The replication checkpoint protects fork stability by releasing transcribed genes from nuclear pores

The Replication Checkpoint Protects Fork Stability by Releasing Transcribed Genes from Nuclear Pores Rodrigo Bermejo, 1,8, *Thelma Capra, 1,8 Rachel Jossen, 1,8 Arianna Colosio, 1 Camilla Frattini, 1 Walter Carotenuto, 1 Andrea Cocito, 1 Ylli Doksani, 1,2 Hannah Klein, 3 Bele ´nGo ´mez-Gonza ´lez, 4 Andre ´s Aguilera, 4 Yuki Katou, 5 Katsuhiko Shirahige, 6 and Marco Foiani 1,7, * 1 Fondazione Istituto FIRC di Oncologia Molecolare (IFOM) at IFOM-IEO Campus, Via Adamello 16, 20139 Milan, Italy 2 Rockefeller University, New York, NY 10065, USA 3 New York University School of Medicine, New York, NY 10016, USA 4 Centro Andaluz de Biologı ´a Molecular y Medicina Regenerativa CABIMER, Universidad de Sevilla, 41004 Seville, Spain 5 Tokyo Institute of Technology, 226-8503 Yokohama City, Japan 6 Research Center for Epigenetic Disease, Institute of Molecular and Cellular Biosciences, Tokyo University, Bunkyo-ku, 108-8639 Tokyo, Japan 7 DSBB-Universita `degli Studi di Milano, 20139 Milan, Italy 8 These authors contributed equally to this work *Correspondence: rodrigo.berm[email protected] (R.B.), [email protected] (M.F.) DOI 10.1016/j.cell.2011.06.033 SUMMARY Transcription hinders replication fork progression and stability, and the Mec1/ATR checkpoint protects fork integrity. Examining checkpoint-dependent mechanisms controlling fork stability, we find that fork reversal and dormant origin firing due to checkpoint defects are rescued in checkpoint mutants lacking THO, TREX-2, or inner-basket nucleoporins. Gene gating tethers transcribed genes to the nuclear periphery and is counteracted by checkpoint kinases through phosphorylation of nucleoporins such as Mlp1. Checkpoint mutants fail to detach transcribed genes from nuclear pores, thus generating topological impediments for incoming forks. Releasing this topological complexity by introducing a doublestrand break between a fork and a transcribed unit prevents fork collapse. Mlp1 mutants mimicking constitutive checkpoint-dependent phosphorylation also alleviate checkpoint defects. We propose that the checkpoint assists fork progression and stability at transcribed genes by phosphorylating key nucleoporins and counteracting gene gating, thus neutralizing the topological tension generated at nuclear pore gated genes. INTRODUCTION ATR/Mec1 and ATM/Tel1 stabilize stalled (Lopes et al., 2001) and terminal replication forks (Doksani et al., 2009), preventing their collapse (Branzei and Foiani, 2010) and accumulation of recombinogenic X-shaped reversed forks (Doksani et al., 2009; Sogo et al., 2002). Reversed forks are processed by the Exo1 nuclease (Cotta-Ramusino et al., 2005; Lopes et al., 2001) into gapped structures (Sogo et al., 2002). Checkpoint mutants experiencing replication stress fail to resume fork progression (Branzei and Foiani, 2010; Segurado and Diffley, 2008) and fire additional replicons (Santocanale and Diffley, 1998; Shirahige et al., 1998). Hydroxyurea (HU) causes dNTP depletion, inhibiting replicative polymerases. Consequently, replication forks stall and accumulate short RPA-ssDNA stretches (Sogo et al., 2002; Zou and Elledge, 2003) that trigger Mec1 activation. Mec1 then activates the Rad53 kinase, which protects stalled fork stability through mechanisms that are still obscure. DNA topology influences fork reversion. In vitro, reversed forks accumulate at positively supercoiled plasmids upon replisome dissociation (Postow et al., 2001b). The thermodynamic energy accumulated as torsional stress in front of replication forks causes fork reversal characterized by parental strands reannealing and nascent strands extrusion and pairing (Postow et al., 2001a; Schvartzman and Stasiak, 2004). Our current knowledge of the in vivo architectural organization of replicating chromosomes is limited. It has been predicted that those sites anchoring chromosomes to membranes behave as topological barriers, preventing the rotation of the DNA helix strands around each other and thus impeding the diffusion of topological changes (Postow et al., 2004; Wang, 2002). In principle, topological barriers might arise at sites where transcribed genes associate to fixed nuclear envelope structures. In eukaryotes, messenger RNA (mRNA) synthesis is coupled with mRNA processing, mRNP (messenger ribonucleoparticle) assembly, and export to the cytoplasm (Aguilera, 2005; Ko ¨hler and Hurt, 2007). These processes are coordinated by protein complexes tethering transcribed chromatin to the nuclear pore complex (NPC). THO is Cell 146, 233–246, July 22, 2011 ª2011 Elsevier Inc. 233 Open access under CC BY-NC-ND license. a complex, composed of Tho2, Hpr1, Mft1 and Thp2, that travels with the RNA polymerase II (Luna et al., 2008). THO also associates with Yra1 and Sub2 (to form the TREX complex) and factors involved in mRNA export, including the TREX-2 (or THSC) complex. TREX-2 is constituted by Sac3, Thp1, Sus1, and Cdc31 proteins (Ko ¨hler and Hurt, 2007). Sac3 interacts with Cdc31 and Sus1 (Jani et al., 2009) and mediates the association of the complex to the nuclear envelope by binding to the Nup1 nucleoporin at the NPC inner basket. Mutants in THO and TREX-2 genes share transcriptional elongation defects, transcriptionassociated hyperrecombination (TAR), and mRNA export defects (Luna et al., 2008). THO and TREX-2 are also required for the association of transcribed genes to the NPC (Cabal et al., 2006; Drubin et al., 2006; Rougemaille et al., 2008). This process, known as gene gating, assists rapid gene expression coupling mRNA transcription and export to the cytoplasm and has been implicated in a myosin-like Mlp1-dependent transcriptional memory mechanism (Tan-Wong et al., 2009). The THO-TREX-2-mediated physical association of the transcribed DNA with the NPC might hinder DNA strand rotation establishing topological barriers. Transcribed genes represent the most abundant sites of replication fork pausing in the yeast genome, and replisome pausing at transcribed genes is independent of the polarity between replication and transcription (Azvolinsky et al., 2009). The mechanisms causing replication/ transcription interference may not simply reflect the clash between forks and transcriptional machineries, as the Rrm3 helicase, which facilitates fork progression by removing protein obstacles, does not act at highly transcribed genes (Azvolinsky et al., 2009). Moreover, DNA topoisomerase II (Top2) seems to modulate DNA topology at sites of replication/transcription interference (Bermejo et al., 2009). We found that mutations in THO and TREX-2 genes counteract HU sensitivity and fork reversal in checkpoint mutants, thus sustaining replication fork progression. Rad53 releases transcribed genes from the nuclear envelope in response to replication stress. Several nucleoporins, including Mlp1, are phosphorylated by checkpoint kinases (Smolka et al., 2005, 2007). mlp1 mutants mimicking constitutive phosphorylation by Rad53 rescue rad53 HU sensitivity. Disrupting the tethering of transcribed genes to the NPC or introducing a double-strand break (DSB) between a fork and a highly transcribed gene alleviates the HU sensitivity and counteracts fork reversal in checkpoint-defective cells. We propose that the Mec1/ATR and Rad53-dependent replication checkpoint promotes replication fork stability by controlling gene gating, thus counteracting topological stress-driven fork reversal. RESULTS Identification of rad53 Suppressors rad53-K227A mutants (Zheng et al., 1993), in 0.2 M HU, accumulate reversed forks and hemireplicated intermediates that can be visualized by electron microscopy and two-dimensional (2D) gels (Lopes et al., 2001; Sogo et al., 2002). Reversed forks migrate on 2D gels as a cone signal characterized by cruciform intermediates spanning from fully duplicated molecules to structures of lower mass, resulting from Exo1-dependent resection (Cotta-Ramusino et al., 2005; Lopes et al., 2001). At lower HU concentrations, Exo1 does not significantly contribute to reversed fork processing and rad53-K227A cells accumulate unresected X spikes (data not shown). Two nonmutually exclusive hypotheses predict that fork reversal results from topological transitions such as accumulation of positive supercoiling (Postow et al., 2001b) and/or runoff of hemicatenane structures likely representing precatenane derivatives (Bermejo et al., 2008; Cotta-Ramusino et al., 2005). We designed genetic screens aimed at identifying factors influencing rad53-K227A viability at low HU doses. We used the yeast deletion library (Tong et al., 2001), searching for those gene deletions rescuing the HU sensitivity caused by overexpression of the dominant-negative rad53-D339A mutation (Fay et al., 1997). The suppressors, validated in the W303 genetic background (Thomas and Rothstein, 1989), were also able to rescue the rad53-K227A mutation. Deletions of SAC3,THO2 HPR1,THP2,MFT1,SUS1, and THP1 (Figure S1 available online) suppressed rad53 and mec1 HU sensitivity (Figures 1A– 1C and data not shown). THO and TREX2 mutations per se do not counteract the inhibitory effects caused by HU because the single-deletion mutants are sensitive to high HU concentrations and are synthetic sick in combination with rad53 (data not shown; Go ´mez-Gonza ´lez et al., 2009). Transcriptome analyses showed that SAC3 ablation did not significantly affect the expression of the Rad53-dependent damage-inducible genes in wild-type or rad53 cells (data not shown). TREX-2 and THO mutations do not accelerate S phase progression at low HU concentrations, as would be the case for mutants exhibiting elevated dNTP levels (data not shown). Although THO/TREX-2 mutants exhibit mild transcriptional defects (Go ´mez-Gonza ´lez et al., 2011), we cannot rule out that the rad53 suppression mechanism is, in part, due to reduced transcriptional obstructions to replication. TREX-2 and THO Mutations Rescue rad53 Fork Defects TREX-2 and THO mutations cause recombinogenic events ascribed to R loops (Gonza ´lez-Aguilera et al., 2008; Huertas and Aguilera, 2003). In vitro and in vivo observations indicate that nascent mRNAs can prime DNA synthesis (Pomerantz and O’Donnell, 2008, 2010). We investigated whether the ability of TREX-2-THO mutations to rescue rad53 defects was due to repriming of DNA synthesis downstream of collapsed forks by engaging trapped mRNAs at R loops. This would imply that the mechanism of suppression would be influenced by RNase H overexpression, as in the case of R loop accumulation and transcriptional and recombination abnormalities in THO mutants (Huertas and Aguilera, 2003). RNH1 (RNase H1) overexpression or ablation of endogenous RNase H1 (RNH1) or RNase H2 subunits (RNH201,RNH202) did not influence the HU sensitivity of rad53 or sac3 rad53 cells (Figures 2A and 2B). Hence, the suppression of TREX-2-THO mutations does not depend on the aberrant accumulation of R loops. We then analyzed the effect of SAC3 ablation on rad53 replicon dynamics (Figure 3A) by immunoprecipitation of BrdU- substituted DNA followed by high-density oligo array hybridization (BrdU-IP-Chip) (Fachinetti et al., 2010). At low doses of HU, wild-type forks emanating from early replication origins 234 Cell 146, 233–246, July 22, 2011 ª2011 Elsevier Inc. progressed slowly (Lopes et al., 2001), generating BrdU tracks spanning several kilobases outward of the origin sequences (Figure 3A). In rad53 cells, shorter BrdU tracks around early origins were observed due to fork collapse and consequent failure to incorporate BrdU (Feng et al., 2006; Lopes et al., 2001; Sogo et al., 2002). Concomitantly, additional BrdU tracks appeared at late/dormant origins (Santocanale and Diffley, 1998; Shirahige et al., 1998). Though SAC3 ablation per se did not cause A B GALrad53-D339A sac3 GALrad53-D339A tho2 YP Galactose 10 mM HU YP Glucose 10 mM HU C Figure 1. TREX-2 and THO Ablations Rescue rad53 HU Sensitivity (A) GAL-rad53-D339A SGA YP galactose or YP glucose plates containing 10 mM HU. White boxes mark suppressors of rad53-D339A HU sensitivity. (B) WT, sac3D,rad53-K227A,rad53-K227A sac3D,thp1D,rad53-K227A thp1D,sus1D, and rad53-K227A sus1Dcells plated without (YPDA) or with 5 mM HU. (C) Summary of THO/TREX, TREX-2, and NPC genes and their genetic interactions with rad53-K227A. Human orthologs are indicated. Null, gene deletion phenotype; rad53 rescue, suppression of rad53-K227A lethality at 5–10 mM HU; ss with rad53, synthetic sickness in combination with rad53-K227A;HU sensitivity, growth defects at 100–200 mM HU; phosphosites, the residues targeted by the checkpoint kinases (in parentheses) are shown. For schematic summary, see Figure S1. Cell 146, 233–246, July 22, 2011 ª2011 Elsevier Inc. 235 significant changes in replicon dynamics, in rad53, it rescued both the short BrdU tracks at early origins and the BrdU incorporation at late/dormant origins. Because SAC3 deletion restores replicon progression at early origins without exhibiting additional origin-unrelated BrdU peaks, we conclude that Sac3 ablation somewhat stabilizes genome-wide rad53 forks. This further argues against reinitiation events mediated by R loops. The finding that, in rad53 sac3 cells, stabilization of early replicons correlates with the lack of firing of late/dormant replicons suggests that the two phenomena are mechanistically linked, at least in this genetic background. Because Sac3 plays a global and detrimental role on the progression of HU-challenged forks Figure 2. RNase H Overexpression Does Not Influence rad53 or rad53 sac3 Cells’ HU Sensitivity (A) WT, sac3D,rad53-K227A, and rad53-K227A sac3Dcells carrying either the vector (pYES) or pGW-RNH1, expressing RNH1 under the GAL1 promoter (GAL RNH1), were plated in glucose (GLU)- or galactose (GAL)-containing media without () or with 5 mM HU. (B) WT, rnh1D,rad53-K227A,rad53-K227A rnh1D, rnh201D,rad53-K227A rnh201D,rnh202D, and rad53-K227A rnh202Dcells were plated without (YPDA) or with 2.5 or 5 mM HU. in rad53 cells, we tested whether Sac3 affected the accumulation of aberrant replication intermediates in rad53 mutants. Replication intermediates from wild-type, sac3,rad53, and rad53 sac3 strains treated with 25 mM HU were visualized by 2D gels (Figure 3B) at the early origin ARS305 (Newlon et al., 1993; Poloumienko et al., 2001). Wild-type and sac3 cells exhibited comparable 2D profiles and accumulated bubbles, which result from origin firing, and large Ys, which arise due to asymmetric fork progression outside of the restriction fragment analyzed (Brewer and Fangman, 1987). At 60–90 min, rad53 cells accumulated X-shaped intermediates corresponding to unprocessed reversed forks (Cotta-Ramusino et al., 2005; Lopes et al., 2001). Reversed forks detection was reduced in sac3 rad53 double mutants (Figure 3B). Hence, SAC3 ablation rescues replicon dynamics and fork reversal in rad53 mutants, suggesting that the TREX-2 complex, and by extension THO, may act in coordination with replication forks or even be an integral part of the moving replisome. However, the Hpr1 ChIP-chip genomic clusters did not colocalize with DNA polymerases at active replication origins in HU-treated cells but overlapped throughout the cell cycle with RNA Pol II clusters (Bermejo et al., 2009; Go ´mez-Gon- za ´lez et al., 2011). This is consistent with a global role for Hpr1 in transcription elongation and mRNP biogenesis. We note that the Hpr1-binding sites also correlate with Top2 and Hmo1 clusters that represent nonpolar pausing elements for incoming forks (Azvolinsky et al., 2009; Bermejo et al., 2009). Rad53 Influences Gene Gating It is reasonable to think that the TREX-2-THO-mediated physical continuity between transcribed chromatin and the nuclear envelope would establish nonpolar barriers preventing the diffusion 236 Cell 146, 233–246, July 22, 2011 ª2011 Elsevier Inc. A B Figure 3. SAC3 Deletion Suppresses rad53 Cells’ Fork Defects (A) BrdU-IP-Chip analysis of replicon dynamics of WT, sac3D,rad53-K227A, and rad53-K227A sac3Dcells 60 min after release from G1 into S phase in 25 mM HU. Orange (BrdU-IP) histogram bars in the y axis show the average signal ratio in log2 scale of loci along the reported region on chromosome III. The x axis shows chromosomal coordinates. Positions of early and late/dormant ARS elements are in red and gray, respectively. Blue and red horizontal bars mark the BrdU incorporation tracks corresponding to forks emanated from early replication origins in WT, sac3D, and rad53-K227A sac3Dcells or rad53-K227A mutants, respectively. Red triangles mark additional BrdU tracks generated by unscheduled dormant origin firing in rad53-K227A cells. (B) 2D gel analysis of replication intermediates in WT (SAC3 RAD53), sac3D(sac3 RAD53), rad53-K227A (SAC3 rad53), and rad53-K227A sac3D(sac3 rad53) cells at the indicated times after release from G1 into S phase in 25 mM HU. A schematic representation of the 2D gel profiles observed in RAD53 and rad53 cells is shown. Histogram plots of the ratio between quantified ‘‘Spike’’ and ‘‘Large Y’’ intermediates signals are shown. Cell 146, 233–246, July 22, 2011 ª2011 Elsevier Inc. 237 of topological stress, as the association of DNA to a fixed nuclear structure would limit the rotation of helix strands around each other (Postow et al., 2001a, 2004). The consequent accumulation of positive supercoiling at transcribed genes might therefore provide the energy source to promote reversal of incoming forks, particularly in a checkpoint-defective genetic context that exhibits an altered replisome-fork association (Cobb et al., 2003; Lucca et al., 2004; Postow et al., 2001b). This hypothesis would have the following expectations: (1) part of the fork instability problems in rad53 mutants should depend on transcription; (2) the Mec1-Rad53 checkpoint might control replicon integrity by releasing the topological barriers imposed by the coupling between mRNA synthesis and nuclear export; and (3) generating a physical discontinuity either at the level of the nuclear porebound TREX-2/THO complexes or in the DNA strands at the border between an incoming fork and a transcribed gene should relieve the topological stress causing fork reversal. We tested the contribution of transcription on fork stability in rad53 mutants. We transformed wild-type, sac3,rad53, and rad53 sac3 cells with a plasmid carrying the URA3-selectable marker and a galactose-inducible LACZ gene that is transcribed head on with the left fork arising from ARS209 (Figure 4A). Wildtype and sac3 cells were able to stably retain the plasmid and grow in the absence of uracil with or without galactose. rad53 AB CD Figure 4. The Replication Checkpoint Negatively Regulates Gene Gating (A) WT, sac3D,rad53-K227A, and rad53-K227A sac3Dcells carrying pGAL-LACZ-IN (IN) were grown in synthetic complete Ura plates containing glucose (SC –Ura) or galactose (SGal –Ura). (B) Representative images of Nup49-GFP, TETRGFP, TETO::GAL1/10/7, Nop1-Cherry glucosegrown cells and percentage of cells showing central or NPC-tethered (peripheral) GAL loci in WT or mlp1Dcells grown in the presence of glucose (GLU) or galactose (GAL). (C) WT and rad53-K227A cells were grown overnight in the presence of galactose (GAL) and treated with 0.2 M HU for the indicated times. The mean percentages of cells showing peripheral GAL loci and standard deviations (histogram error bars) from three independent experiments are shown. (D) Serial dilutions of WT, sac3DCID,rad53- K227A, and rad53-K227A sac3DCID cells plated in the absence (YPDA) or presence of 5 mM HU. cells in galactose were unable to form colonies, suggesting that they failed to retain the URA3 plasmid following galactose-induced transcription of the LACZ gene. We observed the same phenomenon using a plasmid carrying the LACZ gene transcribed codirectionally with forks arising from ARS209 (data not shown). We note that the GAL1 promoter is leaky in the presence of glucose, thus explaining the growth defects of glucose-treated rad53 cells. SAC3 ablation was able to rescue the growth defects in glucose and the cell lethality in galactose owing to the rad53 mutation. We conclude that the TREX-2 complex is detrimental for rad53 viability when forks are destabilized by transcription, raising the possibility that, also in rad53 mutants treated with low doses of HU (and therefore with an already compromised fork processivity), TREX-2 triggers fork abnormalities by tethering transcription with nuclear pore-mediated mRNA export. We then tested whether Rad53 influences the nuclear envelope association of the transcribed GAL genes locus tagged with the TET operator/TETr-GFP system (Berger et al., 2008). We analyzed the percentage of cells exhibiting peripheral distribution of the GFP loci in logarithmically growing cells in the presence of glucose (repressing conditions for GAL genes) or galactose (Figure 4B). In glucose, the GAL1/GAL10/GAL7 gene locus was localized to the nuclear periphery in 39% of the cells, whereas in galactose, in 67%. However, gene gating-defective mlp1 mutants exhibited 37% and 44% peripheral foci in glucose and galactose, respectively, according to previous results (Brickner and Walter, 2004; Cabal et al., 2006; Tan-Wong et al., 2009). We then analyzed the localization of the GAL cluster in wild-type and rad53 cells in the presence of HU and galactose 238 Cell 146, 233–246, July 22, 2011 ª2011 Elsevier Inc. (Figure 4C). Whereas in wild-type cells, the percentage of peripheral GFP loci decreased, in rad53, it remained at high levels. Conversely, Rad53 did not affect GFP distribution in HU-treated cells in the presence of glucose (data not shown). We then interrupted in rad53 cells the physical continuity within the TREX-2 complex by introducing the sac3DCID mutation that precludes the association with Sus1 and Cdc31 (Figure S1 available online) (Jani et al., 2009) and compared HU sensitivity and fork reversal in rad53 and rad53 sac3DCID cells. The sac3DCID mutation rescued rad53 HU sensitivity and fork reversal (Figure 4D) (data not shown). Several NPC factors in the inner basket and some THO subunits are phosphorylated in a checkpoint-dependent manner in yeast and mammals, and in particular, Mpl1 is a Rad53 target (Figure 1C) (Chen et al., 2010; Matsuoka et al., 2007; Smolka et al., 2007). Ablation of MLP1 in rad53 cells prevented the peripheral localization of the GAL locus following HU treatment in galactose, resembling mlp1 mutants (Figure 5A). In principle, Rad53 might counteract gene gating in S phase by negatively regulating NPC factors. The HU sensitivity of rad53 cells might be ascribed to the inability to release the association between transcribed genes and the nuclear envelope and, in turn, to counteract the topological barriers arising when forks approach nuclear envelope-associated transcribed genes. We then ablated those NPC proteins exhibiting gene gating defects (Cabal et al., 2006; Tan-Wong et al., 2009) and tested their ability to alleviate rad53 HU sensitivity. NUP1 or MLP1 deletions recapitulated the rad53 suppression by TREX2 mutations (Figure 5B). Other proteins such as Nup2 and Nup60, which localize in the inner basket of the nuclear pore but do not physically interact with TREX-2 and whose role in gene gating is still controversial (Brickner et al., 2007; Cabal et al., 2006; Light et al., 2010), did not contribute to rad53 HU sensitivity (Figure S2). Altogether, these results suggest that disrupting the association between the nuclear pore and TREX-2 proteins might alleviate the inability of rad53 mutants to detach transcribed genes from the nuclear envelope. We then mutated serine 1710 in Mlp1 (Figure 5C) to an aspartic residue or an alanine to mimic constitutive Mlp1 phosphorylation or dephosphorylation, respectively. mlp1-S1710D counteracted nuclear peripheral localization of transcribed GAL genes, thus resembling loss-of-function mlp1 mutations (Figure 5D). Conversely, mlp1-S1710A cells behaved like wild-type. Moreover, mlp1-S1710D, differently from mlp1-S1710A, also rescued rad53 HU sensitivity (Figure 5E). These observations support the hypothesis that Rad53 inhibits gene gating though phosphorylation of Mlp1 and, possibly, other NPC proteins. Fork Reversal Is Counteracted by Double-Strand Break Formation We then explored the effect of releasing the topological tension within a region between a fork and a transcribed gene on fork reversal. We induced a DSB by overexpressing the HO endonuclease (Lee et al., 2000) that recognizes a specific site inserted between ARS305 and the most proximal transcribed gene, PDI1, which causes fork pausing (Bermejo et al., 2009) and is bound by Top2, Hmo1, Hpr1, and Sac3 (Bermejo et al., 2009 and data not shown). In theory, the DSB-induced discontinuity in the helix strands should permit rotation of their free ends, thus leading to supercoiling relaxation and counteracting fork reversal. We induced DSB formation and released the cells in the presence of HU (Figure 6). We analyzed the replication intermediates of ARS305 and ARS202, an origin located on a different chromosome and not experiencing DSB formation. Strains bearing a noncleavable HO sequence were included as controls (HO-inc). Wild-type cells fired both origins, as indicated by the presence of large Y intermediates in ARS305 and bubbles in ARS202, regardless of the presence or absence of the DSB (Doksani et al., 2009). ARS202 fires later than ARS305, and bubbles can be detected at ARS305 at earlier time points (data not shown). As expected, HU-treated HO-inc rad53 cells fired both origins and accumulated X-shaped intermediates corresponding to reversed replication forks. DSB induction reduced reversed fork accumulation at ARS305, but not at the ARS202 locus (Figure 6). We note that 20% of the AR305-PDI1 region remained uncut following HO induction, thus accounting for the residual Xs observed at ARS305 in rad53 cells. These observations suggest that, in vivo, positive supercoiling is a driving force for fork reversal in checkpoint mutants. Moreover, the findings that DSB formation mimics the effect of Sac3 ablation in preventing fork reversal within the same locus suggest that the integrity of the topological domain spanning the replisome and the nuclear pore-associated transcribed gene (Casolari et al., 2005) influences the fate of stalled forks and that Rad53, Top2, TREX-2, and THO complexes and NPC proteins collaborate in controlling the S phase architecture of transcribed loci to prevent aberrant transitions at replicating chromosomes. DISCUSSION Transcription is coordinated with replication to maintain genome integrity. Because fork advance and the progression of the transcription bubble generate positive supercoiling, a head-on collision between replication and transcription causes topological impediments and fork pausing (Liu and Alberts, 1995; Olavarrieta et al., 2002; Wang, 2002). Fork restart can occur through the displacement of the RNA polymerase complex (Pomerantz and O’Donnell, 2010). In vitro studies showed that codirectional collision between replication and transcription has little effects on fork progression (Liu and Alberts, 1995; Pomerantz and O’Donnell, 2008) unless the RNA polymerase stalls (Elı ´as-Arnanz and Salas, 1997). There is evidence suggesting that the replisome can use mRNA as a primer after colliding codirectionally with RNA polymerase (Pomerantz and O’Donnell, 2008; Kogoma, 1997). Hence, it is not surprising that, in prokaryotes, which initiate DNA synthesis from a single origin, codirectional collision between forks and transcription bubbles is the preferred option (Brewer, 1988; Rocha, 2004). However, in multiorigin species, such as eukaryotes, transcription and replication collide both in head-on or codirectional ways. Indeed, transcribed genes always pause replication forks, regardless of their relative orientation (Azvolinsky et al., 2009; Bermejo et al., 2009), perhaps due to the Top2-mediated architecture of transcribed genes (Bermejo et al., 2009). We show that the TREX-2-THO-dependent coupling of transcription, gene gating, and mRNA biogenesis causes aberrant Cell 146, 233–246, July 22, 2011 ª2011 Elsevier Inc. 239 A C D E B Figure 5. Nucleoporin Mutations Abolishing Gene Gating Suppress rad53 Phenotypes (A) rad53-K227A and rad53-K227A mlp1Dcells were grown overnight with galactose (GAL) and treated with 0.2 M HU. The mean percentages of cells showing peripheral GAL loci and standard deviations (histogram error bars) from three independent experiments are shown. (B) Serial dilutions of WT, nup1D,rad53-K227A,rad53-K227A nup1D,mlp1D, and rad53-K227A mlp1Dcells plated in the absence (YPDA) or presence of 5 mM HU. (C) Evolutionary comparison of a portion of the Mlp1 globular domain containing residues phosphorylated by checkpoint kinases. Conserved residues are labeled in green. (D) Percentage of cells showing central or NPC-tethered (peripheral) GAL loci in WT, mlp1S1710A,mlp1S1710D, and mlp1Dcells grown in the presence of glucose (GLU) or galactose (GAL). (E) Serial dilutions of WT, mlp1S1710D,rad53-K227A, and rad53-K227A mlp1S1710D cells plated in the absence (YPDA) or presence of 5 mM HU. For related data, see Figure S2. 240 Cell 146, 233–246, July 22, 2011 ª2011 Elsevier Inc. transitions at stalled forks in checkpoint-defective cells. These observations, together with the findings that the genome-wide distribution of Hpr1 overlaps with that of RNA polymerase II and Top2 at transcribed genes (Bermejo et al., 2009), the checkpoint phosphorylates nuclear pore proteins (Chen et al., 2010; Smolka et al., 2007), Rad53 stoichiometrically associates with the Kap95 karyopherin (Smolka et al., 2005), and the relative distance between nuclear periphery and transcribed genes depends on the checkpoint, suggest that the physical continuity between transcribed genes and the nuclear envelope is modulated by the checkpoint to assist fork progression. The TREX-2 and THO-dependent topological impediments could, in part, Figure 6. DSB Induction Counteracts Fork Reversal in rad53 Mutants Replication intermediates in WT HO-inc, WT HO, rad53-K227A HO-inc, and rad53-K227A HO cells following afactor-induced G1 arrest, DSB formation by galactose addition, and release in 25 mM HU. Schematic representations of the 2D gel digestion strategies used are shown. Histograms indicate the ratio between reversed fork signal intensities at ARS305 and ARS202 for rad53 cells at each time point. For related data, see Figure S4. Cell 146, 233–246, July 22, 2011 ª2011 Elsevier Inc. 241