The SWR1 histone replacement complex causes genetic instability and genome-wide transcription misregulation in the absence of H2A.Z

The SWR1 Histone Replacement Complex Causes Genetic Instability and Genome-Wide Transcription Misregulation in the Absence of H2A.Z Macarena Morillo-Huesca 1. , Marta Clemente-Ruiz 1. , Eloı ´sa Andu ´jar 2 ,Fe ´lix Prado 1 * 1Department of Molecular Biology, CABIMER-CSIC, Seville, Spain, 2Genomics Unit, CABIMER-CSIC, Seville, Spain Abstract The SWR1 complex replaces the canonical histone H2A with the variant H2A.Z (Htz1 in yeast) at specific chromatin regions. This dynamic alteration in nucleosome structure provides a molecular mechanism to regulate transcription, gene silencing, chromosome segregation and DNA repair. Here we show that genetic instability, sensitivity to drugs impairing different cellular processes and genome-wide transcriptional misregulation in htz1Dcan be partially or totally suppressed if SWR1 is not formed (swr1D), if it forms but cannot bind to chromatin (swc2D) or if it binds to chromatin but lacks histone replacement activity (swc5Dand the ATPase-dead swr1-K727G). These results suggest that in htz1Dthe nucleosome remodelling activity of SWR1 affects chromatin integrity because of an attempt to replace H2A with Htz1 in the absence of the latter. This would impair transcription and, either directly or indirectly, other cellular processes. Specifically, we show that in htz1D, the SWR1 complex causes an accumulation of recombinogenic DNA damage by a mechanism dependent on phosphorylation of H2A at Ser129, a modification that occurs in response to DNA damage, suggesting that the SWR1 complex impairs the repair of spontaneous DNA damage in htz1D. In addition, SWR1 causes DSBs sensitivity in htz1D; consistently, in the absence of Htz1 the SWR1 complex bound near an endonuclease HO-induced DSB at the mating-type (MAT) locus impairs DSB-induced checkpoint activation. Our results support a stepwise mechanism for the replacement of H2A with Htz1 and demonstrate that a tight control of this mechanism is essential to regulate chromatin dynamics but also to prevent the deleterious consequences of an incomplete nucleosome remodelling. Citation: Morillo-Huesca M, Clemente-Ruiz M, Andu ´jar E, Prado F (2010) The SWR1 Histone Replacement Complex Causes Genetic Instability and Genome-Wide Transcription Misregulation in the Absence of H2A.Z. PLoS ONE 5(8): e12143. doi:10.1371/journal.pone.0012143 Editor: Michael Lichten, National Cancer Institute, United States of America Received July 9, 2010; Accepted July 21, 2010; Published August 12, 2010 Copyright: ß2010 Morillo-Huesca et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: The research was funded by the Spanish Ministry of Education and Science (BFU2006-08336 grant) and the Junta de Andalucia (P07-CVI-032000). M.C-R. was a recipient of a pre-doctoral training grant from the Spanish Research Council (CSIC). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected] .These authors contributed equally to this work. Introduction Genome organization and function rely on the precise assembly and dynamics of chromatin. The repeating unit of chromatin, the nucleosome, is formed by 146 bp of DNA wrapped twice around an octamer of histones. Histones H3 and H4 are first assembled into a core (H3/H4) 2 tetramer, which is the stable entity at physiological ionic strength. An H2A/H2B dimer associates on each side of the tetramer to form the histone octamer that is stabilized by the nucleosomal DNA [1]. ATP-dependent remodelling complexes, histone post-translational modifications and replacement of canonical with variant histones can later modify nucleosomes, thus altering the function of specific chromatin regions [2–4]. Variant histones alter the physicochemical properties of nucleosomes and thereby not only the interactions of nucleosomes with other factors but also their stability and DNA accessibility. One such variant, H2A.Z – Htz1 in yeast –, is an evolutionary conserved histone (90% sequence identity across species) with roles in transcription, silencing, genome integrity and cell cycle progression [5,6]. Htz1 is widely distributed throughout the yeast genome (in more than 65% of the genes) occupying preferentially the nucleosomes flanking the nucleosome-free region located at the transcription start site [7–11]. Htz1 is enriched at the promoter of basal/repressed genes where it facilitates transcription activation by histone loss [7–13]. Additionally, Htz1 antagonizes silencing by collaborating in the formation of a boundary that prevents the spreading of heterochromatin proteins [14]. These mechanisms, which appear to be conserved in vertebrate cells [15], provide an explanation for the elevated number of downregulated genes in the absence of Htz1 [14,16,17]. The fact that a similar number of genes is up-regulated in htz1Dhas also led to proposing a role for Htz1 in repression, even though no evidence has been provided yet. H2A.Z/Htz1 is also involved in genome stability. It is a structural component of centromeres [18,19] required for proper chromosome segregation [19,20]. In addition, the absence of Htz1 affects DNA replication and cell cycle progression and causes lethality or sickness in combination with S-phase checkpoint mutants [21]. These results, together with the sensitivity of htz1D to drugs causing DNA damage during DNA replication [16,17], suggest a role for Htz1 in the DNA damage response by replicative PLoS ONE | www.plosone.org 1 August 2010 | Volume 5 | Issue 8 | e12143 stress. Whether or not associated with these phenotypes, Htz1 is transiently recruited to double-strand breaks (DSBs) [22] but its role in DNA repair remains unclear. H2A.Z/Htz1 is incorporated into chromatin by the Swi2/Snf2- related SWR1 complex [10,16,17,23]. The 14-subunit yeast SWR1 has been extensively characterized in vitro. Purified SWR1 complex can specifically replace H2A/H2B with Htz1/ H2B in an ATP-dependent manner [17]. Swr1 is the catalytic subunit of the complex and the main scaffold for the assembly of the remaining subunits; Swc5, Swc2, Yaf9 and Arp4 are also required for histone replacement in vitro. In addition, Swc2, and less strongly the N-terminal region of Swr1, are the two components that interact directly with Htz1 [24,25]. Little is known, however, about the mechanisms of SWR1 targeting and Htz1 replacement in vivo. It has been shown that Swr1, Yaf9, the bromodomain-containing Bdf1 protein and the module formed by Swc2, Swc6, Arp6 and Swc3 are required for Htz1 incorporation into chromatin [9,10,16,17,23], but except for Swr1 the specific function of the remaining subunits and therefore the mechanism of replacement are still obscure. Here we show that in the absence of Htz1 the SWR1 complex causes genetic instability, sensitivity to stress conditions and genome-wide transcriptional misregulation. Our results are consistent with a stepwise mechanism of histone replacement that, in the absence of Htz1, affects chromatin integrity and function. Results The SWR1 complex causes genetic instability in the absence of Htz1 To uncover new mechanisms by which chromatin prevents genetic instability, we have analyzed the effects on recombination of mutants affected either in structural components or in remodelling factors of chromatin. This screening revealed a new function for the histone variant Htz1 in preventing the accumulation of recombinogenic DNA damage, as shown by an increase in the frequency of both genetic recombination between inverted repeats and budded cells with foci of the recombination protein Rad52 fused to the yellow fluorescence protein (Rad52-YFP) in htz1Dcells (Figure 1A Figure 1. The SWR1 complex causes genetic instability in the absence of Htz1. (A) Effect of htz1D,swr1Dand htz1Dswr1D(BY4741) on the frequency of inverted-repeat recombination. (B) Effect of htz1D,swr1D,swc2D,swc5D,htz1Dswr1D,htz1Dswc2D,htz1Dswc5D,swr1Dswc2Dand htz1Dswr1Dswc2D(BY4741) on the frequency of budded cells with Rad52-YFP foci. (C) Effect of htz1D,hta1/2 S129* and htz1Dhta1/2 S129* (W303-1a) on the frequency of budded cells with Rad52-YFP foci. (D) Scheme with the pathways of accumulation of recombinogenic DNA damage in htz1D,swr1D and swc2D. The frequency of recombination and budded cells with Rad52-YFP foci is presented as the average and standard deviation as described in Materials and Methods. Asterisks and circles indicate statistically significant differences compared to wild type and htz1D, respectively, according to a Student’s t-test, where one asterisk/circle represents a P-value ,0.0001, two represents ,0.005, and three represents ,0.025. doi:10.1371/journal.pone.0012143.g001 SWR1-Mediated Defects in htz1D PLoS ONE | www.plosone.org 2 August 2010 | Volume 5 | Issue 8 | e12143 and B, respectively). As expected by the fact that Swr1 is required for the incorporation of Htz1 into chromatin [17,23], the absence of Swr1 led to similar phenotypes (Figure 1A and B). Notably, however, htz1Dswr1Ddisplayed levels of genetic recombination and Rad52-YFP foci close to the wild type (Figure 1A and B). These results therefore support the existence of two pathways that lead to an accumulation of recombinogenic DNA damage, one associated with htz1Dthat depends on Swr1, and another associated with swr1Dthat depends on Htz1 (Figure 1D). The effect of Swr1 in htz1Dis likely to be mediated by the SWR1 complex because SWR1 remains intact in the absence of Htz1 [24]. Given that Swr1 is essential for the integrity of the complex, we decided to study genetic stability in the absence of either Swc2 or Swc5, two SWR1 subunits required for Htz1 transfer in vitro but not for the integrity of the complex [24]. Swc2 binds directly to Htz1 and this interaction is responsible for most of the Htz1 bound to the complex. By contrast, Swc5 is the only subunit absent in the complex purified from swc5Dcells [24,25] and therefore a convenient mutant to explore if the phenotypes mediated by Swr1 in htz1Drequire the histone transfer activity of the SWR1 complex. The absence of Swc2 increased the proportion of budded cells with Rad52-YFP foci (Figure 1B; [26]). However, this increase was also detected in the triple mutant htz1Dswr1Dswc2D, despite the fact that the double mutant htz1Dswr1Ddoes not accumulate Rad52-YFP (Figure 1B), supporting the existence of a swc2D- associated mechanism leading to Rad52-YFP foci that is independent of Swr1 and Htz1 (Figure 1D). By contrast, swc5D caused just a slight increase in the frequency of cells with Rad52- YFP foci, and more importantly, swc5Dpartially suppressed the hyper-recombination phenotype of htz1D(Figure 1B). Phosphorylation of H2A is one of the earliest molecular events in response to DNA damage [27,28] that is required for SWR1 binding to H2A in vitro [29] and to damaged DNA in vivo [30,31]. Therefore, we analyzed the effect of a H2A mutant that cannot be phosphorylated (hta1/2 S129* )onhtz1D-induced genetic instability. As shown in Figure 1C, hta1/2 S129* displayed a 5-fold increase in cells with Rad52-YFP foci, consistent with its defect in NHEJ but not in HR [28], but two-fold lower than in htz1D. Importantly, the effect of hta1/2 S129* was epistatic over htz1D, indicating that the accumulation of Rad52-YFP in htz1Drequires phosphorylation of H2A. The SWR1 complex causes DNA damage sensitivity in the absence of Htz1 The response to DNA damage is not similar in htz1Dand swr1D cells. While htz1Dis highly sensitive to the replication inhibitor hydroxyurea (HU) and the alkylating agent methyl methanesulfonate (MMS), swr1Dis either resistant or moderately sensitive depending on the genetic background (Figures 2A and S1A; [16,17]). Since the high density of cells in the drop-test assays can exacerbate growth defects and does not distinguish between lethality and slow growth, we determined the efficiency of plating in media with drug relative to the controls without drug. At the analyzed concentrations htz1Dcells died in HU and grew slowly in MMS (Figure S2), consistent with the fact these two agents cause different types of replicative DNA damage. We hypothesized that DNA damage sensitivity in htz1Dmight be mediated by Swr1. In agreement with this possibility we observed that swr1Dsuppresses – totally or partially depending on the genetic background – htz1D-mediated DNA damage sensitivity (Figures 2A, S1A and S2), further demonstrating that Swr1 leads to genetic instability in the absence of Htz1. Also, we observed that both swc2Dand swc5D, while displaying the same low sensitivity to HU as swr1D, suppressed htz1Dlethality (Figures 2B and S2A). These results suggest that DNA damage sensitivity in the absence of Htz1 requires the histone exchange activity of SWR1. To further support this point we analyzed a mutant carrying a K to G substitution in the ATP binding site of Swr1 that completely abolish its histone replacement activity but has no effect on the integrity of the complex [17]. We transformed swr1Dand htz1D swr1Dmutants with a plasmid expressing either SWR1 or swr1- K727G and tested their sensitivity to HU and MMS. Expression of the wild-type but not of the Swr1 ATPase-dead protein in htz1D swr1Dcaused DNA damage sensitivity, indicating that the histone replacement activity of SWR1 is responsible for genetic instability in the absence of Htz1 (Figures 2A and S2A). Finally, and in contrast to the accumulation of Rad52 foci, htz1Dsensitivity to HU was independent of H2A phosphorylation (Figure 2A). In response to a DSB the SWR1 complex incorporates Htz1 at the proximity of the break [22,32]. Therefore, we decided to determine whether Swr1 also lead to defective DSB repair in the absence of Htz1, as shown above for HU and MMS. To assess this possibility we first determined the sensitivity of mutant and wild type cells to DSBs generated by the endonuclease PvuII expressed from a GAL1 promoter variant with reduced basal activity [33]. While the growth of wild type and swr1Dwas not affected by the residual expression of PvuII under non-inducing conditions and equally affected upon induction of the GAL1 promoter for 8 hours, htz1Dwas highly sensitive even under residual PvuII expression. Remarkably, swr1Dsuppressed htz1Dsensitivity to DSBs under both conditions (Figures 2C and S1B). Similar results were obtained with the DSB-inducing drug phleomycin; htz1Dbut not swr1Dwas highly sensitive to the drug and this sensitivity was partially or totally suppressed – depending on the genetic background – in the double mutant (Figure 2D and S1B). Further analysis showed that phleomycin caused a loss of viability in htz1D (Figure S2A). These data indicate that the absence of Htz1 at chromatin has a minor impact in DSB repair and that Swr1 causes sensitivity to DSBs in htz1D. However, and despite the fact that SWR1 has been shown to be recruited to DSBs via P-H2A [31], the phosphoacceptor mutant hta1/2 S129* did not suppress htz1D sensitivity to phleomycin (Figure 2D). Since SWR1 is closely related to the chromatin remodelling complex INO80 and both SWR1 and INO80 are recruited to DSBs [29–31,34], where they appear to regulate the level of Htz1 [32], we decided to address the possibility that the effect of SWR1 on DSB repair in htz1Dwas mediated by INO80. Figure 2E shows that both swr1Dand ino80Ddisplayed a similar low sensitivity to phleomycin-induced DSBs, consistent with the minor role, if any, played by SWR1 and INO80 in NHEJ and DSB-induced HR [31,32]. Unlike swr1D, however, ino80Ddid not suppressed htz1D sensitivity, indicating that INO80 is not required for SWR1- dependent htz1Dsensitivity to DSBs. Altogether, our results, performed in two genetic backgrounds, demonstrate that SWR1 causes sensitivity to a number of different DNA lesions. The SWR1 complex impairs DSB-induced checkpoint activation in the absence of Htz1 To understand why SWR1 affects DSB repair in htz1Dwe first analyzed by ChIP the binding of the SWR1 complex to an induced DSB using a Myc-tagged version of Swr1. We used a yeast strain in which an unrepairable DSB at the mating-type (MAT) locus can be synchronously generated by continuous expression of the endonuclease HO from the galactose-inducible GAL1 promoter, and in which the deletion of the donor sequences HML and HMR prevents the repair of the DSB by HR (Figure 3A; SWR1-Mediated Defects in htz1D PLoS ONE | www.plosone.org 3 August 2010 | Volume 5 | Issue 8 | e12143 [35]). The efficiency of HO-induced cleavage at MAT was reduced as compared with previous results due to growth conditions (minimal versus rich medium; data not shown) and slightly affected in htz1D,swr1Dand htz1Dswr1D(Figure 3B; [22]). As shown in Figure 3C, and in contrast to a previous result [31], Swr1 was present at MAT before formation of the DSB and this binding was not altered over HO digestion. The accumulation of SWR1 at MAT before cleavage was not due to incomplete repression of the GAL1 promoter in raffinose because a similar enrichment was detected in glucose (data not shown). Also, Swr1 binding to MAT before and after DSB formation was not affected in htz1D, indicating that the absence of Htz1 has no effect on SWR1 binding to intact and broken DNA molecules. It is well established that during DSB repair the 59-ends generated upon the break are resected leaving 39-ended singlestrand DNA molecules that trigger the activation of the DNA damage checkpoint [36]. Notably, processing of the break has been shown to be different in htz1Dand swr1Dmutants. DNA resection and checkpoint activation – as determined by accumulation of phosphorylated H2A (P-H2A) and Rad53 (P-Rad53) – are affected during HO endonuclease-induced DSB repair in htz1D[22] but not in swr1D[31,32]. We hypothesized that SWR1 might impair DSB processing in the absence of Htz1. To address this possibility we followed the accumulation of P-H2A and P- Rad53 in response to an unrepairable DSB at the MAT locus (Figure 3D). As previously shown, the kinetics of Rad53 and histone H2A phosphorylation were not affected in swr1Dand delayed in htz1Ddespite equivalent cleavage efficiencies. More importantly, the absence of Swr1 suppressed the defect in checkpoint activation associated with htz1D(Figure 3D). Swr1 causes sensitivity to stress conditions in htz1D In addition to DNA damage, htz1Dand swr1Dhave been reported to be sensitive to stress conditions that impair different cell processes. In particular, they are sensitive to the microtubule polymerization inhibitor benomyl, a result consistent with their Figure 2. The SWR1 complex causes DNA damage sensitivity in the absence of Htz1. (A) DNA damage sensitivity of htz1D,swr1D,htz1D swr1D,hta1/2 S129* ,htz1Dhta1/2 S129* and of swr1Dand htz1Dswr1Dtransformed with plasmids pRS416-SWR1-2Flag or p416-swr1-2Flag-k727G (W303- 1a) as determined by plating ten-fold serial dilutions from the same number of mid-log phase cells onto SMM or SMM-U plates with or without HU or MMS, respectively. (B) DNA damage sensitivity of htz1D,swr1D,swc2D,swc5D,htz1Dswr1D,htz1Dswc2Dand htz1Dswc5D(BY4741) as indicated in (A). (C) Sensitivity of htz1D,swr1Dand htz1Dswr1D(W303-1a) to DSBs generated by expression of the endonuclease PvuII from a GAL1 promoter variant. DSB sensitivity of cells transformed with either pV10 (GAL1pr::PvuII) or pRS316 (empty vector) was determined by plating onto SMM-U tenfold serial dilutions from the same number of mid-log phase cells growth under non-inducing (glucose) or inducing (galactose for 8 hours) conditions. (D) DSB sensitivity of htz1D,swr1D,htz1Dswr1D,hta1/2 S129* and htz1Dhta1/2 S129* (W303-1a) as determined by plating ten-fold serial dilutions from the same number of mid-log phase cells onto YPD with or without phleomycin. (E) Phleomycin-induced DSBs sensitivity of htz1D, swr1D,htz1Dswr1D,ino80Dand htz1Dino80D(BY4733) as determined in (D). Cells were incubated at 30uC for 1–3 days as indicated. doi:10.1371/journal.pone.0012143.g002 SWR1-Mediated Defects in htz1D PLoS ONE | www.plosone.org 4 August 2010 | Volume 5 | Issue 8 | e12143 genetic interactions with mutations in components of the kinetochore and the Swr1-mediated deposition of Htz1 at centromeric regions [19]. However, this sensitivity is more pronounced in htz1Dthan in swr1D(Figures 4A; [16,19]). The absence of Htz1 has also been shown to cause sensitivity to the denaturing agent formamide and to the phosphatidylinositol-3- OH kinase related kinases (PIKKs) inhibitor caffeine, these sensitivities also being high in htz1Dand moderate in swr1D (Figures 4A [16]). Finally, the absence of both Htz1 and components of SWR1 display synthetic growth defects with the absence of the transcriptional elongation factor Dst1 [23], a result that, together with the sensitivity of htz1Dto the transcriptional elongation inhibitor 6-azauracil (6-AU) [37], points to a role for Htz1 during transcription elongation. We observed that htz1D sensitivity to 6-AU was less pronounced in swr1Dthan in htz1D (Figure 4B). In all cases, however, the differential sensitivity to drugs of htz1Das compared to swr1Dwas dependent on genetic background (compare Figure 4 with S1C). In addition, these drugs did not lead to a significant loss of viability and only in the case of formamide to a dramatic slow growth, being the growth defects exacerbated by the drop-test assay (Figure S2). More importantly, our study in two different genetic backgrounds showed that the absence of Swr1 suppresses the growth defects of htz1Dunder all tested stress conditions, including unperturbed conditions (Figures 4, S1C and S2). The SWR1 complex causes transcriptional misregulation in the absence of Htz1 A number of studies have provided strong genetic and molecular evidence for a function of Htz1 in transcription regulation [7,9,10,14,17,23]. Interestingly, comparison of the genome-wide transcription profiles for htz1Dand swr1Dshowed a high percentage of genes whose misregulation was specific to each mutation, a result that suggested the existence of non- Figure 3. The SWR1 complex impairs DSB processing in the absence of Htz1. (A) Schematic representation of the MAT locus and deleted HML and HMR donor sequences at chromosome III. The position of the DNA fragments amplified by qPCR for cleavage efficiency and ChIP analysis is indicated. (B) Accumulation of HO-induced DSBs at MAT over time. (C) Swr1-Myc enrichment at both sides of the cleavage site at MAT by ChIP analysis. Both input and ChIP DNA were amplified by real-time PCR with amplicons situated at the regions shown in (A) (see Table S4 for oligos). The enrichment at each zone is graphed relative to the enrichment in the wild-type strain incubated with IgG, taken as 1. Similar results were obtained using as a control the untagged strain incubated with the anti-Myc antibody. (D) DSB-induced phosphorylation of Rad53 (Rad53-P) and histone H2A (P-H2A) over HO digestion by western blot. All the experiments were performed in JKM179. doi:10.1371/journal.pone.0012143.g003 SWR1-Mediated Defects in htz1D PLoS ONE | www.plosone.org 5 August 2010 | Volume 5 | Issue 8 | e12143 overlapping functions for Swr1 and Htz1. This effect was particularly relevant in the case of htz1Dwith percentages of 86% and 64% of the total amount of htz1Dup- and downregulated genes, respectively [17]. In light of our previous results we decided to explore the possibility that Swr1 caused transcriptional misregulation in htz1D. With this aim the transcription profiles of single and double mutants and wild-type cells were determined by whole-genome microarray analysis (Figure 5 and Tables S1 and S2). A 2-fold expression change cutoff relative to wild type yielded 126 and 41 up-regulated genes and 198 and 108 down-regulated genes by htz1Dand swr1D, respectively (Figure 5A). Overall these numbers are 10–20-fold higher than those obtained by Mizugushi and co-workers with the same strains but in a rich medium [17], indicating that the unexpected low number of genes regulated by Swr1 and Htz1, considering that Htz1 is present in most yeast promoters, was mostly due to growth conditions. Apart from this effect and consistent with the spreading of heterochromatin proteins in the absence of Htz1 incorporation into chromatin [14], 25% of the genes down-regulated in htz1D accumulated near telomeres in small clusters (Htz1-activated domains; HZADs) and 60% of these genes were also downregulated by swr1D(Table S2). A comparative analysis of htz1Dand swr1Dtranscriptional profiles shows that the number of misregulated genes was much higher in htz1Dthan in swr1D. Also, most of the genes misregulated by htz1Dwere not misregulated by swr1D(84% and 63% for up- and down-regulated), as opposed to the number of genes misregulated by swr1Dthat were not misregulated by htz1D(51% and 31% for up- and down-regulated) (Figure 5A). More importantly, most of the htz1D-misregulated genes were suppressed by more than 1.5-fold by swr1D(68% and 49% of the up- and down-regulated genes) (Figure 5B) without a preferential association with genes located at HZADs or randomly distributed (Table S2). It is noted that this suppression by swr1Dmainly affected the genes that were specifically misregulated by htz1D(79% and 73% for up- and down-regulated, respectively) and not the genes misregulated by both htz1Dand swr1D(10% and 9% for up- and down-regulated, respectively) (Figure 5B). By contrast, the changes in the level of mRNA caused by the absence of Swr1 were not significantly affected by htz1D(Figure 5E). These results indicate that Swr1 affects the expression of a large number of genes in the absence of Htz1, and suggest that the major role for Htz1 in transcription is mediated by Swr1. Similar results were obtained with 1.5-fold- expression and 1.2-fold-suppression cutoffs (Tables S1 and S2). Next, we decided to determine whether transcriptional misregulation in the absence of Htz1 was also dependent on Swc2 and Swc5. As shown in Figure 5A, the number of genes up- and down-regulated either by htz1Dalone or htz1Dand swc2D(or swc5D) together were similar to those obtained with either htz1D alone or swr1Dand htz1Dtogether, respectively. Also, these groups of genes displayed a significant overlapping (Figures S3A and S3B). Importantly, swc2Dand swc5Dsuppressed a similar (Figure 5C and 5D) and common (Figure 5F) number of htz1D- misregulated genes as did swr1D, indicating that the SWR1 complex impairs transcription in the absence of Htz1 and that the major role of Htz1 in transcription occurs via SWR1-mediated histone replacement. It is also noted that misregulation by swc2D and swc5Dwas not affected by the absence of Htz1, except for a reduced number of genes (23 out of 172) whose swc5D-mediated change in mRNA levels was suppressed by htz1D(Figure 5E and Table S1), and that may reflect a residual activity of the SWR1swc5Dcomplex. Interestingly, the genes misregulated specifically by the absence of Swr1, Swc2 or Swc5 do not show a significant overlapping (Figure S3C), suggesting that the SWR1 complex does not have a physiological role in transcription regulation independent of Htz1. Swc2, but not Swc5 and Htz1, is required for Swr1 binding to chromatin, while Swc5 is required for histone replacement Our previous results indicate that genetic instability, sensitivity to stress and transcriptional misregulation in htz1Dare, to a greater or lesser extent, the consequence of the activity of the SWR1 complex. To get better insight into this mechanism we decided to determine what steps of the histone replacement reaction were prevented in our mutants. First, we analyzed SWR1 targeting to chromatin by ChIP analysis of strains harbouring a TAP-tagged version of Swr1. This construct is functional as indicated by the fact that SWR1-TAP displayed the same resistance to HU as the wild type (Figure 6A). We chose the promoter and an internal region of the BUD3 gene known to be enriched or not in Figure 4. Swr1 causes sensitivity to stress conditions in the absence of Htz1. Stress sensitivity of htz1D,swr1Dand htz1Dswr1D(W303-1a) as determined by plating ten-fold serial dilutions from the same number of mid-log phase cells onto YPD plates with or without benomyl, caffeine or formamide, or SMM without uracil (SMM-U) plates with or without 6-AU. Cells were incubated at 30uC for 2–3 days as indicated. doi:10.1371/journal.pone.0012143.g004 SWR1-Mediated Defects in htz1D PLoS ONE | www.plosone.org 6 August 2010 | Volume 5 | Issue 8 | e12143 Htz1, respectively [9], and three Swr1 enriched promoters (TOA1, SSM4 and YNL116w) [10]. As can be seen in Figure 6C, Swr1- TAP bound to chromatin in wild-type cells and, as shown above for the MAT locus, this binding did not require Htz1 (similar results were obtained with Myc-Swr1 binding to the TOA1 promoter in samples of Figure 3C; Figure S4A). Similarly, Swr1 binding to chromatin was not prevented by the absence of Swc5. However, Swr1 binding to promoters was impaired in swc2D despite this strain displaying wild-type levels of Swr1-TAP as determined by western analysis (Figure 6B). Figure 5. The SWR1 complex causes transcriptional misregulation in the absence of Htz1. (A) Venn diagrams showing the number of genes whose mRNA levels changed by more than 2-fold in mutants relative to the wild type and the number of genes that were commonly misregulated by htz1Dand either swr1D,swc2Dor swc5D.(B, C, D) Fraction of htz1Dmisregulated genes (2-fold cutoff) whose mRNA levels were changed by more than 1.5-fold (either suppressed or increased) in the double mutant htz1Dswr1D(B), htz1Dswc2D(C) or htz1Dswc5D(D) relative to htz1D.(E) Fraction of swr1D,swc2Dor swc5Dmisregulated genes (2-fold cutoff) whose mRNA levels changed by more than 1.5-fold in htz1Dswr1D, htz1Dswc2Dor htz1Dswc5Drelative to swr1D,swc2Dor swc5D, respectively. (F) Venn diagrams showing the number of htz1D-misregulated genes (2- fold cutoff) that were commonly suppressed by swr1D,swc2Dand swc5D. The genome-wide transcriptional analysis was performed in BY4741. doi:10.1371/journal.pone.0012143.g005 SWR1-Mediated Defects in htz1D PLoS ONE | www.plosone.org 7 August 2010 | Volume 5 | Issue 8 | e12143 SWR1-Mediated Defects in htz1D PLoS ONE | www.plosone.org 8 August 2010 | Volume 5 | Issue 8 | e12143 Swc5 has been shown to be required for histone transfer in vitro [24]. We decided to determine whether Swc5 is also required for Htz1 transfer in vivo by ChIP analysis in strains harbouring a functional TAP-tagged version of Htz1, as determined by HU and 6-AU sensitivity (Figure 6A; [23]). Figure 6D shows that Htz1 was not incorporated into chromatin in the absence of Swr1, consistent with previous results [17,23], while the absence of Swc5 nearly eliminated the amount of Htz1 bound to chromatin. Western blot analysis showed that the amount of Htz1-TAP in both mutants was 3–4-fold lower than in the wild type (Figure 6B) despite the levels of mRNA not being affected (data not shown), suggesting that Htz1 incorporation into chromatin prevents its degradation. Our molecular analysis demonstrates that Swc2 is required to target the SWR1 complex to chromatin, while Swc5 participates in the reaction of histone replacement. Altogether, our results let us conclude that genetic instability and transcriptional misregulation in the absence of Htz1 require the binding to chromatin (prevented in swc2D) and the histone replacement (prevented in swc5Dand swr1-K727G) activities of SWR1. The requirement for the histone replacement activity of SWR1 prompted us to look for an alteration in chromatin structure associated with htz1Dand to determine its dependence on Swr1. The analysis of nucleosome positioning and DNA accessibility by MNaseI and DNaseI digestion, respectively, of the INO1,DAN1 and GAL1 promoters did not reveal any difference in chromatin structure between the wild type, htz1D,swr1Dand htz1Dswr1D (Figures 6E and 6F). While a previous work reported a 20 bp shift in the positioning of nucleosome +2 of the GAL1 promoter in htz1Dwith a similar approach, this result was obtained with a wild type strain in which Htz1 was tagged with Myc [7]. We also tried to determine by ChIP analysis if the absence of Htz1 altered the stoichiometry of the nucleosomal histones at three promoters enriched in Htz1, and we did not detect any modification in the H2B/H3 ratio (Figure S4B). Overall our results are in line with both an independent study of four different chromatin regions [8] and a recent genome-wide analysis on nucleosome positioning [38] that have not detected any effect of htz1Don chromatin structure. Discussion The genome-wide distribution of Htz1 and its role in a number of processes from transcription and silencing to DNA repair and chromosome segregation make of this histone a key regulator of the genome dynamics. Here we show that genetic instability, sensitivity to drugs impairing different cellular processes and genome-wide transcription misregulation in htz1Dcan be partially or totally suppressed if the SWR1 complex is not assembled (swr1D), if it is assembled but cannot bind to chromatin (swc2Dand hta1/2 S129* in case of recombinogenic DNA damage) or if it can bind to chromatin but lacks histone replacement activity (swc5D and swr1-K727G), indicating that in the absence of Htz1 the nucleosome remodelling activity of SWR1 affects transcription and genetic stability. The mechanism of histone replacement has been suggested to occur in a stepwise manner [17] even though these steps remain unknown. We show that Swc2 is required for Swr1 binding to chromatin, and that this binding is not mediated by its interaction with Htz1 because it is not prevented by htz1D. These results, together with the fact that the mouse homolog of Swc2 (YL-1) binds to DNA in vitro [39] point to Swc2 being the subunit that targets SWR1 to chromatin. Upon binding, the SWR1 complex may promote histone replacement in a two-step manner, with the destabilization of H2A/H2B followed by deposition of Htz1/H2B via interactions with Swc2 and Swr1. It has been proposed that SWR1, as shown for SWI/SNF, may generate a dynamic DNA loop on the nucleosomal surface that promotes the intrinsic tendency of the histone octamer to dissociate the H2A/H2B dimer [17]. In vitro studies have shown that Swc5 is required for histone replacement [24]; our in vivo results support this conclusion and suggest that it is required for the destabilization of the H2A/H2B dimer. In this regard, the presence in Swc5 of a 60-residue N- terminal domain highly enriched in acidic amino acids (43%) characteristic of histone chaperones might provide the binding module to capture the H2A/H2B dimer as shown for Swi3 in SWI/SNF [40]. In a second step, Swr1 and Swc2 may deposit Htz1/H2B thus restructuring the nucleosome. In this context, the fact that in htz1Dthe SWR1 complex causes transcription misregulation and genetic instability by a mechanism that requires both the binding of SWR1 to chromatin and its histone replacement activity led us to propose that in the absence of Htz1 the nucleosome would remain transiently ‘‘destabilized’’ by SWR1 because of an attempt to replace H2A without Htz1, leading to a loss of chromatin integrity and function. However, we note the absence of changes in chromatin structure associated with htz1D(Figures 6 and S4B; [8,38]), a result that is consistent with subtle and transient alterations in nucleosome structure and/or a reduced population of affected molecules, but also with SWR1 affecting transcription and genetic stability without altering chromatin structure. For instance, SWR1 might be ‘‘trapped’’ at chromatin becoming a steric hindrance for DNA metabolic processes. However, the fact that the cellular defects associated with htz1Drequire the replacement activity of SWR1 (abolished in swc5Dand swr1-K727G) and that the enrichment in Swr1 at promoters and broken DNA ends is not significantly higher in htz1Dthan in wild type and swc5Dquestions this alternative model. Additional analysis will thereby be required to reconcile genetic and molecular data. Notably, the transcriptional effect of the SWR1 complex is more evident in genes that are specifically misregulated by htz1D. This result may be explained if we consider gene regulation by histone replacement as a dynamic process whose turnover rate can be low but not absent in those genes that are regulated by SWR1/Htz1 but do not require Htz1 under our experimental growth Figure 6. Swc2 and Swc5 are required for Swr1 binding to chromatin and histone replacement, respectively. (A) Swr1-TAP and Htz1- TAP functionality as determined by plating ten-fold serial dilutions from the same number of mid-log phase cells onto SMM plates with or without 200 mM HU. Cells were incubated at 30uC for 2–3 days as indicated. (B) Swr1-TAP and Htz1-TAP protein levels in mutants and wild type as determined by western blot. (C) Swr1-TAP and (D) Htz1-TAP enrichment at the promoters of BUD3,TOA1,SSM4 and YNL116w and an internal region of the coding sequence of BUD3 by ChIP analysis. Both input and ChIP DNA from untagged and tagged cells were amplified by real-time PCR with amplicons situated at the indicated regions (see Table S4 for oligos). The enrichment in the tagged strain at each zone is graphed relative to the enrichment in the untagged strain, taken as 1. ChIP experiments were performed in BY4741. (E) Nucleosome positioning at GAL1 in W303-1a and INO1 and DAN1 in BY4741 as determined by MNaseI digestion and indirect-end labelling. Spheroplasts from W303-1a cells were treated with different amounts of MNaseI, while BY4741 cells expressing the MNaseI were incubated with Ca 2+ to activate the nuclease for the indicated times. (F) DNA accessibility at the genes indicated in (E) by DNaseI digestion and indirect-end labelling. A scheme with the position of the upstream activation (UAS) and coding sequences (CDS) is shown on the left of each panel. doi:10.1371/journal.pone.0012143.g006 SWR1-Mediated Defects in htz1D PLoS ONE | www.plosone.org 9 August 2010 | Volume 5 | Issue 8 | e12143