 3. Both ARS313 and ARS314 origins of replication are activated in response to DSB formation at the MATa locus.
The experimental procedure is the one described in figure 1 except that the cells were released in the presence of 200mM HU. Genomic DNA was purified from strains (HO) CY6914 and (HO-inc) CY6965, digested with PmeI, NcoI (A) or PmeI (B), analyzed by 2D gel with the respective probes (dashed lines).
Supplementary Information 4. Replication fork direction in the region downstream of the break.
The experimental procedure is the one described in figure 1. Genomic DNA from HO (CY7184) and HO inc (CY7382) strains was digested with HindIII. DNA was separated in the first dimension, and then slides were treated with EcoRV before the second dimension and southern blotting with the R2 probe (dashed line in A). (A) Restriction digestion strategy. The fragment corresponds to R2 in figure 2B. (B) Expected pattern resulting from EcoRV digestion on forks moving in the either directions. (C) 2D gel analysis.
We wish to thank D. Branzei, F. d’Adda di Fagagna, A. Pellicioli, D. Sherratt, and all members of our laboratories for helpful comments and F. Vanoli for technical suggestions. Work in M.F. laboratory is supported by grants from Italian Association for Cancer Research, Telethon-Italy, European Community, Italian Ministry of Health and Regione Lombardia. Work in J.E.H. laboratory is supported by NIH grants GM20056 and 61766.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.