|Home | About | Journals | Submit | Contact Us | Français|
INO80 is an evolutionarily conserved, ATP-dependent chromatin remodeling enzyme that plays roles in transcription, DNA repair, and replication. Here, we show that yeast INO80 facilitates these diverse processes at least in part by controlling genome-wide distribution of the histone variant H2A.Z. In the absence of INO80, H2A.Z nucleosomes are mis-localized, and H2A.Z levels at promoters show reduced responsiveness to transcriptional changes, suggesting that INO80 controls H2A.Z dynamics. Additionally, we demonstrate that INO80 has a novel histone exchange activity in which the enzyme can replace nucleosomal H2A.Z/H2B with free H2A/H2B dimers. Genetic interactions between ino80 and htz1 support a model in which INO80 catalyzes the removal of unacetylated H2A.Z from chromatin as a novel mechanism to promote genome stability.
DNA damage and aberrant chromosome replication can jeopardize genome integrity with serious effects to an organism’s health and survival. Several mechanisms have evolved in eukaryotic cells to cope with damaged DNA and to promote proper duplication of the genome. During recent years it has become apparent that chromatin structure plays an essential role in maintaining genomic integrity (Groth et al., 2007; Peterson and Cote, 2004). Specialized chromatin structures are formed during the DNA damage response or within S phase, promoting DNA repair and stabilizing replication forks. However, our understanding of how chromatin contributes to genome stability remains limited.
In addition to posttranslational modifications of histones, the building blocks of chromatin, incorporation of variant histones within chromatin regions provides an additional regulatory mechanism (Talbert and Henikoff, 2010). Histone variants such as H3.3 and H2A.Z are expressed throughout the cell cycle, and they can be incorporated into chromatin in the absence of DNA replication. Incorporation of the H2A-like H2A.Z into nucleosomal arrays alters their biophysical properties (Fan et al., 2002; Fan et al., 2004), potentially creating distinct chromatin structures that may regulate diverse metabolic processes. H2A.Z is highly conserved from yeast to human, and likewise the H2A.Z variant is enriched within nucleosomes at the proximal promoter regions of euchromatic genes of all eukaryotes (Mavrich et al., 2008; Raisner et al., 2005; Zhang et al., 2005). H2A.Z is also highly dynamic, being lost from promoters upon transcriptional activation at a rate that exceeds that of the core H3/H4 tetramer (Hardy et al., 2009; Zhang et al., 2005).
The SWI2/SNF2 family of ATP-dependent chromatin remodeling enzymes use the energy of ATP hydrolysis to alter histone-DNA interactions, leading to movements of nucleosomes in cis (“sliding”), loss of some or all histones, or the exchange of H2A/H2B dimers (Clapier and Cairns, 2009). The Ino80 and Swr1 ATPases belong to the INO80 subfamily of the SWI2/SNF2 group of remodeling enzymes (Morrison and Shen, 2009). Both Swr1 and Ino80 are subunits of highly conserved, multisubunit complexes, SWR-C and INO80, that share several common subunits (e.g. Rvb1,2) (Kobor et al., 2004; Krogan et al., 2003; Mizuguchi et al., 2004; Shen et al., 2000). INO80 can catalyze nucleosome sliding in cis (Shen et al., 2000), whereas SWR-C, or its metazoan ortholog SRCAP (Kobor et al., 2004; Krogan et al., 2003; Mizuguchi et al., 2004; Ruhl et al., 2006), directs incorporation of H2A.Z into nucleosomes by a dimer exchange reaction (Mizuguchi et al., 2004). In addition to a role in transcription, genetic studies indicate that INO80 regulates the DNA damage checkpoint response (Morrison et al., 2007; Papamichos-Chronakis et al., 2006) and stabilizes stalled replication forks (Papamichos-Chronakis and Peterson, 2008). Even though the importance of INO80 in genome stability is apparent, it is still unclear how INO80 contributes to these processes.
Here, we investigate the molecular mechanism of INO80 function in budding yeast. We present evidence indicating that INO80 regulates the genome-wide distribution of H2A.Z, and that it promotes the eviction of H2A.Z from promoters during transcriptional induction. We also demonstrate that purified INO80 complex can incorporate H2A into an H2A.Z nucleosome in vitro, indicating that it has a novel histone exchange activity that replaces nucleosomal H2A.Z/H2B with free H2A/H2B dimers. Notably, glutamine substitutions of the four N-terminal acetylatable lysine residues of H2A.Z alleviate the sensitivity of ino80 mutants to both replication stress and DNA damage inducing agents. Our data suggest that removal and replacement of unacetylated H2A.Z from chromatin by INO80 is an essential and novel mechanism for maintaining genome integrity.
Previously, we reported that a partial deletion of the INO80 gene (ino80Δ300) led to increased levels of H2A.Z within chromatin that surrounded a single, unrepaired DNA double strand break (DSB) (Papamichos-Chronakis et al., 2006). When H2A.Z distribution was analyzed by chromatin immunoprecipitation in a strain harboring a complete deletion of INO80 (ino80Δ), altered levels of H2A.Z were detected at many genomic locations even in the absence of a DSB (data not shown). In order to understand how H2A.Z localization may be altered upon inactivation of the INO80 complex, we analyzed the distribution of H2A.Z across 4% of the genome (chromosome III and 230 additional promoter regions) at single nucleosome resolution using a combination of mononucleosome-resolution ChIP with dense tiling microarrays (ChIP-chip). Briefly, cells from WT and ino80Δ strains were arrested in the G1 phase of the cell cycle, cells were fixed with formaldehyde, cross-linked chromatin was digested to mononucleosomes, and chromatin immunoprecipitation was conducted using a polyclonal antibody directed against H2A.Z. Total and immunoprecipitated DNA from the two strains were subsequently amplified, labeled with Cy3 and Cy5, and hybridized to a custom, high-resolution tiling microarray (Liu et al., 2005; Yuan et al., 2005).
In agreement with published results, the mononucleosomal H2A.Z map produced from the WT cells confirmed the distinct enrichment pattern of H2A.Z (Figure 1C, F). H2A.Z enrichment spanned a wide dynamic range, with the 2.5% most-enriched nucleosomes exhibiting 16-fold enrichment of H2A.Z relative to the 2.5% most H2A.Z-depleted nucleosomes (Fig 1A). However, the dynamic range of nucleosomal H2A.Z levels in the ino80Δ mutant was significantly compressed, with 95% of nucleosomes captured within a 6-fold dynamic range (Fig 1A). This is visualized by plotting nucleosome enrichments for H2A.Z in the ino80Δ cells against those of the WT cells, showing that the slope of WT vs. ino80Δ enrichments is well below one (y=0.547x), (Fig. 1B). Because of standard microarray normalization, these data are equally consistent with global over-incorporation of H2A.Z, global under-incorporation, or mixed gain and loss at specific loci. However, equal levels of bulk, chromatin-associated H2A.Z were detected in the WT and ino80Δ strains (Fig. 1D). These results indicate that INO80 does not impact the total amount of H2A.Z that is incorporated into chromatin, but rather there is an extensive reorganization of nucleosomal H2A.Z across the genome in the absence of Ino80. This can be seen at many loci as spurious incorporation of H2A.Z in the ino80Δ mutant strain, together with a drop in H2A.Z levels at H2A.Z-rich domains (Fig 1C and Figure S1).
To gain further insight into the regulation of H2A.Z localization by the INO80 complex, we focused on RNAPII-transcribed genes. The typical open reading frame is characterized by high levels of H2A.Z at the first nucleosome (+1) at the transcription start site (TSS), with variable H2A.Z at the upstream (−1) nucleosome, and low levels of H2A.Z downstream of the +1 nucleosome (Albert et al., 2007; Raisner et al., 2005; Zhang et al., 2005). As shown in Figure 1E, H2A.Z mislocalization was especially pronounced at promoter nucleosomes (y=0.487x). Interestingly, the decrease of H2A.Z at the +1 nucleosome was associated with concomitant gain of H2A.Z at nucleosomes inside the coding sequences (mid and 3’ CDS), (Fig. 1F). Together, these data demonstrate that H2A.Z becomes globally mislocalized in the ino80Δ mutant, and they support a role for the INO80 complex in regulating proper genome-wide H2A.Z localization.
In budding yeast, H2A.Z occupancy negatively correlates with transcription rates, with H2A.Z being highly enriched in most gene promoters but depleted upstream of very highly transcribed genes (Zhang et al., 2005). As an initial test to investigate whether INO80 plays a role in this process, a mononucleosomal ChIP-chip assay for H2A.Z was conducted in G1 and G2/M arrested WT and ino80Δ cells. Scatter plot analysis of H2A.Z nucleosome occupancy demonstrated that H2A.Z genomic occupancy is altered between the two cell cycle phases in the WT strain (y=0.635x; Fig. 2A). However, in the ino80Δ mutant the H2A.Z nucleosomal pattern remains largely unchanged, consistent with INO80 regulating the dynamics of H2A.Z-containing nucleosomes (y=0.953x; Fig. 2B).
Expression of KAR4 is highly induced when cells are arrested in G1 by mating pheromone, and it is repressed in G2 phase (Kurihara et al., 1996). As shown in the heat maps of the KAR4 locus in the WT strain (Figure 2C, left panel), H2A.Z is enriched at the repressed KAR4 promoter in G2/M and becomes, as expected, depleted during transcriptional activation in G1 cells. As expected, the enrichment of H2A.Z at other, non-cell cycle regulated genes remains unchanged between G1 and G2/M samples Figure S2C). In contrast, H2A.Z levels in the ino80Δ mutant remain high and similar to the repressed level in both G1 and G2/M phases (Fig 2C, right panel). Notably, induction of KAR4 expression is not affected by inactivation of Ino80 (Figure S2A), and thus transcription levels do not explain the altered H2A.Z dynamics. These results also indicate that the failure to deplete H2A.Z during transcriptional induction has little effect on KAR4 expression. These results suggest that INO80 controls either the eviction of the H2A.Z/H2B dimers or the loss of H2A.Z-containing nucleosomes that occurs during transcriptional induction.
To investigate how INO80 controls H2A.Z dynamics, nucleosome scanning analysis was used at the KAR4 locus. 45 overlapping primer pairs were used to monitor the translational position and dynamics of H2A.Z nucleosomes at the KAR4 promoter in the presence or absence of INO80. When samples were analyzed from G2/M arrested WT or ino80 cells, this analysis yielded four peaks, indicating four positioned nucleosomes flank the KAR4 promoter in the repressed state (Figure 2D). Notably, the positioning of these promoter proximal nucleosomes was identical in the presence or absence of Ino80 (Figure 2D). Strikingly, KAR4 promoter sequences were also severely depleted from mononucleosome samples of both G1-arrested WT and ino80 cells, indicating that nucleosomes are depleted when KAR4 is expressed (Figure 2D, E).
To measure H2A.Z occupancy at KAR4 promoter nucleosomes, chromatin immunoprecipitation of H2A.Z was conducted. In the repressed state (G2/M), H2A.Z is enriched at each of the four promoter nucleosomes, with the highest levels seen for nucleosome +1 (Figure 2F). At this promoter, similar levels are observed in the ino80Δ mutant, as expected from Figure 2C. Upon activation of KAR4 (G1 cells), the amount of H2A.Z per nucleosome is lower at several KAR4 nucleosomes in the WT strain, indicating that H2A.Z is evicted (Figure 2G and Figure S2B). In contrast, the amount of H2A.Z per nucleosome does not decrease in the ino80Δ mutant, with levels remaining at the repressed, G2/M level or even higher (Figure 2G and Fig. S2B). Taken together, these results suggest two independent and complementary pathways for H2A.Z eviction -- the first pathway is driven by complete nucleosome loss, and the second, H2A.Z-specific eviction, requires INO80.
These data indicate that INO80 regulates the genome-wide localization of H2A.Z as well as the eviction of H2A.Z during transcriptional induction. One simple possibility is that INO80 might preferentially slide H2A.Z nucleosomes or evict H2A.Z octamers during transcriptional induction. However, INO80 shows no detectable octamer eviction activity with either H2A or H2A.Z mononucleosomes, and we find that INO80 mobilizes H2A or H2A.Z nucleosomes with equal efficiency (Figure S3A, B). Since Ino80 and Swr1 belong to the same subfamily of Snf2-like ATPases, we tested whether Ino80 might catalyze an ATP-dependent H2A.Z/H2B dimer exchange event that removes H2A.Z and incorporates H2A.
Histone exchange assays were performed using mononucleosomes reconstituted with recombinant yeast histones. Initially, mononucleosomes were assembled with H2A/H2B dimers, and these substrates were incubated with remodeling enzyme and free HA-tagged H2A.Z/H2B dimers. Following incubation, the reactions were electrophoresed on native PAGE to separate bona fide mononucleosome products from free histones or other types of nucleosomal products. Histone exchange was evaluated by western blotting, probing for loss of H2A and incorporation of HA-tagged H2A.Z into the mononucleosome (Fig. 3A). In all experiments, mononucleosome integrity was analyzed by both western analysis of histone H2B and by visualizing DNA with ethidium bromide (Figure 3 and data not shown). As expected, the SWR-C complex showed robust, ATP-dependent incorporation of HA-H2A.Z and significant loss of H2A. In contrast, and consistent with previous studies, Ino80 showed little activity in this H2A.Z incorporation assay (Figure 3B and Figure S3E). Importantly, these same preparations of Ino80 showed robust ATPase and nucleosomal sliding activities (Figure S3B–D).
Next, mononucleosomes were assembled with H2A.Z/H2B dimers and incubated with remodeling enzyme and FLAG-tagged H2A/H2B dimers. Strikingly, Ino80 catalyzed ATP-dependent incorporation of FLAG-H2A into a mononucleosome product, whereas SWR-C, SWI/SNF, and RSC were inactive on this substrate (Figure 3C and data not shown). The INO80-dependent incorporation of FLAG-H2A was concentration dependent and increased with time of incubation (Figure 3D and data not shown), and titration of the mononucleosome substrate indicates that efficient exchange activity for 5nM INO80 requires >50 nM nucleosomes (data not shown). Strikingly, INO80 action catalyzed removal of 35% of the H2A.Z from the initial mononucleosome substrate, while the levels of H2B remained constant (Figure 3E). Notably, INO80 does not exhibit promiscuous dimer eviction activity, as INO80 does not catalyze H2A eviction from an H2A-containing mononucleosome (Figure S3E). Thus, these data indicate that INO80 catalyzes a dimer exchange reaction in which nucleosomal H2A.Z/H2B is replaced with an H2A/H2B dimer.
Previous analyses of ATP-dependent dimer exchange activities have used biotinylated chromatin substrates immobilized on streptavidin magnetic beads (Mizuguchi et al., 2004). In these assays, the immobilized substrate is incubated with remodeling enzyme and free histones, and exchange events are monitored by western blot following magnetic bead capture of the chromatin substrate. We performed this strategy with biotinylated mononucleosomes reconstituted with H2A.Z/H2B dimers, and we found that Ino80 catalyzed FLAG-H2A incorporation in this assay as well (Figure 3F). Furthermore, no detectable incorporation of FLAG-H2A was observed when INO80 was incubated with a mononucleosome reconstituted with an H2A/H2B dimer (Figure 3G). Together, both types of assays indicate that INO80 can specifically replace nucleosomal H2A.Z with H2A.
INO80 plays roles in many nuclear events, including gene transcription, DNA replication, DNA repair, and sister chromatid cohesion (Conaway and Conaway, 2009). One possibility is that INO80 regulates these diverse events by its action on H2A.Z, and consequently the defects observed in an ino80 mutant may be due to the mis-localization and aberrant chromatin dynamics of H2A.Z. One simple prediction of this model is that H2A.Z depletion might rescue the defects of an ino80 mutant. Unfortunately, htz1Δ ino80Δ and swr1Δ ino80Δ double mutants are inviable, suggesting that H2A.Z and Ino80 may play additional, redundant role(s) in an essential function (Figure S4A and data not shown). To overcome this problem, we created isogenic WT and ino80Δ strains in which HTZ1 is expressed from a chromosomal, truncated promoter at ~10% WT levels (HTZ1CP) (Fig 4A). This reduced expression of HTZ1 leads to a 4-fold decrease in bulk H2A.Z chromatin association, and a ~2-fold decrease at the positioned nucleosomes of the KAR4 locus (Figure 4B and Figure S4B). The HTZ1 CP allele fully complements the growth defect and thiobendazol sensitivity of an htz1Δ strain (data not shown), indicating that this level of H2A.Z is sufficient to perform its known functions.
Previously we showed that ino80 cells are incapable of completing DNA replication when exposed to replication stress conditions (Papamichos-Chronakis and Peterson, 2008). We investigated whether decreased expression of H2A.Z can rescue this replication defect. WT and ino80Δ cells that expressed either normal (HTZ1) or low levels of H2A.Z (HTZ1CP) were arrested in G1 and then released into media containing 40mM HU and their progression through S phase was followed by FACS. Both HTZ1 and HTZ1CP WT strains progressed normally through S phase (Fig 4C). As we showed previously, ino80Δ cells that express normal levels of H2A.Z are rapidly blocked in S phase (Fig. 4C, left panel). Interestingly, lowering the expression of H2A.Z restored a normal rate of S phase progression in HU media in the absence of Ino80 (Fig. 4C, right panel). In contrast, both the ino80 and HTZ1CP ino80 cells failed to grow in media lacking inositol (data not shown), indicating that lowered expression of HTZ1 cannot support transcription of the INO1 gene in the absence of INO80. Thus, these results implicate a close functional relationship between INO80 and H2A.Z, and suggest that aberrant H2A.Z incorporation may have a negative impact on DNA replication fork stability.
The N-terminal domain of yeast H2A.Z is acetylated in vitro at lysines 3, 8, 10 and 14 by the NuA4 HAT complex (Babiarz et al., 2006; Keogh et al., 2006; Millar et al., 2006), and acetylation occurs at promoter nucleosomes in vivo after incorporation of H2A.Z into chromatin by SWR-C (Keogh et al., 2006). Given that H2A.Z was mislocalized in the absence of Ino80, we tested whether H2A.Z acetylation levels might be altered in the ino80Δ mutant. Strikingly, H2A.Z-K14 acetylation levels were much lower in the ino80Δ strain compared to WT (Fig. 5A).
We entertained the possibility that this defect in H2A.Z acetylation contributes to the genome instability phenotypes of the ino80Δ mutant. However, a strain that harbors a derivative of H2A.Z that cannot be acetylated, H2A.Z-K3,8,10,14R, does not show sensitivity to DNA damage or replication stress agents (Millar et al., 2006), indicating that the lack of H2A.Z acetylation is insufficient to cause genome instability phenotypes. Interestingly, H2A.Z-K3,8,10,14R shows synthetic sensitivity to replication stress and DNA damage agents when expressed in an ino80Δ strain (Figure 5B). These results reveal a novel role for Htz1 acetylation in DDR and replication stress survival and suggest a functional connection between H2A.Z acetylation and INO80.
HDA1 encodes a histone deacetylase that regulates H2A.Z acetylation (Lin et al., 2008). As shown in Figure 5A, inactivation of Hda1 led to a large increase in H2A.Z-K14 acetylation in both the WT and ino80 strains. These data support a simple model in which H2A.Z can be acetylated in the absence of INO80, but that it is deacetylated by Hda1, possibly due to its mis-localization.
This data raise the interesting possibility that the accumulation of deacetylated H2A.Z in the ino80 mutant might be detrimental to genome stability. Deletion of HDA1 renders cells sensitive to DNA damage-inducing agents like MMS and Zeocin, but not to replication stress induced by hydroxyurea (Begley et al., 2002) and Figure S5). In our initial studies, we found that an ino80Δ hda1Δ double mutant has a severe slow growth phenotype that made growth assays problematic. To circumvent this issue, we monitored the phenotype of an arp8Δ hda1Δ double mutant that did not show this synthetic phenotype. The Arp8 subunit is essential for the chromatin remodeling activities of INO80, and an arp8Δ mutant shows sensitivity to replication stress (HU) and DNA damaging agents (zeocin). Strikingly, deletion of HDA1 suppresses the HU sensitivity of an arp8Δ mutant (Figure 5C). These results suggest that the replication defects caused by inactivation of the INO80 complex can be rescued by removing the Hda1 HDAC.
To further test whether constitutive H2A.Z acetylation can alleviate ino80Δ phenotypes, we created a putative H2A.Z acetyl mimic (HTZ1 K3,8,10,14Q). Initially, we tested whether expression of H2A.Z-K3,8,10,14Q could rescue the replication defects of an arp8 mutant during replication stress conditions. WT and arp8 cells that express either H2A.Z or H2A.Z-K3,8,10,14Q were arrested in G1 and subsequently released into 40mM HU, and their progression through S phase was followed by FACS. As shown in Figure 6A, WT cells progressed through S phase and completed DNA replication in approximately 100 minutes. In contrast, the arp8 cells proceeded through S phase slowly, unable to fully replicate their genome even after almost 6 hours in HU. However, expression of the H2A.Z panacetyl-mimic in the arp8 strain enabled cells to duplicate their genome, albeit slowly (Fig. 6A). In addition, expression of the H2A.Z panacetyl-mimic appears to alleviate the fork collapse phenotype of an ino80 mutant, as wildtype levels of DNA polα are recovered at a stalled replication fork in the absence of Ino80 (Figure S6A). Expression of H2A.Z-K3,8,10,14Q also alleviated the growth sensitivity of arp8, arp5, and ino80 mutants to HU, as well as to the DNA damage inducing agents MMS and zeocin (Fig 6B). Importantly, the htz1-4KQ strain has no apparent phenotype in the presence of INO80 (Fig. S6B). In contrast, expression of H2A.Z-K3,8,10,14Q did not alleviate the MMS or HU sensitivity of an mre11Δ mutant (Figure S6C), indicating that suppression is specific to mutations that disrupt the INO80 complex. Interestingly, expression of H2A.Z-K3,8,10,14Q did not suppress the inositol auxotrophy of an arp8 mutant, and ARP8-dependent transcription of the INO1 gene was not alleviated by H2A.Z-K3,8,10,14Q (Figure 6C,D). Importantly, suppression of arp8Δ genome stability phenotypes by the panacetyl mimic are eliminated after re-introduction of a wildtype copy of HTZ1 (Fig 6E). These data indicate that H2A.Z-K3,8,10,14Q is a potent suppressor of the genomic instability phenotypes of strains that lack the INO80 complex.
One simple explanation for why the H2A.Z-K3,8,10,14Q might suppress ino80 phenotypes posits that this H2A.Z derivative is not properly expressed or that it restores the wildtype chromatin distribution and dynamics of H2A.Z in the absence of Ino80. We find however, that H2A.Z-K3,8,10,14Q is expressed and incorporated into chromatin at levels similar to wild type as measured by ChIP and nucleosome scanning assays (Figure S6B,D, E). Moreover, mapping of H2A.Z-K3,8,10,14Q at KAR4 nucleosomes demonstrated that both H2A.Z-K3,8,10,14Q and wild type H2A.Z were incorporated in high amounts in the arp8 mutant compared to the WT strain, and neither H2A.Z-K3,8,10,14Q or H2A.Z were lost from KAR4 promoter nucleosomes upon transcriptional induction in the absence of INO80 (Fig 6F,G). These results suggest that the activity of INO80 is not sensitive to the acetylation status of H2A.Z, and that both WT and the H2A.Z-K3,8,10,14Q derivative require INO80 action for proper localization. Consistent with this view, the in vitro histone exchange activity of INO80 is not affected by substitution of H2A.Z N-terminal lysines by either arginine or glutamine residues (Figure S6F and data not shown). Collectively, these results are consistent with a model in which the mis-localization of unacetylated H2A.Z in the absence of INO80 is detrimental to genome integrity, but that constitutive H2A.Z acetylation counteracts these inhibitory effects.
Whereas previous studies have focused on the key role of the yeast SWR-C and mammalian SRCAP remodeling enzymes in directing the ATP-dependent deposition of the H2A.Z histone variant, here we have shown that the related INO80 enzyme catalyzes the replacement of nucleosomal H2A.Z for H2A within coding regions and the eviction of H2A.Z during transcriptional activation. Interestingly, this role for INO80 appears essential for the maintenance of genome stability, as decreased expression of H2A.Z or expression of a H2A.Z pan-acetyl mimic alleviates the sensitivity of ino80, arp5, or arp8 mutants to DNA damaging or replication stress agents. Thus, our genetic interactions suggest that aberrant accumulation of unacetylated H2A.Z has a negative impact on DNA double strand break repair and DNA replication fork stability.
Although members of the SWI/SNF subfamily of remodeling enzymes are able to evict histone dimers or entire octamers from nucleosomal substrates, only members of the INO80 subfamily exhibit histone dimer deposition and/or exchange activity. Both the yeast SWR-C and mammalian SRCAP members can replace nucleosomal H2A with H2A.Z, and in this study we report that yeast INO80 can perform the opposite reaction, converting an H2A.Z mononucleosome into one that contains H2A. Why hasn’t the dimer exchange activity of INO80 been detected previously? We find that optimal dimer exchange activity requires nucleosome concentrations >50 nM (S. Watanabe, unpublished observations), whereas most previously published dimer exchange assays have used much lower nucleosome concentrations (<10 nM; Mizuguchi et al., 2004). The Ino80 and Swr1 ATPases are the only members of the Snf2 family of ATPases that contain very large, ~300–500 amino acid insertions between Helicase/ATPase motifs III and IV, and it seems likely that this insertion influences the outcome of the remodeling reaction (Clapier and Cairns, 2009). In addition, each enzyme has a unique complement of histone binding subunits that may determine the specificity of the deposition reaction. For instance, the SWR-C contains the SWC2 subunit, key for H2A.Z recognition (Wu et al., 2005), as well as the Yaf9 subunit which harbors a YEATS domain that binds H3/H4 (Wang et al., 2009). Furthermore, the SWR-C and INO80 complexes each harbor the Arp4 and Arp5 subunits that interact with H2A/H2B dimers (Shen et al., 2003), and previous mass spectrometry data indicate that both H2A and H2A.Z are associated with the purified INO80 complex even in the absence of DNA damage (Mizuguchi et al., 2004).
Why is H2A.Z mislocalized in the absence of INO80 and how is INO80 action targeted to create the wildtype pattern? The SWR-C enzyme is localized predominantly to the +1 and/or −1 nucleosomes proximal to many RNAPII promoters, consistent with the deposition of H2A.Z at these locations (Venters and Pugh, 2009; Shimada et al., 2008). One possibility is that SWR-C typically incorporates H2A.Z within a large number of nucleosomes that encompass and flank a target promoter (Figure 7). In this case, we envision that INO80 confers boundary function, removing the H2A.Z from coding region nucleosomes, reinforcing the targeted, SWR-C-dependent deposition at promoter nucleosomes. In this model, both SWR-C and INO80 may be recruited to the promoter or coding region of a target gene, or alternatively, INO80 may function in a more general fashion, much like that proposed for the global action of histone deacetylases. Two studies have provided evidence that INO80 may be targeted to the coding regions of many genes transcribed by RNAPII (Klopf et al., 2009; Venters and Pugh, 2009), perhaps through interactions with the transcription elongation complex.
A second model is based on the fact that INO80 is associated with stalled and elongating replication forks (Papamichos-Chronakis and Peterson, 2008; Shimada et al., 2008; Vincent et al., 2008), and in this capacity INO80 controls elongation rate and fork stability. One attractive possibility is that INO80 at the replication fork may remove or replace H2A.Z that might be mis-localized during the chromatin assembly process that occurs following fork passage. In this case, H2A.Z may be deposited ectopically by fork-associated histone chaperones, or deposited aberrantly by SWR-C. In either case, removal by Ino80 might then facilitate the re-incorporation of H2A.Z at the “proper” locations by SWR-C.
H2A.Z contains several lysine residues that are subject to reversible acetylation in all systems where it has been investigated. In Tetrahymena, these H2A.Z lysines are essential for cell viability (Ren and Gorovsky, 2001), whereas in budding and fission yeast the substitution of H2A.Z N-terminal lysines by arginine results in sensitivity to drugs that impact chromosome segregation, but no other obvious phenotypes (Keogh et al., 2006; Kim et al., 2009). Acetylated H2A.Z is enriched at transcriptionally active promoters where H2A.Z is preferentially evicted, and it has been suggested that H2A.Z acetylation may facilitate re-assembly of H2A.Z nucleosomes during gene repression (Millar et al., 2006). However, substitution alleles that remove H2A.Z lysines do not have a major impact on gene expression profiles (Millar et al., 2006). Unfortunately, commercial antibodies that recognize acetylated H2A.Z do not function in ChIP assays, so detailed analysis of the distribution and dynamics of acetylated H2A.Z has not been possible (Keogh et al, 2006; M.P. and C.L.P, unpublished results).
We were surprised to find that INO80 has a large impact on the steady state levels of H2A.Z acetylation. Indeed, the analysis of bulk H2A.Z acetylation suggests that most if not all of the mis-localized H2A.Z is likely to be unacetylated in an ino80 mutant. Strikingly, the genetics indicate that it is the mis-localization of unacetylated H2A.Z that has a major impact on genome stability, not mis-localization of H2A.Z per se. The combination of mutations that disrupt the INO80 complex and the H2A.Z 4K-Q version suppresses the sensitivity of ino80, arp5, or arp8 mutants to DNA damage and replication stress agents. In contrast, expression of the H2A.Z 4K ->R version causes an enhanced sensitivity to these same agents. These data indicate that mis-localization of unacetylated H2A.Z is an inhibitor of genome stability that must either be acetylated or be removed by INO80.
The distinctive enrichment of the H2A.Z histone variant at promoter proximal nucleosomes has led to the pervasive view that H2A.Z is a key regulator of transcription that creates a more permissive environment for transcriptional activation. However, at least in yeast, loss of H2A.Z has a relatively minor effect on gene expression profiles, typically affecting only the transcriptional kinetics of a subset of inducible genes (Meneghini et al., 2003). However, our studies indicate that mis-localized H2A.Z exerts a general, repressive effect on processes that prevent genomic instability. Thus, although the promoter localization of H2A.Z provides a sensitive readout for proper deposition, perhaps the prevention of H2A.Z mis-localization by INO80 is more important than actual promoter proximal positioning. One intriguing possibility is that promoter localization places H2A.Z in a location that enhances its removal, thereby limiting its inhibitory effects on genome stability and allowing it to be used as a mechanism of transcriptional regulation.
Why does mis-localized, unacetylated H2A.Z impact DSB repair and replisome function? One possibility is that nucleosomal arrays that contain large amounts of H2A.Z assume more compact, folded states that block access of repair enzymes or destabilize stalled forks. Indeed, in vitro studies indicate that H2A.Z incorporation facilitates formation of condensed 30nm-like fibers (Fan et al., 2002; Fan et al., 2004). Alternatively, perhaps H2A.Z nucleosomes are inherently more dynamic, and genome stability is impacted by the inappropriate localization of dynamic nucleosome hotspots. And finally, the acetylation state of H2A.Z may regulate interactions with protein(s) that promote or hinder genome stability. In either case, the H2A.Z pan acetyl mimic suppresses defects in both DNA damage repair and replication stress pathways due to loss of INO80, suggesting a common function for acetylated H2A.Z.
Elucidating the mechanisms that protect genome stability is an essential step towards understanding and fighting devastating diseases like cancer (Halazonetis et al., 2008). Our work has uncovered a novel chromatin-mediated pathway essential for the maintenance of genome integrity that implicates the function of the INO80 chromatin remodeling enzyme on H2A.Z-containing chromatin. Recently, two groups provided evidence that the human INO80 complex also participates in DNA damage repair and in DNA replication, promoting genome stability (Hur et al., 2010; Wu et al., 2007). Additionally, studies in cancer patients have reported over-expression of H2A.Z in several major types of malignancies (Dunican et al., 2002; Rhodes et al., 2004; Svotelis et al., 2010; Zucchi et al., 2004). Given that the INO80 complex is highly conserved throughout evolution, both structurally and functionally (Conaway and Conaway, 2009), it would be particularly interesting to test whether the metazoan INO80 complex, similar to its yeast counterpart, regulates the localization and dynamics of the H2A.Z histone variant in higher eukaryotes.
ChIPs were performed as described ( Liu et al., 2005; Papamichos-Chronakis and Peterson, 2008) using commercially available polyclonal antibodies raised against H2A.Z (Millipore and Abcam antibodies were used for microarray analyses; Millipore and Active Motif antibodies were used for nucleosome scanning assays). Antibody specificity was confirmed by both ChIP and western analyses (Figure S7). Mononucleosomes were prepared as described (Liu et al., 2005). The recovered DNA was subjected to quantitative real-time PCR. All ChIPs were performed at least twice and the variation between experiments was 10–25%. Primers used in the PCR reactions are available upon request. Microarray hybridization and analysis were conducted as described (Liu et al., 2005). Data availability: All microarray data used in this study have been deposited to GEO (http://www.ncbi.nlm.nih.gov/projects/geo/), accession: GSE25722.
Chromatin fractionation was conducted as described (Liang and Stillman, 1997; Wang et al., 2009). For MNase release of nucleosome associated proteins from the chromatin pellet, pellets were resuspended in 200µl Lysis-1% Triton X buffer containing 1mM CaCl2 and 15 units of MNase. Samples were incubated at 37°C for 20 minutes and reaction was stopped by the addition of 1mM EGTA and 1mM EDTA. Samples were subsequently centrifuged at 14K rpm for 5 minutes at 4°C and the supernatant was recovered for protein and DNA analysis. Equal MNase digestion was confirmed by agarose gel visualization of the released DNA.
Cell-cycle arrest and FACS were performed as described (Papamichos-Chronakis and Peterson, 2008).
INO80-TAP and SWR1-TAP were purified as described (Sinha et al., 2009). ATPase assay and remodeling assays were performed as described (Logie and Peterson, 1999). Recombinant yeast histones were expressed and purified from E.coli, and octamers were reconstituted as described (Luger et al., 1999a, b).
Mononucleosomes were reconstituted by salt dialysis onto a 200 bp DNA fragment containing the 601 nucleosome-positioning sequence. Mononucleosomes were incubated with remodeling enzymes, free histone dimers and 2 mM ATP in Exchange buffer (70 mM NaCl, 10 mM Tris-HCl [pH8.0], 5 mM MgCl2, 0.1 mg/ml BSA and 1 mM DTT at 30 °C for 60 min). To reconstitute biotinylated mononucleosomes, 200 bp 601 DNA fragment was generated by PCR using biotinylated DNA primers. The biotinylated mononuclesomes were immobilized onto Dynabeads M-280 (Invitrogen). After washing to remove unbound mononucleosomes, the immobilized mononucleosomes were incubated with remodeling enzymes, free histone dimers and 2 mM ATP in Exchange buffer at room temperature for 60 min. The immobilized mononucleosomes were washed three times with Exchange buffer and subjected to SDS-PAGE and Western blotting.
This work was supported by grants from the National Institutes of Health to C.L.P. (GM54096). OJR is supported in part by a Career Award in the Biomedical Sciences from the Burroughs Wellcome Fund and grants from the NIGMS and HFSP. M.P.C is supported by the Avenir Program from Inserm. We thank John Holik (UMMS) for assistance with the tiling array studies, Jerry Workman (Stowers Institute) for yeast histone expression vectors, John Lescyz (UMMS) for mass spectrometry analysis, Nicholas Adkins for the western blot with Millipore α-Htz1 sera (Fig. S7), Erica Hong for help with Figure 7, members of the Peterson lab for helpful discussions throughout the course of this work, and Genevieve Almouzni, Angela Taddei and Valerie Borde for critical reading of the manuscript.
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.